\begin{filecontents*}[overwrite]{leannote.sty} \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{leannote}[2026/11/18 Lean margin notes] \RequirePackage{xcolor,url,needspace,zref-savepos,zref-abspage,etoolbox} \makeatletter \newcommand*\leannotefont{\fontfamily{zi4}\fontseries{m}\fontshape{n}\selectfont} \newcommand*\leannotecolor{\color[gray]{0.33}} \newcommand*\leannotestyle{% \everypar{}\normalfont\scriptsize\upshape \setlength\baselineskip{9.2pt}% \parindent\z@\leftskip\z@\rightskip\z@\@plus 2.5em\relax \parfillskip\z@\@plus 1fil\relax\spaceskip\fontdimen2\font\xspaceskip\z@ \hyphenpenalty 4000\exhyphenpenalty 4000\pretolerance\m@ne\tolerance 9000\relax \emergencystretch 1em\hbadness 10000\relax \let\nolinkurl\leandecl\let\url\leandecl} \DeclareUrlCommand\leandecl{% \def\UrlFont{\leannotefont\color[gray]{0.18}}% \def\UrlBreaks{\do\.\do\_}\def\UrlBigBreaks{}\def\UrlNoBreaks{}% \def\UrlSpecials{}\Urlmuskip=0mu plus 1mu\relax} \@namedef{PFln@lead@same}{Same route.} \@namedef{PFln@lead@formal}{Formalized.} \@namedef{PFln@lead@route}{Different route.} \@namedef{PFln@lead@narrower}{Narrower in Lean.} \@namedef{PFln@lead@different}{Different form in Lean.} \@namedef{PFln@lead@conditional}{Conditional in Lean.} \@namedef{PFln@lead@details}{Details in Lean.} \@namedef{PFln@lead@hand}{Not formalized.} \@namedef{PFln@lead@none}{} \newcommand*\PFln@lead[1]{% \ifcsname PFln@lead@#1\endcsname \edef\PFln@tmp{\csname PFln@lead@#1\endcsname}% \ifx\PFln@tmp\@empty\else{\itshape\PFln@tmp}\space\fi \else\PackageWarning{leannote}{Unknown note kind `#1'}\fi} \newcommand*\PFln@map[2]{\@namedef{PFln@map@#1}{#2}} \PFln@map{Same route}{same} \PFln@map{Same route (planned)}{same} \PFln@map{Same statement}{formal} \PFln@map{Same statement, different route}{route} \PFln@map{Same statement, different route (planned)}{route} \PFln@map{Same statement, different route (source-checked)}{route} \PFln@map{Different route (planned)}{route} \PFln@map{Lean scope narrower}{narrower} \PFln@map{Lean scope narrower (planned)}{narrower} \PFln@map{Lean scope narrower (source-checked)}{narrower} \PFln@map{Different form}{different} \PFln@map{Lean scope conditional}{conditional} \PFln@map{Proof by formalization (planned)}{details} \PFln@map{Not formalized}{hand} \PFln@map{Lean endpoint; route not compared}{none} \PFln@map{Lean status not yet assigned}{none} \newcommand\provisionalref[4]{% \ifcsname PFln@map@#3\endcsname \edef\PFln@key{\csname PFln@map@#3\endcsname}% \else\def\PFln@key{none}\fi \expandafter\leannote\expandafter{\PFln@key}{#4}} \newcommand\PFln@needspace[1]{% \ifhmode\par\fi \@tempskipb\lastskip\vskip-\@tempskipb\penalty-100 \begingroup\setlength\dimen@{#1}\advance\dimen@\@tempskipb \dimen@ii\pagegoal\advance\dimen@ii-\pagetotal \ifdim\dimen@>\dimen@ii \PackageInfo{leannote}{page break before note: needs \the\dimen@, has \the\dimen@ii}% \ifdim\dimen@ii>\z@\vfil\fi\break \else\vskip\@tempskipb\fi \endgroup} \newsavebox\PFln@box \newcount\PFln@cnt \newif\ifPFln@armed \newcommand\PFln@build[1]{% \global\setbox\PFln@box\vtop{\hsize\marginparwidth\leannotestyle\leavevmode{\leannotecolor#1}\par}} \newcommand\leannote[2]{\PFln@build{\PFln@lead{#1}#2}\PFln@dispatch} \newcommand\leannoteexplain[1]{\PFln@build{#1}\PFln@dispatch} \newcommand\PFln@dispatch{% \ifvmode \if@nobreak\PackageInfo{leannote}{note after a heading: no room requested}% \else\PFln@needspace{\dimexpr\dp\PFln@box+2\baselineskip\relax}\fi \global\PFln@armedtrue \expandafter\PFln@look \else \PFln@place \fi} \newcommand\PFln@look{\@ifnextchar\begin{}{\@ifnextchar\paragraph\PFln@para\PFln@leave}} \newcommand\PFln@para[2]{#1{#2}\leavevmode\PFln@fire\ignorespaces} \newcommand\PFln@leave{\leavevmode\PFln@fire} \newcommand\PFln@fire{\ifPFln@armed\global\PFln@armedfalse\PFln@place\fi} \let\PFln@orig@dth\deferred@thm@head \def\deferred@thm@head#1{\PFln@orig@dth{\PFln@fire#1}} \newcommand\PFln@place{% \global\advance\PFln@cnt\@ne \zsavepos{PFln\the\PFln@cnt}% \setbox\z@\hbox{\copy\PFln@box}% \PackageInfo{leannote}{note \the\PFln@cnt: depth \the\dp\z@\space height \the\ht\z@}% \immediate\write\@mainaux{\string\PFlnsize{\the\PFln@cnt}{\the\ht\z@}{\the\dp\z@}}% \ht\z@\z@\dp\z@\z@ \rlap{\kern\dimexpr1in+\hoffset+\oddsidemargin+\textwidth+\marginparsep -\zposx{PFln\the\PFln@cnt}sp\relax\box\z@}} \providecommand\PFlnsize[3]{} \AtBeginDocument{\zref@addprop{savepos}{abspage}} \makeatother \endinput \end{filecontents*} \begin{filecontents*}[overwrite]{proofnote-cover.sty} \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{proofnote-cover}[2026/11/17 AI proof note cover helpers (mock-up)] \RequirePackage{xcolor,tikz,array,tabularx,booktabs,enumitem} \newcommand{\tbd}[1]{\tikz[baseline=(n.base)]\node[draw,dashed,line width=0.35pt, inner xsep=1.6pt,inner ysep=0.9pt,rounded corners=0.8pt](n){\normalfont\sffamily\scriptsize\color[gray]{0.35}#1};} \newcommand{\stSame}{\tikz[baseline=-0.6ex]\fill (0,0) circle (0.9mm);} \newcommand{\stNarrow}{\tikz[baseline=-0.6ex]{\draw[line width=0.4pt] (0,0) circle (0.9mm); \fill (0,-0.9mm) arc (-90:90:0.9mm) -- cycle;}} \newcommand{\stHand}{\tikz[baseline=-0.6ex]\draw[line width=0.5pt] (0,0) circle (0.9mm);} \newcommand{\stImp}{\tikz[baseline=-0.6ex]\draw[line width=0.5pt] (-0.8mm,-0.8mm) rectangle (0.8mm,0.8mm);} \newcommand{\pnband}[2]{% \noindent\begin{tikzpicture} \useasboundingbox (0,2mm) rectangle (\linewidth,-6.5mm); \node[anchor=west,inner sep=0pt] at (0,0) {\normalfont\sffamily\bfseries\small\addfontfeature{}AI\ \ PROOF\ \ NOTE}; \node[anchor=east,inner sep=0pt] at (\linewidth,0) {\normalfont\sffamily\small #1}; \draw[line width=1.2pt] (0,-2.6mm) -- (\linewidth,-2.6mm); \node[anchor=west,inner sep=0pt] at (0,-5.4mm) {\normalfont\sffamily\scriptsize #2}; \end{tikzpicture}\par} \providecommand{\addfontfeature}[1]{} \newlength{\pnlab}\setlength{\pnlab}{0.92in} \newlength{\pnsep}\setlength{\pnsep}{0.16in} \newenvironment{pnrow}[1]{% \par\noindent\begin{minipage}[t]{\pnlab}\raggedright\normalfont\sffamily\scriptsize\bfseries \MakeUppercase{#1}\end{minipage}\hspace{\pnsep}% \begin{minipage}[t]{\dimexpr\linewidth-\pnlab-\pnsep\relax}\small}{% \end{minipage}\par} \newcommand{\pnrule}{\par\vspace{0.7ex}\noindent{\color[gray]{0.55}\rule{\linewidth}{0.3pt}}\par\vspace{0.6ex}} \newenvironment{pnthm}[2]{% \par\noindent\textbf{#1}\ \textit{(#2)}\ \ignorespaces}{\par} \newcommand{\pnstatus}[1]{\par\vspace{0.5ex}\noindent{\raggedright\normalfont\sffamily\scriptsize #1\par}} \endinput \end{filecontents*} \documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage[type1]{newpxtext} \usepackage{amsmath,amsthm,amssymb,mathtools} \usepackage{newpxmath} \usepackage[letterpaper,textwidth=5.4in,textheight=8.6in,left=0.8in,marginparsep=0.2in,marginparwidth=1.3in]{geometry} \PassOptionsToPackage{noadjust}{marginnote}\usepackage{xcolor,marginnote,microtype,graphicx,caption,booktabs,tabularx,xltabular,needspace,array,enumitem,tikz} \usepackage[super,sort&compress]{natbib} \usepackage[hidelinks,bookmarksnumbered]{hyperref} \hypersetup{pdftitle={Finite certificates for sequential measurements in the exceptional Jordan algebra},pdfauthor={Robert R. MacGregor}} \setlength{\emergencystretch}{2em} \captionsetup{font=small,hypcap=false} \makeatletter\setlength{\@fptop}{0pt}\makeatother \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{example}[theorem]{Example} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \numberwithin{equation}{section} \newenvironment{conceptfigure}{\begin{figure}[htbp]\centering}{\end{figure}} \usepackage{leannote} \usepackage[titles]{tocloft} \renewcommand{\cftdot}{} \setlength{\cftbeforesecskip}{3pt} \renewcommand{\cftsecfont}{\bfseries} \renewcommand{\cftsecpagefont}{\bfseries} \setcounter{tocdepth}{2} \newcommand{\R}{\mathbb{R}} \newcommand{\C}{\mathbb{C}} \newcommand{\Ha}{\mathbb{H}} \newcommand{\Oct}{\mathbb{O}} \newcommand{\Q}{\mathbb{Q}} \newcommand{\J}{J} \newcommand{\HR}{H_3(\R)} \newcommand{\HC}{H_3(\C)} \newcommand{\HH}{H_3(\Ha)} \newcommand{\HO}{H_3(\Oct)} \newcommand{\OP}{\Oct P^2} \newcommand{\tr}{\operatorname{tr}} \newcommand{\Tr}{\operatorname{Tr}} \newcommand{\Jord}{\operatorname{Jord}} \newcommand{\jp}{\circ} \newcommand{\Aut}{\operatorname{Aut}} \newcommand{\Spin}{\operatorname{Spin}} \newcommand{\Ffour}{F_4} \newcommand{\Esix}{E_{6(-26)}} \newcommand{\spn}{\operatorname{span}} \newcommand{\ip}[2]{\langle #1,#2\rangle} \newcommand{\Acal}{\mathcal{A}} \newcommand{\Dtet}{\mathcal{D}} \newcommand{\BH}{B(\mathcal{H})} \newcommand{\Hil}{\mathcal{H}} \newcommand{\sa}{\mathrm{sa}} \newcommand{\TV}{\operatorname{TV}} \newcommand{\id}{\mathrm{id}} \newcommand{\CHECK}[1]{\textcolor{orange}{\textsf{[check: #1]}}} \usepackage{proofnote-cover} \newcommand{\leanrepo}{\href{https://github.com/macgregor-sudo/albert-sequential-measurements}{\nolinkurl{github.com/macgregor-sudo/albert-sequential-measurements}}} \newcommand{\coverhead}[1]{{\normalfont\bfseries\scshape #1}\par\vspace{-0.25\baselineskip}} \usepackage{etoolbox} \newcommand{\defcore}[2]{\csdef{PFcore@#1}{#2}} \newcommand{\corenum}[3]{\csdef{PFcoreK@#1}{#2}\csdef{PFcoreN@#1}{#3}} \newcommand{\PFcorepart}[1]{\if\relax\detokenize{#1}\relax\else#1\fi} \newcommand{\PFcorenumber}[2]{\ifcsdef{PFcoreN@#2}{\ifcsempty{PFcoreN@#2}{~\PFcorepart{#1}~(core paper)}{\ifcsstring{PFcoreK@#2}{eq}{~(core~(\csuse{PFcoreN@#2})\PFcorepart{#1})}{~(core~\csuse{PFcoreK@#2}~\csuse{PFcoreN@#2}\PFcorepart{#1})}}}{\PackageWarning{coreref}{No core number for #2}~\PFcorepart{#1}~(core paper)}} \newcommand{\PFcorebody}[2]{\ifcsdef{PFcore@#2}{\csuse{PFcore@#2}}{\PackageWarning{coreref}{No name for core label #2}\textbf{[#2]}}\PFcorenumber{#1}{#2}} \newcommand{\coreref}[2][]{the~\PFcorebody{#1}{#2}} \newcommand{\Coreref}[2][]{The~\PFcorebody{#1}{#2}} \newcommand{\corename}[2][]{the~\PFcorebody{#1}{#2}} \defcore{sec:set}{section on ideal tests and their sequences} \defcore{sec:set:eja}{summary of Euclidean Jordan algebras} \defcore{lem:set:support}{support lemma} \defcore{def:ideal}{definition of an ideal test} \defcore{thm:ideal}{theorem that ideal tests are L\"uders instruments} \defcore{def:idealmodel}{definition of sequential projective representations} \defcore{def:memorymodel}{definition of completely positive realizations} \defcore{thm:bridgeJ}{criterion for the Albert algebra} \defcore{lem:bri:gns}{real cyclic representation} \defcore{lem:purity}{purity lemma} \defcore{sec:bri:recon}{reconstruction of a Jordan homomorphism} \defcore{lem:bri:orbit}{orbit-span lemma} \defcore{eq:bri:refl}{reflection formula} \defcore{sec:four}{four-point section} \defcore{prop:three}{three-atom proposition} \defcore{def:traces}{thirteen traces} \defcore{thm:fourpoint}{four-point criterion} \defcore{prop:windowdet}{formula for $\Acal$ as a determinant in a complex subalgebra} \defcore{cor:fp:rank}{rank trichotomy} \defcore{lem:fp:access}{lemma on relations forced by the statistics} \defcore{eq:fp:access}{access formulas} \defcore{sec:fp:tetra}{section on the tetrahedron identity} \defcore{thm:tetra}{tetrahedron identity} \defcore{prop:Dacal}{identity $\Dtet\prod o_{ab}=\Acal$} \defcore{cor:fp:noideal}{four-atom no-go corollary} \defcore{prop:max}{maximum of $\Acal$} \defcore{cor:fp:maxmixed}{corollary that the maximiser looks mixed} \defcore{eq:fp:assoc}{normal form of $\Acal$} \defcore{eq:fp:DeltaW}{formula for the triangle factor} \defcore{eq:fp:maximiser}{maximiser of $\Acal$} \defcore{sec:cer}{section on finite certificates} \defcore{thm:cer:uniform}{uniform length-seven theorem} \defcore{prop:cer:universalsix}{six-step models for every generating quadruple} \defcore{sec:cer:eleven}{section on words of length eleven} \defcore{sec:cer:frame}{section on finite lists} \defcore{lem:cer:recon}{finite reconstruction lemma} \defcore{sec:cer:horizon}{length-seven argument} \defcore{lem:cer:short}{relations at short histories} \defcore{lem:cer:gram28}{lemma that the words fix the Gram matrix} \defcore{thm:cer:seven}{length-seven theorem for $\mathsf M$ and $\mathsf S$} \defcore{prop:horizon}{six-step model for $\mathsf M$ and $\mathsf S$} \defcore{rem:cer:variance}{variance remark} \defcore{thm:eleven}{length-eleven theorem} \defcore{prop:Dfails}{model with $\Dtet>0$ and no purity} \defcore{sec:cer:equality}{section on linear equalities} \defcore{prop:cer:noequality}{proposition on linear equalities} \defcore{rem:cer:fivestep}{remark on equalities through five steps} \defcore{prop:robust}{robustness proposition} \defcore{prop:cer:nearideal}{stability under face calibration} \defcore{thm:pentad}{five-test theorem} \defcore{sec:fiv:coord}{coordinates for the five tests} \defcore{cor:helly}{Helly-number corollary} \defcore{sec:ins}{section on the projection postulate} \defcore{thm:cp69}{finite realization with non-L\"uders instruments} \defcore{cor:ins:nohalf}{realization omitting the initial memory state} \defcore{rem:ins:notideal}{remark that the realization is not projective} \defcore{prop:ins:lower}{lower bound on the realization dimension} \defcore{sec:ins:dilation}{section on sharp effects with arbitrary channels} \defcore{lem:ins:factor}{square-root factorisation lemma} \defcore{prop:ins:sharpdilation}{sharp dilation} \defcore{cor:ins:sharpalbert}{sharp-effect realization for $\J$} \defcore{rem:ins:sharpnotideal}{remark that what fails is ideality} \defcore{prop:markov}{Markov property of complete measurements} \defcore{sec:fiv:local}{discussion of local and global after the Helly-number corollary} \defcore{app:conv}{appendix on conventions} \defcore{app:A:oct}{octonion conventions} \defcore{lem:A:slots}{slot lemma} \defcore{eq:A:matrix}{matrix coordinates of $\J$} \defcore{app:A:points}{description of atoms and dyads} \defcore{lem:A:restrict}{restriction lemma} \defcore{lem:A:three}{normal form for three atoms} \defcore{lem:A:degenerate}{lemma on degenerate pairs} \defcore{app:A:F4}{facts on $\Ffour$} \defcore{app:A:Acal}{expansion of $\Acal$ in full} \defcore{eq:A:Acal52}{expansion of $\Acal$ in the thirteen traces} \defcore{lem:five:sub}{lemma on proper subalgebras} \defcore{app:mem:cp69}{spinor realization} \defcore{app:cert:seven}{detailed length-seven argument} \defcore{lem:cer:gram7}{Gram formulas at length seven} \defcore{app:cert:conv}{identification of the configurations} \defcore{app:cert:models}{explicit models on $\C^{12}$} \corenum{sec:set}{Section}{3} \corenum{sec:set:eja}{Section}{3.1} \corenum{lem:set:support}{Lemma}{3.1} \corenum{def:ideal}{Definition}{3.2} \corenum{thm:ideal}{Theorem}{3.3} \corenum{def:idealmodel}{Definition}{3.4} \corenum{def:memorymodel}{Definition}{3.6} \corenum{thm:bridgeJ}{Theorem}{4.1} \corenum{lem:bri:gns}{Lemma}{4.3} \corenum{lem:purity}{Lemma}{4.5} \corenum{sec:bri:recon}{Section}{4.3} \corenum{lem:bri:orbit}{Lemma}{4.7} \corenum{eq:bri:refl}{eq}{4.4} \corenum{sec:four}{Section}{5} \corenum{prop:three}{Proposition}{5.1} \corenum{def:traces}{Definition}{5.2} \corenum{thm:fourpoint}{Theorem}{5.5} \corenum{prop:windowdet}{Proposition}{5.6} \corenum{cor:fp:rank}{Corollary}{5.7} \corenum{lem:fp:access}{Lemma}{5.8} \corenum{eq:fp:access}{eq}{5.2} \corenum{sec:fp:tetra}{Section}{5.4} \corenum{thm:tetra}{Theorem}{5.9} \corenum{prop:Dacal}{Proposition}{5.10} \corenum{cor:fp:noideal}{Corollary}{5.11} \corenum{prop:max}{Proposition}{5.12} \corenum{cor:fp:maxmixed}{Corollary}{5.14} \corenum{eq:fp:assoc}{eq}{5.7} \corenum{eq:fp:DeltaW}{eq}{5.8} \corenum{eq:fp:maximiser}{eq}{5.13} \corenum{sec:cer}{Section}{6} \corenum{thm:cer:uniform}{Theorem}{6.1} \corenum{prop:cer:universalsix}{Proposition}{6.2} \corenum{sec:cer:eleven}{Section}{6.3} \corenum{sec:cer:frame}{Section}{6.1} \corenum{lem:cer:recon}{Lemma}{6.4} \corenum{sec:cer:horizon}{Section}{6.2} \corenum{lem:cer:short}{Lemma}{6.5} \corenum{lem:cer:gram28}{Lemma}{6.8} \corenum{thm:cer:seven}{Theorem}{6.10} \corenum{prop:horizon}{Proposition}{6.11} \corenum{rem:cer:variance}{Remark}{6.12} \corenum{thm:eleven}{Theorem}{6.15} \corenum{prop:Dfails}{Proposition}{6.16} \corenum{sec:cer:equality}{Section}{6.5} \corenum{prop:cer:noequality}{Proposition}{6.17} \corenum{rem:cer:fivestep}{Remark}{6.18} \corenum{prop:robust}{Proposition}{6.19} \corenum{prop:cer:nearideal}{Proposition}{6.20} \corenum{thm:pentad}{Theorem}{7.2} \corenum{sec:fiv:coord}{Section}{7.1} \corenum{cor:helly}{Corollary}{7.5} \corenum{sec:ins}{Section}{8} \corenum{thm:cp69}{Theorem}{8.1} \corenum{cor:ins:nohalf}{Corollary}{8.2} \corenum{rem:ins:notideal}{Remark}{8.3} \corenum{prop:ins:lower}{Proposition}{8.4} \corenum{sec:ins:dilation}{Section}{8.2} \corenum{lem:ins:factor}{Lemma}{8.5} \corenum{prop:ins:sharpdilation}{Proposition}{8.6} \corenum{cor:ins:sharpalbert}{Corollary}{8.7} \corenum{rem:ins:sharpnotideal}{Remark}{8.8} \corenum{prop:markov}{Proposition}{8.10} \corenum{sec:fiv:local}{}{}\corenum{app:conv}{Appendix}{A} \corenum{app:A:oct}{Appendix}{A.2} \corenum{lem:A:slots}{Lemma}{A.1} \corenum{eq:A:matrix}{eq}{A.2} \corenum{app:A:points}{Appendix}{A.4} \corenum{lem:A:restrict}{Lemma}{A.2} \corenum{lem:A:three}{Lemma}{A.3} \corenum{lem:A:degenerate}{Lemma}{A.6} \corenum{app:A:F4}{Appendix}{A.5} \corenum{app:A:Acal}{Appendix}{A.9} \corenum{eq:A:Acal52}{eq}{5.4} \corenum{lem:five:sub}{Lemma}{E.1} \corenum{app:mem:cp69}{Appendix}{D} \corenum{app:cert:seven}{Appendix}{C.2} \corenum{lem:cer:gram7}{Lemma}{C.1} \corenum{app:cert:conv}{Appendix}{C.4} \corenum{app:cert:models}{Appendix}{C.5} \expandafter\def\csname PFref@AbramskyBrandenburger2011\endcsname{S.~Abramsky and A.~Brandenburger, The sheaf-theoretic structure of non-locality and contextuality, \emph{New J. Phys.} \textbf{13} (2011) 113036. \href{https://doi.org/10.1088/1367-2630/13/11/113036}{doi:10.1088/1367-2630/13/11/113036}. \href{https://arxiv.org/abs/1102.0264}{arXiv:1102.0264}} \expandafter\def\csname PFref@Albert1934\endcsname{A.~A. Albert, On a certain algebra of quantum mechanics, \emph{Ann. of Math.} (2) \textbf{35} (1934) 65--73. \href{https://doi.org/10.2307/1968118}{doi:10.2307/1968118}} \expandafter\def\csname PFref@AlbertPaige1959\endcsname{A.~A. Albert and L.~J. Paige, On a homomorphism property of certain Jordan algebras, \emph{Trans. Amer. Math. Soc.} \textbf{93} (1959) 20--29. \href{https://doi.org/10.1090/S0002-9947-1959-0108524-2}{doi:10.1090/S0002-9947-1959-0108524-2}} \expandafter\def\csname PFref@AlfsenShultz2003\endcsname{E.~M. Alfsen and F.~W. Shultz, \emph{Geometry of State Spaces of Operator Algebras}, Birkh\"auser, Boston, 2003} \expandafter\def\csname PFref@AlfsenShultzStormer1978\endcsname{E.~M. Alfsen, F.~W. Shultz and E.~St{\o}rmer, A Gelfand--Neumark theorem for Jordan algebras, \emph{Adv. Math.} \textbf{28} (1978) 11--56. \href{https://doi.org/10.1016/0001-8708(78)90044-0}{doi:10.1016/0001-8708(78)90044-0}} \expandafter\def\csname PFref@Araki1980b\endcsname{H.~Araki, On a characterization of the state space of quantum mechanics, \emph{Commun. Math. Phys.} \textbf{75} (1980) 1--24. \href{https://doi.org/10.1007/BF01962588}{doi:10.1007/BF01962588}} \expandafter\def\csname PFref@AsoreyScolariciSolombrino2007\endcsname{M.~Asorey, G.~Scolarici and L.~Solombrino, Complex projections of completely positive quaternionic maps, \emph{Theoret. and Math. Phys.} \textbf{151} (2007) 735--743. \href{https://www.mathnet.ru/eng/tmf6051}{Math-Net.Ru}} \expandafter\def\csname PFref@Baez2002\endcsname{J.~C. Baez, The octonions, \emph{Bull. Amer. Math. Soc.} \textbf{39} (2002) 145--205. \href{https://doi.org/10.1090/S0273-0979-01-00934-X}{doi:10.1090/S0273-0979-01-00934-X}. \href{https://arxiv.org/abs/math/0105155}{arXiv:math/0105155}} \expandafter\def\csname PFref@Baez2012\endcsname{J.~C. Baez, Division algebras and quantum theory, \emph{Found. Phys.} \textbf{42} (2012) 819--855. \href{https://doi.org/10.1007/s10701-011-9566-z}{doi:10.1007/s10701-011-9566-z}. \href{https://arxiv.org/abs/1101.5690}{arXiv:1101.5690}} \expandafter\def\csname PFref@BaezSchwahn2026\endcsname{J.~C. Baez and P.~Schwahn, The Standard Model gauge group from the exceptional Jordan algebra, \href{https://arxiv.org/abs/2606.15235}{arXiv:2606.15235} (2026)} \expandafter\def\csname PFref@BarnumHilgert2020\endcsname{H.~Barnum and J.~Hilgert, Spectral properties of convex bodies, \emph{J. Lie Theory} \textbf{30} (2020) 315--344. \href{https://doi.org/10.5802/jolt.1118}{doi:10.5802/jolt.1118}. \href{https://arxiv.org/abs/1904.03753}{arXiv:1904.03753} (preprint title: Strongly symmetric spectral convex bodies are Jordan algebra state spaces)} \expandafter\def\csname PFref@BarnumWilce2014\endcsname{H.~Barnum and A.~Wilce, Local tomography and the Jordan structure of quantum theory, \emph{Found. Phys.} \textbf{44} (2014) 192--212. \href{https://doi.org/10.1007/s10701-014-9777-1}{doi:10.1007/s10701-014-9777-1}. \href{https://arxiv.org/abs/1202.4513}{arXiv:1202.4513}} \expandafter\def\csname PFref@BGW2015\endcsname{H.~Barnum, M.~A. Graydon and A.~Wilce, Some nearly quantum theories, in \emph{Proc. 12th Int. Workshop on Quantum Physics and Logic (QPL 2015)}, Electron. Proc. Theor. Comput. Sci. \textbf{195} (2015) 59--70. \href{https://doi.org/10.4204/EPTCS.195.5}{doi:10.4204/EPTCS.195.5}. \href{https://arxiv.org/abs/1507.06278}{arXiv:1507.06278}} \expandafter\def\csname PFref@BGW2020\endcsname{H.~Barnum, M.~A. Graydon and A.~Wilce, Composites and categories of Euclidean Jordan algebras, \emph{Quantum} \textbf{4} (2020) 359. \href{https://doi.org/10.22331/q-2020-11-08-359}{doi:10.22331/q-2020-11-08-359}. \href{https://arxiv.org/abs/1606.09331}{arXiv:1606.09331}} \expandafter\def\csname PFref@BMU2014\endcsname{H.~Barnum, M.~P. M\"uller and C.~Ududec, Higher-order interference and single-system postulates characterizing quantum theory, \emph{New J. Phys.} \textbf{16} (2014) 123029. \href{https://doi.org/10.1088/1367-2630/16/12/123029}{doi:10.1088/1367-2630/16/12/123029}. \href{https://arxiv.org/abs/1403.4147}{arXiv:1403.4147}} \expandafter\def\csname PFref@Boyle2020\endcsname{L.~Boyle, The standard model, the exceptional Jordan algebra, and triality, \emph{J. Math. Phys.} \textbf{67} (2026) 071701. \href{https://doi.org/10.1063/5.0336175}{doi:10.1063/5.0336175}. \href{https://arxiv.org/abs/2006.16265}{arXiv:2006.16265}} \expandafter\def\csname PFref@BoyleFarnsworth2020\endcsname{L.~Boyle and S.~Farnsworth, The standard model, the Pati--Salam model, and ``Jordan geometry'', \emph{New J. Phys.} \textbf{22} (2020) 073023. \href{https://doi.org/10.1088/1367-2630/ab9709}{doi:10.1088/1367-2630/ab9709}. \href{https://arxiv.org/abs/1910.11888}{arXiv:1910.11888}} \expandafter\def\csname PFref@Brehm1990\endcsname{U.~Brehm, The shape invariant of triangles and trigonometry in two-point homogeneous spaces, \emph{Geom. Dedicata} \textbf{33} (1990) 59--76. \href{https://doi.org/10.1007/BF00147601}{doi:10.1007/BF00147601}} \expandafter\def\csname PFref@BrehmEtTaoui2001\endcsname{U.~Brehm and B.~Et-Taoui, Congruence criteria for finite subsets of quaternionic elliptic and quaternionic hyperbolic spaces, \emph{Geom. Dedicata} \textbf{84} (2001) 261--269. \href{https://doi.org/10.1023/A:1010306423797}{doi:10.1023/A:1010306423797}} \expandafter\def\csname PFref@BMKG2013\endcsname{C.~Budroni, T.~Moroder, M.~Kleinmann and O.~G\"uhne, Bounding temporal quantum correlations, \emph{Phys. Rev. Lett.} \textbf{111} (2013) 020403. \href{https://doi.org/10.1103/PhysRevLett.111.020403}{doi:10.1103/PhysRevLett.111.020403}. \href{https://arxiv.org/abs/1302.6223}{arXiv:1302.6223}} \expandafter\def\csname PFref@BudroniEmary2014\endcsname{C.~Budroni and C.~Emary, Temporal quantum correlations and Leggett--Garg inequalities in multilevel systems, \emph{Phys. Rev. Lett.} \textbf{113} (2014) 050401. \href{https://doi.org/10.1103/PhysRevLett.113.050401}{doi:10.1103/PhysRevLett.113.050401}. \href{https://arxiv.org/abs/1309.3678}{arXiv:1309.3678}} \expandafter\def\csname PFref@BudroniFagundesKleinmann2019\endcsname{C.~Budroni, G.~Fagundes and M.~Kleinmann, Memory cost of temporal correlations, \emph{New J. Phys.} \textbf{21} (2019) 093018. \href{https://doi.org/10.1088/1367-2630/ab3cb4}{doi:10.1088/1367-2630/ab3cb4}. \href{https://arxiv.org/abs/1902.06517}{arXiv:1902.06517}} \expandafter\def\csname PFref@BuschGrabowskiLahti1995\endcsname{P.~Busch, M.~Grabowski and P.~J. Lahti, Repeatable measurements in quantum theory: their role and feasibility, \emph{Found. Phys.} \textbf{25} (1995) 1239--1266. \href{https://doi.org/10.1007/BF02055331}{doi:10.1007/BF02055331}} \expandafter\def\csname PFref@BuschSingh1998\endcsname{P.~Busch and J.~Singh, L\"uders theorem for unsharp quantum measurements, \emph{Phys. Lett. A} \textbf{249} (1998) 10--12. \href{https://doi.org/10.1016/S0375-9601(98)00704-X}{doi:10.1016/S0375-9601(98)00704-X}. \href{https://arxiv.org/abs/1304.0054}{arXiv:1304.0054}} \expandafter\def\csname PFref@CafutaKlepPovh2015\endcsname{K.~Cafuta, I.~Klep and J.~Povh, Rational sums of hermitian squares of free noncommutative polynomials, \emph{Ars Math. Contemp.} \textbf{9} (2015) 243--259} \expandafter\def\csname PFref@Catto2003\endcsname{S.~Catto, Exceptional projective geometries and internal symmetries, \href{https://arxiv.org/abs/hep-th/0302079}{arXiv:hep-th/0302079} (2003)} \expandafter\def\csname PFref@ChiribellaYuan2014\endcsname{G.~Chiribella and X.~Yuan, Measurement sharpness cuts nonlocality and contextuality in every physical theory, \href{https://arxiv.org/abs/1404.3348}{arXiv:1404.3348} (2014)} \expandafter\def\csname PFref@ChiribellaYuan2018\endcsname{G.~Chiribella and X.~Yuan, Bridging the gap between general probabilistic theories and the device-independent framework for nonlocality and contextuality, \emph{Inf. Comput.} \textbf{250} (2016) 15--49. \href{https://doi.org/10.1016/j.ic.2016.02.006}{doi:10.1016/j.ic.2016.02.006}. \href{https://arxiv.org/abs/1504.02395}{arXiv:1504.02395}} \expandafter\def\csname PFref@Cho2019\endcsname{K.~Cho, \emph{Effectuses in Categorical Quantum Foundations}, PhD thesis, Radboud University Nijmegen, 2019. \href{https://arxiv.org/abs/1910.12198}{arXiv:1910.12198}} \expandafter\def\csname PFref@ClementeKofler2016\endcsname{L.~Clemente and J.~Kofler, No fine theorem for macrorealism: limitations of the Leggett--Garg inequality, \emph{Phys. Rev. Lett.} \textbf{116} (2016) 150401. \href{https://arxiv.org/abs/1509.00348}{arXiv:1509.00348}} \expandafter\def\csname PFref@DakicPaterekBrukner2014\endcsname{B.~Daki\'c, T.~Paterek and \v{C}.~Brukner, Density cubes and higher-order interference theories, \emph{New J. Phys.} \textbf{16} (2014) 023028. \href{https://doi.org/10.1088/1367-2630/16/2/023028}{doi:10.1088/1367-2630/16/2/023028}. \href{https://arxiv.org/abs/1308.2822}{arXiv:1308.2822}} \expandafter\def\csname PFref@DaviesLewis1970\endcsname{E.~B. Davies and J.~T. Lewis, An operational approach to quantum probability, \emph{Commun. Math. Phys.} \textbf{17} (1970) 239--260} \expandafter\def\csname PFref@DeMedtsRowenSegev2023\endcsname{T.~De~Medts, L.~Rowen and Y.~Segev, Primitive 4-generated axial algebras of Jordan type, \emph{Proc. Amer. Math. Soc.} \textbf{152} (2024) 537--551. \href{https://doi.org/10.1090/proc/16571}{doi:10.1090/proc/16571}. \href{https://arxiv.org/abs/2301.02509}{arXiv:2301.02509}} \expandafter\def\csname PFref@DiazGarciaGutierrezJaimez2010\endcsname{J.~A. D\'{\i}az-Garc\'{\i}a and R.~Guti\'errez-J\'aimez, On Wishart distribution, \href{https://arxiv.org/abs/1010.1799}{arXiv:1010.1799} (2010)} \expandafter\def\csname PFref@DuboisViolette2016\endcsname{M.~Dubois-Violette, Exceptional quantum geometry and particle physics, \emph{Nucl. Phys. B} \textbf{912} (2016) 426--449. \href{https://doi.org/10.1016/j.nuclphysb.2016.04.018}{doi:10.1016/j.nuclphysb.2016.04.018}. \href{https://arxiv.org/abs/1604.01247}{arXiv:1604.01247}} \expandafter\def\csname PFref@DuboisVioletteTodorov2018\endcsname{M.~Dubois-Violette and I.~Todorov, Exceptional quantum geometry and particle physics~II, \emph{Nucl. Phys. B} \textbf{938} (2019) 751--761. \href{https://doi.org/10.1016/j.nuclphysb.2018.12.012}{doi:10.1016/j.nuclphysb.2018.12.012}. \href{https://arxiv.org/abs/1808.08110}{arXiv:1808.08110}} \expandafter\def\csname PFref@Ehrlich2026\endcsname{B.~Ehrlich, \texttt{sequential-products-lean}: classification of sequential products on Hermitian matrix algebras (Lean formalisation), repository \texttt{ehrlich-b/sequential-products-lean}, 2026; Palomar registry entries PALOMAR-2026-08-21-000009 and PALOMAR-2026-08-22-000006 (commit \texttt{9c6d92f5}), Apache License~2.0} \expandafter\def\csname PFref@Ehrlich2026ej\endcsname{B.~Ehrlich, \texttt{euclidean-jordan}: Euclidean Jordan algebras in Lean~4, repository \texttt{ehrlich-b/euclidean-jordan}, 2026; Palomar registry entries PALOMAR-2026-08-23-000005, PALOMAR-2026-08-23-000006, PALOMAR-2026-08-29-000001 and PALOMAR-2026-08-29-000008, Apache License~2.0} \expandafter\def\csname PFref@ElduqueIltyakov1999\endcsname{A.~Elduque and A.~V. Iltyakov, On polynomial invariants of exceptional simple algebraic groups, \emph{Canad. J. Math.} \textbf{51} (1999) 506--522. \href{https://doi.org/10.4153/CJM-1999-023-2}{doi:10.4153/CJM-1999-023-2}} \expandafter\def\csname PFref@FLW2024\endcsname{M.~Fanizza, J.~Lumbreras and A.~Winter, Quantum theory in finite dimension cannot explain every general process with finite memory, \emph{Commun. Math. Phys.} \textbf{405} (2024) 50. \href{https://doi.org/10.1007/s00220-023-04913-4}{doi:10.1007/s00220-023-04913-4}. \href{https://arxiv.org/abs/2209.11225}{arXiv:2209.11225}. (Journal Theorems~3--4 are Theorems~7--8 of arXiv v2.)} \expandafter\def\csname PFref@FarautKoranyi1994\endcsname{J.~Faraut and A.~Kor\'anyi, \emph{Analysis on Symmetric Cones}, Oxford Mathematical Monographs, The Clarendon Press, Oxford University Press, New York, 1994} \expandafter\def\csname PFref@FernandesWagnerNovoGalvao2024\endcsname{C.~Fernandes, R.~Wagner, L.~Novo and E.~F. Galv\~ao, Unitary-invariant witnesses of quantum imaginarity, \emph{Phys. Rev. Lett.} \textbf{133} (2024) 190201. \href{https://arxiv.org/abs/2403.15066}{arXiv:2403.15066}} \expandafter\def\csname PFref@Forrester2016\endcsname{P.~J. Forrester, Octonions in random matrix theory, \href{https://arxiv.org/abs/1610.08081}{arXiv:1610.08081} (2016)} \expandafter\def\csname PFref@GarnerMuller2020\endcsname{A.~J.~P. Garner and M.~P. M\"uller, Characterization of the probabilistic models that can be embedded in quantum theory, \href{https://arxiv.org/abs/2004.06136}{arXiv:2004.06136} (2020)} \expandafter\def\csname PFref@Glennie1966\endcsname{C.~M. Glennie, Some identities valid in special Jordan algebras but not valid in all Jordan algebras, \emph{Pacific J. Math.} \textbf{16} (1966) 47--59. \href{https://doi.org/10.2140/pjm.1966.16.47}{doi:10.2140/pjm.1966.16.47}} \expandafter\def\csname PFref@GogicTomasevic2026\endcsname{I.~Gogi\'c and M.~Toma\v{s}evi\'c, Idempotent words in finite-dimensional simple Jordan algebras, \href{https://arxiv.org/abs/2609.23581}{arXiv:2609.23581} (2026)} \expandafter\def\csname PFref@GMS2025\endcsname{I.~Gorshkov, A.~Mamontov and A.~Staroletov, On Jordan algebras that are factors of Matsuo algebras, \emph{Int. Electron. J. Algebra} \textbf{37} (2025) 70--90. \href{https://doi.org/10.24330/ieja.1488467}{doi:10.24330/ieja.1488467}. \href{https://arxiv.org/abs/2305.10958}{arXiv:2305.10958}} \expandafter\def\csname PFref@GorshkovStaroletov2020\endcsname{I.~Gorshkov and A.~Staroletov, On primitive 3-generated axial algebras of Jordan type, \emph{J. Algebra} \textbf{563} (2020) 74--99. \href{https://doi.org/10.1016/j.jalgebra.2020.07.014}{doi:10.1016/j.jalgebra.2020.07.014}. \href{https://arxiv.org/abs/2005.13791}{arXiv:2005.13791}} \expandafter\def\csname PFref@GudderGreechie2002\endcsname{S.~Gudder and R.~Greechie, Sequential products on effect algebras, \emph{Rep. Math. Phys.} \textbf{49} (2002) 87--111. \href{https://doi.org/10.1016/S0034-4877(02)80007-6}{doi:10.1016/S0034-4877(02)80007-6}} \expandafter\def\csname PFref@GPR1978\endcsname{M.~G\"unaydin, C.~Piron and H.~Ruegg, Moufang plane and octonionic quantum mechanics, \emph{Commun. Math. Phys.} \textbf{61} (1978) 69--85. \href{https://doi.org/10.1007/BF01609468}{doi:10.1007/BF01609468}} \expandafter\def\csname PFref@GunaydinGursey1973\endcsname{M.~G\"unaydin and F.~G\"ursey, Quark structure and octonions, \emph{J. Math. Phys.} \textbf{14} (1973) 1651--1667. \href{https://doi.org/10.1063/1.1666240}{doi:10.1063/1.1666240}} \expandafter\def\csname PFref@HarveyLawson1982\endcsname{R.~Harvey and H.~B. Lawson, Jr., Calibrated geometries, \emph{Acta Math.} \textbf{148} (1982) 47--157. \href{https://doi.org/10.1007/BF02392726}{doi:10.1007/BF02392726}} \expandafter\def\csname PFref@HassairiLajmi2018\endcsname{A.~Hassairi and S.~Lajmi, Singular Riesz measures on symmetric cones, \emph{Adv. Oper. Theory} \textbf{3} (2018) 337--350. \href{https://doi.org/10.15352/aot.1706-1183}{doi:10.15352/aot.1706-1183}. \href{https://arxiv.org/abs/1705.04057}{arXiv:1705.04057}} \expandafter\def\csname PFref@HeinosaariReitznerStano2008\endcsname{T.~Heinosaari, D.~Reitzner and P.~Stano, Notes on joint measurability of quantum observables, \emph{Found. Phys.} \textbf{38} (2008) 1133--1147. \href{https://doi.org/10.1007/s10701-008-9256-7}{doi:10.1007/s10701-008-9256-7}. \href{https://arxiv.org/abs/0811.0783}{arXiv:0811.0783}} \expandafter\def\csname PFref@Helton2002\endcsname{J.~W. Helton, ``Positive'' noncommutative polynomials are sums of squares, \emph{Ann. of Math.} (2) \textbf{156} (2002) 675--694} \expandafter\def\csname PFref@HeltonMcCullough2004\endcsname{J.~W. Helton and S.~A. McCullough, A Positivstellensatz for non-commutative polynomials, \emph{Trans. Amer. Math. Soc.} \textbf{356} (2004) 3721--3737} \expandafter\def\csname PFref@HSGB2018\endcsname{J.~Hoffmann, C.~Spee, O.~G\"uhne and C.~Budroni, Structure of temporal correlations of a qubit, \emph{New J. Phys.} \textbf{20} (2018) 102001. \href{https://arxiv.org/abs/1712.01234}{arXiv:1712.01234}} \expandafter\def\csname PFref@Hoggar1977\endcsname{S.~G. Hoggar, New sets of equi-isoclinic $n$-planes from old, \emph{Proc. Edinburgh Math. Soc.} \textbf{20} (1977) 287--291. \href{https://doi.org/10.1017/S0013091500026511}{doi:10.1017/S0013091500026511}} \expandafter\def\csname PFref@Hsiang1989\endcsname{W.-Y. Hsiang, On the laws of trigonometries of two-point homogeneous spaces, \emph{Ann. Global Anal. Geom.} \textbf{7} (1989) 29--45. \href{https://doi.org/10.1007/BF00137400}{doi:10.1007/BF00137400}} \expandafter\def\csname PFref@Iltyakov1997\endcsname{A.~V. Iltyakov, On generators of the algebra of polynomial invariants of group $F_4$, University of Sydney Research Report 97-4, 1997} \expandafter\def\csname PFref@Jacobson1954\endcsname{N.~Jacobson, Structure of alternative and Jordan bimodules, \emph{Osaka Math. J.} \textbf{6} (1954) 1--71. \href{https://ir.library.osaka-u.ac.jp/repo/ouka/all/8608/omj06_01_01.pdf}{ir.library.osaka-u.ac.jp/repo/ouka/all/8608}} \expandafter\def\csname PFref@JordanvNWigner1934\endcsname{P.~Jordan, J.~von Neumann and E.~Wigner, On an algebraic generalization of the quantum mechanical formalism, \emph{Ann. of Math.} \textbf{35} (1934) 29--64. \href{https://doi.org/10.2307/1968117}{doi:10.2307/1968117}} \expandafter\def\csname PFref@KhandelwalTavakoli2025\endcsname{S.~Khandelwal and A.~Tavakoli, Simulating quantum instruments with projective measurements and quantum post-processing, \emph{Phys. Rev. Lett.} \textbf{135} (2025) 040202. \href{https://arxiv.org/abs/2503.00956}{arXiv:2503.00956}} \expandafter\def\csname PFref@Kleinmann2014\endcsname{M.~Kleinmann, Sequences of projective measurements in generalized probabilistic models, \emph{J. Phys. A} \textbf{47} (2014) 455304. \href{https://doi.org/10.1088/1751-8113/47/45/455304}{doi:10.1088/1751-8113/47/45/455304}. \href{https://arxiv.org/abs/1402.3583}{arXiv:1402.3583}} \expandafter\def\csname PFref@KLM2025\endcsname{I.~Klep, J.~Levenson and S.~McCullough, Fej\'er--Riesz factorization for positive noncommutative trigonometric polynomials, \href{https://arxiv.org/abs/2511.09267v3}{arXiv:2511.09267v3} (2026)} \expandafter\def\csname PFref@Kollross\endcsname{A.~Kollross, Octonions, triality, the exceptional Lie algebra $F_4$ and polar actions on the Cayley hyperbolic plane, \emph{Internat. J. Math.} \textbf{31} (2020) 2050051. \href{https://doi.org/10.1142/S0129167X20500512}{doi:10.1142/S0129167X20500512}. \href{https://arxiv.org/abs/1802.08075}{arXiv:1802.08075}} \expandafter\def\csname PFref@Krasnov2019\endcsname{K.~Krasnov, SO(9) characterization of the Standard Model gauge group, \emph{J. Math. Phys.} \textbf{62} (2021) 021703. \href{https://doi.org/10.1063/5.0039941}{doi:10.1063/5.0039941}. \href{https://arxiv.org/abs/1912.11282}{arXiv:1912.11282}} \expandafter\def\csname PFref@Luders1951\endcsname{G.~L\"uders, \"Uber die Zustands\"anderung durch den Me{\ss}proze{\ss}, \emph{Ann. Phys. (Leipzig)} (6) \textbf{8} (1950) 322--328. \href{https://doi.org/10.1002/andp.19504430510}{doi:10.1002/andp.19504430510}; English translation: Concerning the state-change due to the measurement process, \emph{Ann. Phys. (Leipzig)} \textbf{15} (2006) 663--670, \href{https://doi.org/10.1002/andp.20065180904}{doi:10.1002/andp.20065180904}} \expandafter\def\csname PFref@MacGregorPhi2026\endcsname{R.~R. MacGregor, Finite observations and recognition of the compact Albert algebra, Palomar registry entry PALOMAR-2026-10-16-000013, 16 October 2026} \expandafter\def\csname PFref@MasilloScolariciSozzo2009\endcsname{F.~Masillo, G.~Scolarici and S.~Sozzo, Proper versus improper mixtures: Toward a quaternionic quantum mechanics, \emph{Theoret. and Math. Phys.} \textbf{160} (2009) 1006--1013. \href{https://arxiv.org/abs/0901.0795}{arXiv:0901.0795}} \expandafter\def\csname PFref@McCrimmon2004\endcsname{K.~McCrimmon, \emph{A Taste of Jordan Algebras}, Universitext, Springer, New York, 2004} \expandafter\def\csname PFref@McCrimmonIdentities2004\endcsname{K.~McCrimmon, The role of identities in Jordan algebras, \emph{Resenhas IME-USP} \textbf{6} (2004), no.~2/3, 265--280. \href{https://doi.org/10.11606/resimeusp.v6i2/3.75294}{doi:10.11606/resimeusp.v6i2/3.75294}} \expandafter\def\csname PFref@MonrasWinter2016\endcsname{A.~Monr\`as and A.~Winter, Quantum learning of classical stochastic processes: the completely-positive realization problem, \emph{J. Math. Phys.} \textbf{57} (2016) 015219. \href{https://doi.org/10.1063/1.4936935}{doi:10.1063/1.4936935}. \href{https://arxiv.org/abs/1412.3634}{arXiv:1412.3634}} \expandafter\def\csname PFref@MullerGarner2023\endcsname{M.~P. M\"uller and A.~J.~P. Garner, Testing quantum theory by generalizing noncontextuality, \emph{Phys. Rev. X} \textbf{13} (2023) 041001. \href{https://doi.org/10.1103/PhysRevX.13.041001}{doi:10.1103/PhysRevX.13.041001}. \href{https://arxiv.org/abs/2112.09719}{arXiv:2112.09719}} \expandafter\def\csname PFref@NPA2008\endcsname{M.~Navascu\'es, S.~Pironio and A.~Ac\'in, A convergent hierarchy of semidefinite programs characterizing the set of quantum correlations, \emph{New J. Phys.} \textbf{10} (2008) 073013. \href{https://doi.org/10.1088/1367-2630/10/7/073013}{doi:10.1088/1367-2630/10/7/073013}. \href{https://arxiv.org/abs/0803.4290}{arXiv:0803.4290}} \expandafter\def\csname PFref@Niestegge2008\endcsname{G.~Niestegge, A representation of quantum measurement in order-unit spaces, \emph{Found. Phys.} \textbf{38} (2008) 783--795. \href{https://doi.org/10.1007/s10701-008-9236-y}{doi:10.1007/s10701-008-9236-y}. \href{https://arxiv.org/abs/1001.3633}{arXiv:1001.3633}} \expandafter\def\csname PFref@Niestegge2012\endcsname{G.~Niestegge, Conditional probability, three-slit experiments, and the Jordan algebra structure of quantum mechanics, \emph{Adv. Math. Phys.} \textbf{2012} (2012) 156573. \href{https://doi.org/10.1155/2012/156573}{doi:10.1155/2012/156573}. \href{https://arxiv.org/abs/0912.0203}{arXiv:0912.0203}} \expandafter\def\csname PFref@Niestegge2020a\endcsname{G.~Niestegge, A simple and quantum-mechanically motivated characterization of the formally real Jordan algebras, \emph{Proc. R. Soc. A} \textbf{476} (2020) 20190604. \href{https://doi.org/10.1098/rspa.2019.0604}{doi:10.1098/rspa.2019.0604}. \href{https://arxiv.org/abs/1905.04189}{arXiv:1905.04189}} \expandafter\def\csname PFref@Niestegge2020b\endcsname{G.~Niestegge, Quantum probability's algebraic origin, \emph{Entropy} \textbf{22} (2020) 1196. \href{https://doi.org/10.3390/e22111196}{doi:10.3390/e22111196}. \href{https://arxiv.org/abs/2009.08489}{arXiv:2009.08489}} \expandafter\def\csname PFref@Niestegge2026a\endcsname{G.~Niestegge, On the homogeneity of the quantum transition probability, \href{https://arxiv.org/abs/2601.01936}{arXiv:2601.01936} (2026)} \expandafter\def\csname PFref@Nasmith2022\endcsname{B.~Nasmith, Octonions and the two strictly projective tight 5-designs, \emph{Algebr. Comb.} \textbf{5} (2022) 401--411. \href{https://doi.org/10.5802/alco.215}{doi:10.5802/alco.215}} \expandafter\def\csname PFref@Nasmith2022integers\endcsname{B.~Nasmith, Octonion integers and tight 5-designs, \href{https://arxiv.org/abs/2209.02398v1}{arXiv:2209.02398v1} (2022)} \expandafter\def\csname PFref@Hoehn2026lattice\endcsname{G.~H\"ohn, The unique extremal threemodular lattice of rank 26, the generalized hexagon (2,8), and the tight Cayley-plane 5-design, \href{https://arxiv.org/abs/2609.25086v1}{arXiv:2609.25086v1} (2026)} \expandafter\def\csname PFref@Hoehn2026albert\endcsname{G.~H\"ohn, Definite integral Albert algebras: arithmetic and finite geometry, \href{https://arxiv.org/abs/2609.27550v1}{arXiv:2609.27550v1} (2026)} \expandafter\def\csname PFref@Niestegge2026b\endcsname{G.~Niestegge, The bare necessities of a physically reasonable mathematical model for quantum theory, \href{https://arxiv.org/abs/2607.24474}{arXiv:2607.24474} (2026)} \expandafter\def\csname PFref@OrtegaSantander2002\endcsname{R.~Ortega and M.~Santander, Trigonometry of `complex Hermitian' type homogeneous symmetric spaces, \emph{J. Phys. A} \textbf{35} (2002) 7877--7918. \href{https://arxiv.org/abs/math-ph/0112030}{arXiv:math-ph/0112030}} \expandafter\def\csname PFref@OszmaniecBrodGalvao2021\endcsname{M.~Oszmaniec, D.~J. Brod and E.~F. Galv\~ao, Measuring relational information between quantum states, and applications, \href{https://arxiv.org/abs/2109.10006}{arXiv:2109.10006} (2021)} \expandafter\def\csname PFref@Ozawa1984\endcsname{M.~Ozawa, Quantum measuring processes of continuous observables, \emph{J. Math. Phys.} \textbf{25} (1984) 79--87} \expandafter\def\csname PFref@Peres1979\endcsname{A.~Peres, Proposed test for complex versus quaternion quantum theory, \emph{Phys. Rev. Lett.} \textbf{42} (1979) 683--686} \expandafter\def\csname PFref@PinedaRua2026\endcsname{S.~Pineda Montoya and J.~H. Rua Munoz, Operational foundations for quaternionic and octonionic quantum models: exact quaternionic channels and error correction, para-linear operators, and categorical closure boundaries beyond associativity, \href{https://arxiv.org/abs/2608.20259}{arXiv:2608.20259} (2026)} \expandafter\def\csname PFref@PeyrlParrilo2008\endcsname{H.~Peyrl and P.~A. Parrilo, Computing sum of squares decompositions with rational coefficients, \emph{Theoret. Comput. Sci.} \textbf{409} (2008) 269--281} \expandafter\def\csname PFref@PNA2010\endcsname{S.~Pironio, M.~Navascu\'es and A.~Ac\'in, Convergent relaxations of polynomial optimization problems with non-commuting variables, \emph{SIAM J. Optim.} \textbf{20} (2010) 2157--2180. \href{https://doi.org/10.1137/090760155}{doi:10.1137/090760155}. \href{https://arxiv.org/abs/0903.4368}{arXiv:0903.4368}} \expandafter\def\csname PFref@PopovaSamoilenko2006\endcsname{N.~D. Popova and Yu.~S. Samoilenko, On the existence of configurations of subspaces in a Hilbert space with fixed angles, \emph{SIGMA} \textbf{2} (2006) 055. \href{https://doi.org/10.3842/SIGMA.2006.055}{doi:10.3842/SIGMA.2006.055}. \href{https://arxiv.org/abs/math/0605717}{arXiv:math/0605717}} \expandafter\def\csname PFref@Racine1977\endcsname{M.~L. Racine, Maximal subalgebras of exceptional Jordan algebras, \emph{J. Algebra} \textbf{46} (1977) 12--21. \href{https://doi.org/10.1016/0021-8693(77)90391-X}{doi:10.1016/0021-8693(77)90391-X}} \expandafter\def\csname PFref@RoelandsTiersma2026\endcsname{M.~Roelands and S.~Tiersma, On the Shirshov--Cohn theorem for JB-algebras, \href{https://arxiv.org/abs/2605.24427v1}{arXiv:2605.24427v1} (2026)} \expandafter\def\csname PFref@SahaSantosAugusiak2020\endcsname{D.~Saha, R.~Santos and R.~Augusiak, Sum-of-squares decompositions for a family of noncontextuality inequalities and self-testing of quantum devices, \emph{Quantum} \textbf{4} (2020) 302. \href{https://doi.org/10.22331/q-2020-08-03-302}{doi:10.22331/q-2020-08-03-302}. \href{https://arxiv.org/abs/2002.12216}{arXiv:2002.12216}} \expandafter\def\csname PFref@SahaXuCabello2026\endcsname{D.~Saha, Z.-P.~Xu and A.~Cabello, Self-testing ideal quantum measurements, preprint (2026). \href{https://arxiv.org/abs/2610.03168v1}{arXiv:2610.03168v1}} \expandafter\def\csname PFref@SalamonWalpuski2017\endcsname{D.~A. Salamon and T.~Walpuski, Notes on the octonions, in \emph{Proceedings of the 23rd G\"okova Geometry--Topology Conference} (2017) 1--85. \href{https://arxiv.org/abs/1005.2820v4}{arXiv:1005.2820v4}} \expandafter\def\csname PFref@SamoilenkoYershova2006\endcsname{Yu.~S. Samoilenko and Yu.~Yu. Yershova, On equiangular configurations of subspaces of a Hilbert space, \emph{Methods Funct. Anal. Topology} \textbf{12} (2006) 84--96} \expandafter\def\csname PFref@Serre1980\endcsname{J.-P. Serre, \emph{Trees}, translated from the French by J.~Stillwell, Springer, Berlin, 1980; corrected second printing, Springer Monographs in Mathematics, 2003} \expandafter\def\csname PFref@SlomczynskiSzczepanek2018\endcsname{W.~S{\l}omczy\'nski and A.~Szczepanek, Quantum dynamical entropy, chaotic unitaries and complex Hadamard matrices, \emph{IEEE Trans. Inf. Theory} \textbf{63} (2017) 7821--7831. \href{https://doi.org/10.1109/TIT.2017.2751507}{doi:10.1109/TIT.2017.2751507}. \href{https://arxiv.org/abs/1612.03363}{arXiv:1612.03363}} \expandafter\def\csname PFref@Sokal2011\endcsname{A.~D. Sokal, When is a Riesz distribution a complex measure?, \emph{Bull. Soc. Math. France} \textbf{139} (2011) 519--534. \href{https://doi.org/10.24033/bsmf.2617}{doi:10.24033/bsmf.2617}. \href{https://arxiv.org/abs/0906.3027v2}{arXiv:0906.3027v2}} \expandafter\def\csname PFref@Sorkin1994\endcsname{R.~D. Sorkin, Quantum mechanics as quantum measure theory, \emph{Mod. Phys. Lett. A} \textbf{9} (1994) 3119. \href{https://arxiv.org/abs/gr-qc/9401003}{arXiv:gr-qc/9401003}} \expandafter\def\csname PFref@SpringerVeldkamp2000\endcsname{T.~A. Springer and F.~D. Veldkamp, \emph{Octonions, Jordan Algebras and Exceptional Groups}, Springer Monographs in Mathematics, Springer, Berlin, 2000. \href{https://doi.org/10.1007/978-3-662-12622-6}{doi:10.1007/978-3-662-12622-6}} \expandafter\def\csname PFref@SupicBowles2020\endcsname{I.~\v{S}upi\'c and J.~Bowles, Self-testing of quantum systems: a review, \emph{Quantum} \textbf{4} (2020) 337. \href{https://doi.org/10.22331/q-2020-09-30-337}{doi:10.22331/q-2020-09-30-337}. \href{https://arxiv.org/abs/1904.10042}{arXiv:1904.10042}} \expandafter\def\csname PFref@UdudecBarnumEmerson2011\endcsname{C.~Ududec, H.~Barnum and J.~Emerson, Three slit experiments and the structure of quantum theory, \emph{Found. Phys.} \textbf{41} (2011) 396--405. \href{https://doi.org/10.1007/s10701-010-9429-z}{doi:10.1007/s10701-010-9429-z}. \href{https://arxiv.org/abs/0909.4787}{arXiv:0909.4787}} \expandafter\def\csname PFref@vandeWetering\endcsname{J.~van de Wetering, Three characterisations of the sequential product, \emph{J. Math. Phys.} \textbf{59} (2018) 082202. \href{https://doi.org/10.1063/1.5031089}{doi:10.1063/1.5031089}. \href{https://arxiv.org/abs/1803.08453}{arXiv:1803.08453}} \expandafter\def\csname PFref@vandeWetering2019a\endcsname{J.~van de Wetering, Sequential product spaces are Jordan algebras, \emph{J. Math. Phys.} \textbf{60} (2019) 062201. \href{https://doi.org/10.1063/1.5093504}{doi:10.1063/1.5093504}. \href{https://arxiv.org/abs/1803.11139}{arXiv:1803.11139}} \expandafter\def\csname PFref@VitaglianoBudroni2022\endcsname{G.~Vitagliano and C.~Budroni, Leggett--Garg macrorealism and temporal correlations, \emph{Phys. Rev. A} \textbf{107} (2023) 040101. \href{https://doi.org/10.1103/PhysRevA.107.040101}{doi:10.1103/PhysRevA.107.040101}. \href{https://arxiv.org/abs/2212.11616}{arXiv:2212.11616}} \expandafter\def\csname PFref@Vorobev1962\endcsname{N.~N. Vorob'ev, Consistent families of measures and their extensions, \emph{Theory Probab. Appl.} \textbf{7} (1962) 147--163. \href{https://doi.org/10.1137/1107014}{doi:10.1137/1107014}} \expandafter\def\csname PFref@WWvdW2022\endcsname{B.~Westerbaan and J.~van de Wetering, A computer scientist's reconstruction of quantum theory, \emph{J. Phys. A} \textbf{55} (2022) 384002. \href{https://doi.org/10.1088/1751-8121/ac8459}{doi:10.1088/1751-8121/ac8459}. \href{https://arxiv.org/abs/2109.10707}{arXiv:2109.10707}} \expandafter\def\csname PFref@Wigner1963\endcsname{E.~P. Wigner, The problem of measurement, \emph{Am. J. Phys.} \textbf{31} (1963) 6--15} \expandafter\def\csname PFref@Wilce2019\endcsname{A.~Wilce, Conjugates, filters and quantum mechanics, \emph{Quantum} \textbf{3} (2019) 158. \href{https://doi.org/10.22331/q-2019-07-08-158}{doi:10.22331/q-2019-07-08-158}. \href{https://arxiv.org/abs/1206.2897}{arXiv:1206.2897}} \expandafter\def\csname PFref@Wolf1963elliptic\endcsname{J.~A. Wolf, Elliptic spaces in Grassmann manifolds, \emph{Illinois J. Math.} \textbf{7} (1963) 447--462. \href{https://doi.org/10.1215/ijm/1255644952}{doi:10.1215/i{}jm/1255644952}. Correction: D.~B. Shapiro and J.~A. Wolf, On isoclinic closure: correction to ``Elliptic spaces in Grassmann manifolds'', \emph{Illinois J. Math.} \textbf{19} (1975) 316--317. \href{https://doi.org/10.1215/ijm/1256050819}{doi:10.1215/i{}jm/1256050819}} \expandafter\def\csname PFref@Wolf1963geodesic\endcsname{J.~A. Wolf, Geodesic spheres in Grassmann manifolds, \emph{Illinois J. Math.} \textbf{7} (1963) 425--446. \href{https://doi.org/10.1215/ijm/1255644951}{doi:10.1215/i{}jm/1255644951}} \expandafter\def\csname PFref@Wong1961\endcsname{Y.-C. Wong, Isoclinic $n$-planes in Euclidean $2n$-space, Clifford parallels in elliptic $(2n-1)$-space, and the Hurwitz matrix equations, \emph{Mem. Amer. Math. Soc.} \textbf{41} (1961). \href{https://doi.org/10.1090/memo/0041}{doi:10.1090/memo/0041}} \expandafter\def\csname PFref@ZyczkowskiSommers2001\endcsname{K.~\.Zyczkowski and H.-J. Sommers, Induced measures in the space of mixed quantum states, \emph{J. Phys. A} \textbf{34} (2001) 7111--7125. \href{https://doi.org/10.1088/0305-4470/34/35/335}{doi:10.1088/0305-4470/34/35/335}. \href{https://arxiv.org/abs/quant-ph/0012101}{arXiv:quant-ph/0012101}} \expandafter\def\csname PFshort@BrehmEtTaoui2001\endcsname{Brehm and Et-Taoui, \emph{Congruence criteria}} \expandafter\def\csname PFshort@AsoreyScolariciSolombrino2007\endcsname{Asorey, Scolarici and Solombrino, \emph{Complex projections of quaternionic maps}} \expandafter\def\csname PFshort@Catto2003\endcsname{Catto, \emph{Exceptional projective geometries}} \expandafter\def\csname PFshort@MasilloScolariciSozzo2009\endcsname{Masillo, Scolarici and Sozzo, \emph{Proper versus improper mixtures}} \expandafter\def\csname PFshort@AbramskyBrandenburger2011\endcsname{Abramsky and Brandenburger, \emph{The sheaf-theoretic structure}} \expandafter\def\csname PFshort@Albert1934\endcsname{Albert, ``On a certain algebra of quantum mechanics''} \expandafter\def\csname PFshort@AlbertPaige1959\endcsname{Albert and Paige, ``On a homomorphism property of certain Jordan algebras''} \expandafter\def\csname PFshort@AlfsenShultz2003\endcsname{Alfsen and Shultz, \emph{Geometry of State Spaces}} \expandafter\def\csname PFshort@AlfsenShultzStormer1978\endcsname{Alfsen, Shultz and St{\o}rmer, \emph{A Gelfand--Neumark theorem for Jordan algebras}} \expandafter\def\csname PFshort@Araki1980b\endcsname{Araki, \emph{On a characterization of the state space}} \expandafter\def\csname PFshort@Baez2002\endcsname{Baez, ``The octonions''} \expandafter\def\csname PFshort@Baez2012\endcsname{Baez, ``Division algebras and quantum theory''} \expandafter\def\csname PFshort@BaezSchwahn2026\endcsname{Baez and Schwahn, \emph{The Standard Model gauge group}} \expandafter\def\csname PFshort@BarnumHilgert2020\endcsname{Barnum and Hilgert, ``Spectral properties of convex bodies''} \expandafter\def\csname PFshort@BarnumWilce2014\endcsname{Barnum and Wilce, \emph{Local tomography}} \expandafter\def\csname PFshort@BGW2015\endcsname{Barnum, Graydon and Wilce, \emph{Some nearly quantum theories}} \expandafter\def\csname PFshort@BGW2020\endcsname{Barnum, Graydon and Wilce, \emph{Composites and categories}} \expandafter\def\csname PFshort@BMU2014\endcsname{Barnum, M\"uller and Ududec, \emph{Higher-order interference and single-system postulates}} \expandafter\def\csname PFshort@Boyle2020\endcsname{Boyle, \emph{The standard model, the exceptional Jordan algebra, and triality}} \expandafter\def\csname PFshort@BoyleFarnsworth2020\endcsname{Boyle and Farnsworth, \emph{The standard model, the Pati--Salam model, and `Jordan geometry'}} \expandafter\def\csname PFshort@Brehm1990\endcsname{Brehm, \emph{The shape invariant of triangles}} \expandafter\def\csname PFshort@BMKG2013\endcsname{Budroni et al., ``Bounding temporal quantum correlations''} \expandafter\def\csname PFshort@BudroniEmary2014\endcsname{Budroni and Emary, \emph{Temporal quantum correlations}} \expandafter\def\csname PFshort@BudroniFagundesKleinmann2019\endcsname{Budroni, Fagundes and Kleinmann, \emph{Memory cost of temporal correlations}} \expandafter\def\csname PFshort@BuschGrabowskiLahti1995\endcsname{Busch, Grabowski and Lahti, \emph{Repeatable measurements in quantum theory}} \expandafter\def\csname PFshort@BuschSingh1998\endcsname{Busch and Singh, ``L\"uders theorem for unsharp quantum measurements''} \expandafter\def\csname PFshort@CafutaKlepPovh2015\endcsname{Cafuta, Klep and Povh, \emph{Rational sums of hermitian squares}} \expandafter\def\csname PFshort@ChiribellaYuan2014\endcsname{Chiribella and Yuan, \emph{Measurement sharpness}} \expandafter\def\csname PFshort@ChiribellaYuan2018\endcsname{Chiribella and Yuan, \emph{Bridging the gap}} \expandafter\def\csname PFshort@Cho2019\endcsname{Cho, \emph{Effectuses in Categorical Quantum Foundations}} \expandafter\def\csname PFshort@ClementeKofler2016\endcsname{Clemente and Kofler, ``No fine theorem for macrorealism: limitations of the Leggett--Garg inequality''} \expandafter\def\csname PFshort@DakicPaterekBrukner2014\endcsname{Daki\'c, Paterek and Brukner, \emph{Density cubes and higher-order interference theories}} \expandafter\def\csname PFshort@DaviesLewis1970\endcsname{Davies and Lewis, ``An operational approach to quantum probability''} \expandafter\def\csname PFshort@DeMedtsRowenSegev2023\endcsname{De Medts, Rowen and Segev, \emph{Primitive 4-generated axial algebras}} \expandafter\def\csname PFshort@DiazGarciaGutierrezJaimez2010\endcsname{D\'{\i}az-Garc\'{\i}a and Guti\'errez-J\'aimez, ``On Wishart distribution''} \expandafter\def\csname PFshort@DuboisViolette2016\endcsname{Dubois-Violette, ``Exceptional quantum geometry and particle physics''} \expandafter\def\csname PFshort@DuboisVioletteTodorov2018\endcsname{Dubois-Violette and Todorov, ``Exceptional quantum geometry and particle physics~II''} \expandafter\def\csname PFshort@Ehrlich2026\endcsname{Ehrlich, \emph{\texttt{sequential-products-lean}}} \expandafter\def\csname PFshort@Ehrlich2026ej\endcsname{Ehrlich, \emph{\texttt{euclidean-jordan}}} \expandafter\def\csname PFshort@ElduqueIltyakov1999\endcsname{Elduque and Iltyakov, \emph{Polynomial invariants of exceptional simple algebraic groups}} \expandafter\def\csname PFshort@FLW2024\endcsname{Fanizza, Lumbreras and Winter, \emph{Quantum theory in finite dimension}} \expandafter\def\csname PFshort@FarautKoranyi1994\endcsname{Faraut and Kor\'anyi, \emph{Analysis on Symmetric Cones}} \expandafter\def\csname PFshort@FernandesWagnerNovoGalvao2024\endcsname{Fernandes et al., \emph{Unitary-invariant witnesses of quantum imaginarity}} \expandafter\def\csname PFshort@Forrester2016\endcsname{Forrester, ``Octonions in random matrix theory''} \expandafter\def\csname PFshort@GarnerMuller2020\endcsname{Garner and M\"uller, \emph{Characterization of the probabilistic models}} \expandafter\def\csname PFshort@Glennie1966\endcsname{Glennie, ``Some identities valid in special Jordan algebras''} \expandafter\def\csname PFshort@GogicTomasevic2026\endcsname{Gogi\'c and Toma\v{s}evi\'c, \emph{Idempotent words}} \expandafter\def\csname PFshort@GMS2025\endcsname{Gorshkov, Mamontov and Staroletov, \emph{Jordan algebras that are factors of Matsuo algebras}} \expandafter\def\csname PFshort@GorshkovStaroletov2020\endcsname{Gorshkov and Staroletov, \emph{Primitive 3-generated axial algebras}} \expandafter\def\csname PFshort@GudderGreechie2002\endcsname{Gudder and Greechie, ``Sequential products on effect algebras''} \expandafter\def\csname PFshort@GPR1978\endcsname{G\"unaydin, Piron and Ruegg, \emph{Moufang plane and octonionic quantum mechanics}} \expandafter\def\csname PFshort@GunaydinGursey1973\endcsname{G\"unaydin and G\"ursey, ``Quark structure and octonions''} \expandafter\def\csname PFshort@HarveyLawson1982\endcsname{Harvey and Lawson, ``Calibrated geometries''} \expandafter\def\csname PFshort@HassairiLajmi2018\endcsname{Hassairi and Lajmi, ``Singular Riesz measures on symmetric cones''} \expandafter\def\csname PFshort@HeinosaariReitznerStano2008\endcsname{Heinosaari, Reitzner and Stano, \emph{Notes on joint measurability}} \expandafter\def\csname PFshort@Helton2002\endcsname{Helton, \emph{`Positive' noncommutative polynomials}} \expandafter\def\csname PFshort@HeltonMcCullough2004\endcsname{Helton and McCullough, ``A Positivstellensatz for non-commutative polynomials''} \expandafter\def\csname PFshort@HSGB2018\endcsname{Hoffmann et al., \emph{Structure of temporal correlations of a qubit}} \expandafter\def\csname PFshort@Hoggar1977\endcsname{Hoggar, ``New sets of equi-isoclinic $n$-planes from old''} \expandafter\def\csname PFshort@Hsiang1989\endcsname{Hsiang, \emph{Laws of trigonometries}} \expandafter\def\csname PFshort@Iltyakov1997\endcsname{Iltyakov, \emph{Polynomial invariants of group $F_4$}} \expandafter\def\csname PFshort@Jacobson1954\endcsname{Jacobson, ``Structure of alternative and Jordan bimodules''} \expandafter\def\csname PFshort@JordanvNWigner1934\endcsname{Jordan, von Neumann and Wigner, \emph{On an algebraic generalization}} \expandafter\def\csname PFshort@KhandelwalTavakoli2025\endcsname{Khandelwal and Tavakoli, \emph{Simulating quantum instruments}} \expandafter\def\csname PFshort@Kleinmann2014\endcsname{Kleinmann, \emph{Sequences of projective measurements}} \expandafter\def\csname PFshort@KLM2025\endcsname{Klep, Levenson and McCullough, \emph{Fej\'er--Riesz factorization}} \expandafter\def\csname PFshort@Kollross\endcsname{Kollross, \emph{Octonions, triality, the exceptional Lie algebra $F_4$}} \expandafter\def\csname PFshort@Krasnov2019\endcsname{Krasnov, ``SO(9) characterization of the Standard Model gauge group''} \expandafter\def\csname PFshort@Luders1951\endcsname{L\"uders, ``\"Uber die Zustands\"anderung durch den Me{\ss}proze{\ss}''} \expandafter\def\csname PFshort@MacGregorPhi2026\endcsname{MacGregor, ``Finite observations and recognition of the compact Albert algebra''} \expandafter\def\csname PFshort@McCrimmon2004\endcsname{McCrimmon, \emph{A Taste of Jordan Algebras}} \expandafter\def\csname PFshort@McCrimmonIdentities2004\endcsname{McCrimmon, ``The role of identities in Jordan algebras''} \expandafter\def\csname PFshort@MonrasWinter2016\endcsname{Monr\`as and Winter, \emph{Quantum learning of classical stochastic processes}} \expandafter\def\csname PFshort@MullerGarner2023\endcsname{M\"uller and Garner, \emph{Testing quantum theory}} \expandafter\def\csname PFshort@NPA2008\endcsname{Navascu\'es, Pironio and Ac\'in, \emph{A convergent hierarchy}} \expandafter\def\csname PFshort@Niestegge2008\endcsname{Niestegge, \emph{Quantum measurement in order-unit spaces}} \expandafter\def\csname PFshort@Niestegge2012\endcsname{Niestegge, \emph{Conditional probability, three-slit experiments}} \expandafter\def\csname PFshort@Niestegge2020a\endcsname{Niestegge, \emph{A simple and quantum-mechanically motivated characterization}} \expandafter\def\csname PFshort@Niestegge2020b\endcsname{Niestegge, \emph{Quantum probability's algebraic origin}} \expandafter\def\csname PFshort@Niestegge2026a\endcsname{Niestegge, \emph{Homogeneity of the quantum transition probability}} \expandafter\def\csname PFshort@Nasmith2022\endcsname{Nasmith, ``Octonions and the two strictly projective tight 5-designs''} \expandafter\def\csname PFshort@Nasmith2022integers\endcsname{Nasmith, \emph{Octonion integers and tight 5-designs}} \expandafter\def\csname PFshort@Hoehn2026lattice\endcsname{H\"ohn, \emph{The unique extremal threemodular lattice of rank 26}} \expandafter\def\csname PFshort@Hoehn2026albert\endcsname{H\"ohn, \emph{Definite integral Albert algebras}} \expandafter\def\csname PFshort@Niestegge2026b\endcsname{Niestegge, \emph{The bare necessities}} \expandafter\def\csname PFshort@OrtegaSantander2002\endcsname{Ortega and Santander, \emph{Trigonometry of `complex Hermitian' type}} \expandafter\def\csname PFshort@OszmaniecBrodGalvao2021\endcsname{Oszmaniec, Brod and Galv\~ao, \emph{Measuring relational information}} \expandafter\def\csname PFshort@Ozawa1984\endcsname{Ozawa, ``Quantum measuring processes of continuous observables''} \expandafter\def\csname PFshort@Peres1979\endcsname{Peres, ``Proposed test for complex versus quaternion quantum theory''} \expandafter\def\csname PFshort@PinedaRua2026\endcsname{Pineda Montoya and Rua Munoz, \emph{Operational foundations}} \expandafter\def\csname PFshort@PeyrlParrilo2008\endcsname{Peyrl and Parrilo, \emph{Computing sum of squares decompositions}} \expandafter\def\csname PFshort@PNA2010\endcsname{Pironio, Navascu\'es and Ac\'in, \emph{Convergent relaxations}} \expandafter\def\csname PFshort@PopovaSamoilenko2006\endcsname{Popova and Samoilenko, \emph{Configurations of subspaces with fixed angles}} \expandafter\def\csname PFshort@Racine1977\endcsname{Racine, ``Maximal subalgebras of exceptional Jordan algebras''} \expandafter\def\csname PFshort@RoelandsTiersma2026\endcsname{Roelands and Tiersma, ``On the Shirshov--Cohn theorem for JB-algebras''} \expandafter\def\csname PFshort@SahaSantosAugusiak2020\endcsname{Saha, Santos and Augusiak, \emph{Sum-of-squares decompositions}} \expandafter\def\csname PFshort@SahaXuCabello2026\endcsname{Saha, Xu and Cabello, \emph{Self-testing ideal quantum measurements}} \expandafter\def\csname PFshort@SalamonWalpuski2017\endcsname{Salamon and Walpuski, ``Notes on the octonions''} \expandafter\def\csname PFshort@SamoilenkoYershova2006\endcsname{Samoilenko and Yershova, \emph{Equiangular configurations of subspaces}} \expandafter\def\csname PFshort@SlomczynskiSzczepanek2018\endcsname{S{\l}omczy\'nski and Szczepanek, \emph{Quantum dynamical entropy}} \expandafter\def\csname PFshort@Sokal2011\endcsname{Sokal, ``When is a Riesz distribution a complex measure?''} \expandafter\def\csname PFshort@Sorkin1994\endcsname{Sorkin, ``Quantum mechanics as quantum measure theory''} \expandafter\def\csname PFshort@Serre1980\endcsname{Serre, \emph{Trees}} \expandafter\def\csname PFshort@SpringerVeldkamp2000\endcsname{Springer and Veldkamp, \emph{Octonions, Jordan Algebras and Exceptional Groups}} \expandafter\def\csname PFshort@SupicBowles2020\endcsname{\v{S}upi\'c and Bowles, ``Self-testing of quantum systems: a review''} \expandafter\def\csname PFshort@UdudecBarnumEmerson2011\endcsname{Ududec, Barnum and Emerson, \emph{Three slit experiments}} \expandafter\def\csname PFshort@vandeWetering\endcsname{van de Wetering, ``Three characterisations of the sequential product''} \expandafter\def\csname PFshort@vandeWetering2019a\endcsname{van de Wetering, ``Sequential product spaces are Jordan algebras''} \expandafter\def\csname PFshort@VitaglianoBudroni2022\endcsname{Vitagliano and Budroni, ``Leggett--Garg macrorealism and temporal correlations''} \expandafter\def\csname PFshort@Vorobev1962\endcsname{Vorob'ev, ``Consistent families of measures and their extensions''} \expandafter\def\csname PFshort@WWvdW2022\endcsname{Westerbaan and van de Wetering, ``A computer scientist's reconstruction of quantum theory''} \expandafter\def\csname PFshort@Wigner1963\endcsname{Wigner, ``The problem of measurement''} \expandafter\def\csname PFshort@Wilce2019\endcsname{Wilce, ``Conjugates, filters and quantum mechanics''} \expandafter\def\csname PFshort@Wolf1963elliptic\endcsname{Wolf, ``Elliptic spaces in Grassmann manifolds''} \expandafter\def\csname PFshort@Wolf1963geodesic\endcsname{Wolf, ``Geodesic spheres in Grassmann manifolds''} \expandafter\def\csname PFshort@Wong1961\endcsname{Wong, \emph{Isoclinic $n$-planes}} \expandafter\def\csname PFshort@ZyczkowskiSommers2001\endcsname{\.Zyczkowski and Sommers, \emph{Induced measures}} \expandafter\def\csname PFref@Clerc2022\endcsname{J.-L. Clerc, Unitarizable vector-valued holomorphic discrete series and the Laplace transform: an example, \emph{J. Lie Theory} \textbf{33} (2023) 133--148. \href{https://doi.org/10.5802/jolt.1277}{doi:10.5802/jolt.1277}. \href{https://arxiv.org/abs/2206.07407v2}{arXiv:2206.07407v2}. Section~1 (scalar Riesz measures)} \expandafter\def\csname PFref@CorreiaPachecoSvensson2021\endcsname{N.~Correia, R.~Pacheco and M.~Svensson, Harmonic surfaces in the Cayley plane, \emph{J. London Math. Soc.} (2) \textbf{103} (2021) 353--371. \href{https://doi.org/10.1112/jlms.12376}{doi:10.1112/jlms.12376}} \expandafter\def\csname PFref@FSW2018\endcsname{M.~Freedman, M.~Shokrian-Zini and Z.~Wang, Quantum computing with octonions, \emph{Peking Math. J.} \textbf{2} (2019) 239--273. \href{https://doi.org/10.1007/s42543-019-00020-3}{doi:10.1007/s42543-019-00020-3}. \href{https://arxiv.org/abs/1811.08580}{arXiv:1811.08580}} \expandafter\def\csname PFref@Iltyakov1997laplace\endcsname{A.~V. Iltyakov, Laplace operator and polynomial invariants, University of Sydney Research Report 97-23, 1997} \expandafter\def\csname PFref@KnarrStroppel2011\endcsname{N.~Knarr and M.~Stroppel, Polarities and planar collineations of Moufang planes, \emph{Monatsh. Math.} \textbf{169} (2013) 383--395. \href{https://doi.org/10.1007/s00605-012-0409-6}{doi:10.1007/s00605-012-0409-6}} \expandafter\def\csname PFref@KunjwalHeunenFritz2014\endcsname{R.~Kunjwal, C.~Heunen and T.~Fritz, Quantum realization of arbitrary joint measurability structures, \emph{Phys. Rev. A} \textbf{89} (2014) 052126. \href{https://doi.org/10.1103/PhysRevA.89.052126}{doi:10.1103/PhysRevA.89.052126}. \href{https://arxiv.org/abs/1311.5948}{arXiv:1311.5948}} \expandafter\def\csname PFref@Core\endcsname{R.~R. MacGregor, On sequential ideal measurements in the exceptional Jordan algebra, AI proof note, 18 November 2026} \expandafter\def\csname PFref@Pickert1955\endcsname{G.~Pickert, \emph{Projektive Ebenen}, Grundlehren der mathematischen Wissenschaften, Springer, Berlin, 1955. \href{https://doi.org/10.1007/978-3-662-00110-3}{doi:10.1007/978-3-662-00110-3}} \expandafter\def\csname PFref@vanGaansKalauchRoelands2024\endcsname{O.~van Gaans, A.~Kalauch and M.~Roelands, Order theoretical structures in atomic JBW-algebras: disjointness, bands, and centres, \emph{Positivity} \textbf{28} (2024) 10. \href{https://doi.org/10.1007/s11117-023-01024-1}{doi:10.1007/s11117-023-01024-1}. \href{https://arxiv.org/abs/2305.19086}{arXiv:2305.19086}} \expandafter\def\csname PFshort@Clerc2022\endcsname{Clerc, \emph{Unitarizable vector-valued holomorphic discrete series}} \expandafter\def\csname PFshort@CorreiaPachecoSvensson2021\endcsname{Correia, Pacheco and Svensson, \emph{Harmonic surfaces in the Cayley plane}} \expandafter\def\csname PFshort@FSW2018\endcsname{Freedman, Shokrian-Zini and Wang, \emph{Quantum computing with octonions}} \expandafter\def\csname PFshort@Iltyakov1997laplace\endcsname{Iltyakov, \emph{Laplace operator and polynomial invariants}} \expandafter\def\csname PFshort@KnarrStroppel2011\endcsname{Knarr and Stroppel, \emph{Polarities and planar collineations of Moufang planes}} \expandafter\def\csname PFshort@KunjwalHeunenFritz2014\endcsname{Kunjwal, Heunen and Fritz, \emph{Quantum realization of arbitrary joint measurability structures}} \expandafter\def\csname PFshort@Core\endcsname{MacGregor, \emph{On sequential ideal measurements}} \expandafter\def\csname PFshort@Pickert1955\endcsname{Pickert, \emph{Projektive Ebenen}} \expandafter\def\csname PFshort@vanGaansKalauchRoelands2024\endcsname{van Gaans, Kalauch and Roelands, \emph{Order theoretical structures}} \expandafter\def\csname PFref@Schwarz1988\endcsname{G.~W. Schwarz, Invariant theory of $G_2$ and $\mathrm{Spin}_7$, \emph{Comment. Math. Helv.} \textbf{63} (1988) 624--663. \href{https://doi.org/10.1007/BF02566782}{doi:10.1007/BF02566782}} \expandafter\def\csname PFref@Lean4\endcsname{L.~de Moura and S.~Ullrich, The Lean~4 theorem prover and programming language, in \emph{Automated Deduction -- CADE~28}, Lecture Notes in Comput. Sci. \textbf{12699}, Springer, Cham, 2021, 625--635. \href{https://doi.org/10.1007/978-3-030-79876-5_37}{doi:10.1007/978-3-030-79876-5\_37}} \expandafter\def\csname PFref@Mathlib2020\endcsname{The mathlib Community, The Lean mathematical library, in \emph{Proc. 9th ACM SIGPLAN Int. Conf. on Certified Programs and Proofs (CPP 2020)}, ACM, 2020, 367--381. \href{https://doi.org/10.1145/3372885.3373824}{doi:10.1145/3372885.3373824}. \href{https://arxiv.org/abs/1910.09336}{arXiv:1910.09336}} \expandafter\def\csname PFshort@Schwarz1988\endcsname{Schwarz, ``Invariant theory of $G_2$ and $\mathrm{Spin}_7$''} \expandafter\def\csname PFshort@Lean4\endcsname{de Moura and Ullrich, ``The Lean~4 theorem prover''} \expandafter\def\csname PFshort@Mathlib2020\endcsname{The mathlib Community, ``The Lean mathematical library''} \expandafter\def\csname PFref@Springer1960\endcsname{T.~A. Springer, The projective octave plane. I, II, \emph{Indag. Math.} \textbf{22} (1960) 74--101} \expandafter\def\csname PFshort@Springer1960\endcsname{Springer, ``The projective octave plane''} \makeatletter \let\PFfootnote\footnote \newcommand{\PFlastcite}{} \pretocmd{\footnote}{\gdef\PFlastcite{}}{}{} \newcounter{PFnoterefcount} \newcommand{\PFciteone}[1]{\ifcsdef{PFref@#1}{}{\PackageWarning{PFcite}{No reference text for #1}}\PFfirstfalse \ifcsdef{PFseen@#1}{\ifPFintable\edef\PFtmp{\thetable}\ifcsequal{PFfirsttab@#1}{PFtmp}{\PFfirsttrue}{}\fi}{\PFfirsttrue}\ifPFfirst \Hy@raisedlink{\hypertarget{PFcite.#1}{}}\csname PFref@#1\endcsname \csxdef{PFptr@#1}{\PFhere}\csgdef{PFseen@#1}{1}\ifPFintable\csxdef{PFfirsttab@#1}{\thetable}\fi \else \csname PFshort@#1\endcsname, \hyperlink{PFcite.#1}{\csname PFptr@#1\endcsname}\fi} \newif\ifPFfirst\newif\ifPFintable \newcommand{\PFnoteref}[1]{\ifnum\value{PFnoterefcount}>0 ; \fi \stepcounter{PFnoterefcount}\PFciteone{#1}} \newcommand{\PFhere}{} \renewcommand{\cite}[2][]{\unskip\PFfootnote{\edef\PFhere{note~\the\value{footnote}}\def\PFcurrentcite{#2}\ifx\PFcurrentcite\PFlastcite Ibid.\else \setcounter{PFnoterefcount}{0}\forcsvlist{\PFnoteref}{#2}.\fi \if\relax\detokenize{#1}\relax\else\space #1.\fi \setcounter{PFnoterefcount}{0}\forcsvlist{\PFcountcite}{#2}\ifnum\value{PFnoterefcount}=1 \gdef\PFlastcite{#2}\else\gdef\PFlastcite{}\fi}} \newcommand{\PFcountcite}[1]{\stepcounter{PFnoterefcount}} \newcommand{\PFtablecite}[2][]{\edef\PFhere{Table~\thetable}\setcounter{PFnoterefcount}{0}[\forcsvlist{\PFnoteref}{#2}\if\relax\detokenize{#1}\relax\else, #1\fi]} \AtBeginEnvironment{table}{\let\cite\PFtablecite\PFintabletrue\gdef\PFlastcite{}} \AtEndEnvironment{table}{\gdef\PFlastcite{}} \makeatother \begin{document} \newgeometry{hmargin={1.55in,1.55in},top=1.1in,bottom=1.0in} \pagenumbering{roman} \thispagestyle{empty} \begingroup \raggedright \setlength{\parindent}{0pt} \setlength{\parskip}{0.55\baselineskip} {\sffamily\footnotesize AI proof note\hfill 18 November 2026}\par \vspace{0.5in} {\Large Finite certificates for sequential measurements\\[0.25ex] in the exceptional Jordan algebra\par} \vspace{0.08in} {\small Companion paper to \emph{On sequential ideal measurements in the exceptional Jordan algebra}\par} \vspace{0.12in} {\large Robert R. MacGregor\par} \vspace{0.35in} \coverhead{Abstract} Ideal yes--no tests of four pure states that generate the Albert algebra $H_3(\mathbb O)$ have no model by ideal measurements on a real, complex or quaternionic Hilbert space. A fixed list of $2{,}041$ outcome sequences of length at most eleven witnesses this for every such quadruple from the maximally mixed state, four relabelled copies do so from every state, and a linear inequality on these sequences separates with a positive margin. For two explicit quadruples an inequality on sequences of length at most seven has an explicit rational margin, which survives in reduced form for generalised L\"uders instruments of nearly projective effects. No linear equality on sequences of length at most five separates any four pure states from any state, and along a path of pure states leaving a complex subalgebra every equality valid for ideal measurements has a residual of second order, and of third order from states in the subalgebra. For an explicit family of five tests, every four of which have a model, the least separating length is ten from one of the five pure states and eleven from the maximally mixed state. Without the projection postulate, the tests are ideal on the closed convex hull of the states reachable from the maximally mixed state, a completely positive realization of fixed dimension exists exactly when one exists for every finite set of tests and preparations, and $n\ge1$ tests have a realization with one fixed projection per test in dimension $26n-7$. A uniformly random pure state, projected onto a complex subalgebra, has the law of the reduced state on $\mathbb C^3$ of a uniformly random pure state on $\mathbb C^3\otimes\mathbb C^4$. \vspace{0.3in} \coverhead{AI Statement} All text, mathematics and proofs herein, and all Lean~4 code apart from the adapted files noted in the core paper, were produced by Claude Fable~5.2, Claude Opus~5.5, GPT-6.2 Astra, GPT-6.2 Sol, Gemini Carbon and Gemini Hydrogen, and select theorems are machine-checked in Lean. \emph{Except for this statement, all content herein was generated by AI and has not been understood without AI interpretation.} \vfill {\sffamily\footnotesize Text CC BY 4.0\quad Lean code Apache-2.0\\ Lean repository: \leanrepo\\ ORCID: \href{https://orcid.org/0009-0003-1734-8054}{0009-0003-1734-8054}\par} \endgroup \clearpage \restoregeometry \pagenumbering{arabic} \pagenumbering{roman}\setcounter{page}{2} {\small\tableofcontents} \clearpage \pagenumbering{arabic} \section{Introduction}\label{sec:intro} A yes--no test on a quantum system is \emph{ideal} if it disturbs the state no more than its answer requires; in Hilbert space this is the L\"uders rule on both outcomes. The core paper, \emph{On sequential ideal measurements in the exceptional Jordan algebra}~\cite{Core}, shows that ideal tests of pure states (atoms) of the Albert algebra $\J=H_3(\Oct)$ have a model by ideal measurements on a real, complex or quaternionic Hilbert space, of any dimension and from any state, exactly when the atoms generate a proper subalgebra of $\J$. Three atoms never generate $\J$; four atoms generate it exactly when a single invariant $\Acal$ is positive; and for a generating quadruple, from the maximally mixed state $e/3$, the words of length at most six are matched by a mixture of two quaternionic models, while no model by ideal measurements matches all words of length at most seven. The present paper collects what follows from those results when one asks for explicit finite witnesses, for robustness, and for weaker measurement models. Section~\ref{sec:cer} gives a list of $2{,}041$ words of length at most eleven that excludes every sequential projective representation for every generating quadruple of atoms, with relabelled copies that work from every state, the soundness of sum-of-squares certificates, an inequality of length seven with explicit margin for two fixed configurations, and for one of them an inequality of length eleven that fails by at least $\frac7{828}$ from every state. Section~\ref{sec:eq} shows that no linear equality between word probabilities separates through length five, and that near a complex subalgebra every such equality is weak to second order. Section~\ref{sec:five} shows that words of length eleven exclude a family of five tests every four of which have a model, that eleven is the least such length from $e/3$, and that the Helly number of the quaternionic subalgebras is six. Section~\ref{sec:c:four} adds remarks on three and four atoms and the law of a random atom seen in a complex subalgebra. Section~\ref{sec:c:sharp} leaves the projection postulate and treats instruments with arbitrary channels: ideality holds on the states reachable from $e/3$, the existence of a realization of fixed dimension is decided on finite menus (finite sets of preparation and test labels), and sharp realizations need only $26n-7$ dimensions for $n\ge1$ tests. Section~\ref{sec:prior} discusses related work. \paragraph{Conventions.} Notation and terminology are those of the core paper, set out in \corename{sec:set} and \corename{app:conv}. Results of the core paper are cited by name and number, as in \coreref{thm:tetra}, and are not proved again; numbered results are those of the present paper. A \emph{sequential projective representation} (class~(I) of the core paper) consists of a real, complex or quaternionic Hilbert space of any dimension, one orthogonal projection of any rank per test, the L\"uders update on both outcomes, and any state; a \emph{completely positive realization} (class~(II)) replaces the projections and L\"uders updates by arbitrary completely positive instruments. Words are finite sequences of test outcomes; $\Pr_\rho(w)$ is the octonionic probability of the word $w$ from the state $\rho$ of $\J$, and $L_{\min}$ is the least length $L$ such that no sequential projective representation reproduces the octonionic probabilities from $e/3$ of all words of length at most $L$; Corollary~\ref{cor:five:minimum} writes $L_{\min}(\rho)$ for other initial states. \paragraph{Proofs and formalization.} Every result has a written proof, or a proof sketch where the text says so. Margin notes name the Lean declarations behind the formalized results, using the conventions of the core paper; Appendix~\ref{app:c:lean} lists them, and results without a note are not formalized. \section{Finite certificates for four tests}\label{sec:cer} This section proves results on four tests of atoms that the core paper states or uses with a proof only in outline: a list of words of length at most eleven that works for every generating quadruple, the soundness lemma behind every sum-of-squares certificate, explicit error margins for the two configurations $\mathsf M$ (the maximiser of $\Acal$) and $\mathsf S$ (a rational variant of it), and a model that matches all thirteen traces while violating a purity relation. The model class is class~(I). Notation follows \coreref{sec:cer}: reduced words, sequential products $E_w$, word operators $P_w$, the real cyclic representation $(\pi,\Omega)$, the purity operators $z_q(v)$ of \coreref{lem:purity}, the relations (R1)--(R3) of \coreref{lem:fp:access}, and the holonomies $h_a=H_a+N_a$, numbers $c_a$, $S$, matrix $M$ and volume element $\Gamma=\frac12(N_1N_2N_3-N_3N_2N_1)$ of \coreref{sec:fp:tetra}. \subsection{The tetrahedron identity on finitely many vectors}\label{sec:cer:finite} As explained in \coreref{sec:cer:eleven}, a finite list of probabilities gives each relation only on vectors on which it has been tested, so the proof of \coreref{thm:tetra} is rerun in that form. The base $j=A$, the order $(k,l,m)=(B,C,D)$ and the holonomy degree of a vector $\pi(P_t)\Omega$ are those of the core paper. \begin{lemma}[Finite-vector form of the tetrahedron identity]\label{lem:cer:finite} Let $P_j,P_k,P_l,P_m$ be orthogonal projections on a real Hilbert space $\mathcal K$, let $\Omega\in\mathcal K$ and $\xi=P_j\Omega$, and let $o_{ab}>0$, $R_{abc}$ and $Q_{j;\cdot}$ be real numbers. Form $h_a,H_a,N_a,\Gamma$ from the projections and $c_a,M,S,\Dtet$ from the numbers as in \coreref{sec:fp:tetra}. For $d=0,1,2$ let $\mathcal K_d$ be the span of the vectors $u_1\cdots u_r\xi$ with $r\le d$ and $u_s\in\{h_1,h_2,h_3,h_1^*,h_2^*,h_3^*\}$. Suppose \begin{enumerate}[label=(\roman*),leftmargin=2em] \item $H_a\eta=c_a\eta$ for all $\eta\in\mathcal K_2$ and $a=1,2,3$; \item $h_a^*h_a\eta=h_ah_a^*\eta=\eta$ for all $\eta\in\mathcal K_2$ and $a=1,2,3$; \item $(N_aN_b+N_bN_a)\eta=-2M_{ab}\eta$ for all $\eta\in\mathcal K_1$ and $a\ne b$; \item $\frac12\bigl(h_1h_2h_3+(h_1h_2h_3)^*\bigr)\xi=c_4\xi$. \end{enumerate} Then \begin{equation}\label{eq:cer:data} \Gamma\xi=(c_4-S)\,\xi \end{equation} and \begin{equation}\label{eq:cer:gram} \|\Gamma\xi\|^2=\det M\,\|\xi\|^2 . \end{equation} Hence $\Dtet\,\|\xi\|^2=0$. \end{lemma} \begin{proof} First, (i) and (ii) give the square relations $N_a^2\eta=-M_{aa}\eta$ on $\mathcal K_2$: since $h_a=H_a+N_a$ and $h_a^*=H_a-N_a$, $N_a^2=H_a^2-\frac12(h_ah_a^*+h_a^*h_a)$, and (i) applied twice gives $H_a^2\eta=c_a^2\eta$. Each $N_a$ maps $\mathcal K_d$ into $\mathcal K_{d+1}$. Expanding $h_1h_2h_3\xi=(c_1+N_1)(c_2+N_2)(c_3+N_3)\xi$ uses (i) on vectors of degree at most two, and likewise for $(h_1h_2h_3)^*\xi=(c_3-N_3)(c_2-N_2)(c_1-N_1)\xi$. In half the sum the terms linear in the $N_a$ cancel, and (iii) on $\xi$ turns the quadratic terms into $-c_1M_{23}-c_2M_{13}-c_3M_{12}$. So the left side of (iv) is $(S+\Gamma)\xi$, and (iv) gives \eqref{eq:cer:data}. On the other hand, moving the exactly skew operators $N_a$ across inner products and using the square relations on degree at most two and (iii) on degree at most one, the Gram matrix of $\xi$, $N_a\xi$ and $N_1N_2N_3\xi$ gives \eqref{eq:cer:gram}; the computation is in Appendix~\ref{app:cert:gram}, and the undetermined number $\ip\xi{N_1N_2N_3\xi}$ drops out. Comparing, $\Dtet\|\xi\|^2=(\det M-(c_4-S)^2)\|\xi\|^2=0$. \end{proof} \subsection{Words of length eleven for every configuration}\label{sec:cer:eleven} Table~\ref{tab:cer:ledger} gives the relation applications of \coreref{sec:cer:eleven}, keyed to the hypotheses of Lemma~\ref{lem:cer:finite} on $\xi=P_j\Omega$. The list $\mathcal F_A$ is the set of all words these applications require, after merging adjacent equal letters and removing duplicates. There are $1{,}673$ applications and $2{,}041$ distinct words; Appendix~\ref{app:cert:count} lists them as \eqref{eq:cer:elevenlist} and counts them. \begin{table}[htbp]\centering\footnotesize \caption{Where the eleven letters go (base $j$). A relation (R1), (R2), (R3) applied to a vector whose word has length $n$ and ends in the base of the relation needs words of length up to $n+2$, $n+4$, $n+6$ (one more if the word does not end in the base).} \label{tab:cer:ledger} \begin{tabularx}{\linewidth}{@{}>{\raggedright\arraybackslash}p{0.22\linewidth}>{\raggedright\arraybackslash}p{0.17\linewidth}>{\raggedright\arraybackslash}X r@{}}\toprule Hypothesis of Lemma~\ref{lem:cer:finite} & Relation & Vectors & Longest word\\\midrule (i) $H_a=c_a$ & (R2) at $j$ & holonomy degree $\le2$ (length $\le7$) & $11$\\ (ii) $h_a$ unitary & (R1) $\times3$ & holonomy degree $\le2$, and one letter more & $11$\\ (iii) Clifford relations & (R1), (R3) at $j$ & holonomy degree $\le1$ (length $\le4$) & $10$\\ (iv) fourth face & (R2) at $k$, (R1) & $P_kP_j\Omega$, $P_mP_kP_j\Omega$, $P_lP_kP_j\Omega$, $\xi$ & $6$\\ norm of $\xi$ & none & $P_j\Omega$ & $1$\\\bottomrule \end{tabularx} \end{table} \provisionalref{Length eleven}{thm:eleven}{Different form}{\leandecl{BlockB.thm_eleven_with_overlaps}, which joins \leandecl{BlockB.thm_eleven_full} and the side assertion \leandecl{MatchingModels.overlap_pos_lt_one_of_fourPoint_pos}; the counts: \leandecl{Ends.Eleven.elevenBase_card}, \leandecl{Ends.Eleven.elevenBase_hist}, \leandecl{BlockB.elevenUnion_card}. In Lean the list is generated from an encoding of the 1,673 relation applications behind Table~\ref{tab:cer:ledger} (\leandecl{BlockB.elevenLedgerWords_eq_elevenBase}).} \begin{theorem}[Words of length eleven suffice]\label{thm:eleven} Let $\mathcal F_A$ be the list of $2{,}041$ reduced words in four tests $A,B,C,D$ defined above from Table~\ref{tab:cer:ledger} and printed as \eqref{eq:cer:elevenlist} in Appendix~\ref{app:cert:count}; each begins with ``yes'' on $A$ and has length at most eleven. Let $p_A,p_B,p_C,p_D$ be atoms of $\J$ with $\Acal>0$, equivalently (\coreref{thm:fourpoint}) atoms generating $\J$; then all six transition probabilities lie in $(0,1)$. The model class is (I). \begin{enumerate}[label=(\alph*),leftmargin=2em] \item No sequential projective representation reproduces $\Pr_{e/3}(w)$ for all $w\in\mathcal F_A$. \item Let $\mathcal F_B,\mathcal F_C,\mathcal F_D$ be obtained from $\mathcal F_A$ by exchanging the label $A$ with $B$, $C$, $D$, and $\mathcal F=\mathcal F_A\cup\dots\cup\mathcal F_D$ ($8{,}164$ words). For every state $\rho$ of $\J$, no sequential projective representation reproduces $\Pr_\rho(w)$ for all $w\in\mathcal F$. \item There are a linear functional $\Lambda$ on the probabilities of the words of $\mathcal F$ and the empty word, and $\gamma>0$, such that $\Lambda\ge0$ on every sequential projective representation and $\Lambda\le-\gamma$ on the octonionic table from every state. \end{enumerate} The list does not depend on the atoms. Its words number $1$, $3$, $15$, $36$, $32$, $94$, $240$, $180$, $432$, $576$, $432$ by length $1,\dots,11$. \end{theorem} The theorem is \coreref{thm:eleven}, where its proof is given only in outline. \begin{proof} (a) Suppose a representation reproduces the list, and pass to its real cyclic representation, with $\xi=P_j\Omega$ and $\|\xi\|^2=\Pr(j)=\frac13$. All transition probabilities lie in $(0,1)$ because $\Acal>0$ (proof of \coreref{cor:fp:noideal}), so the numbers of \coreref{sec:fp:tetra} are defined. Each of (R1), (R2), (R3) is a signed sum of purity operators $z_q(v)$ with $|v|\le1,2,3$ respectively (proof of \coreref{lem:fp:access}). The argument applies such a relation to a vector $\pi(P_t)\Omega$ only when the three words $(t,q,v,\check v,q)$, $(t,q,v)$, $(t,q)$ are on the list, so that \coreref{lem:purity} applies; for $t$ of length $n$ ending in the base of the relation this costs words of length $n+2$, $n+4$, $n+6$. The hypotheses of Lemma~\ref{lem:cer:finite} follow, as in the proof of \coreref{thm:tetra} but on vectors only. (i) is (R2) at $j$ on vectors of holonomy degree at most two (words of length at most seven), hence length at most eleven. (ii) is three contractions by (R1) on those vectors and on the vectors one letter longer, length at most eleven. (iii) is one contraction by (R1) and (R3) at $j$ on vectors of degree at most one (words of length at most four), length at most ten. (iv) is four contractions by (R1) on $P_kP_j\Omega$, $P_mP_kP_j\Omega$, $P_lP_kP_j\Omega$ (two in $h_1h_2h_3\xi$, two in its adjoint), (R2) at $k$, and one contraction by (R1) at $j$ on $\xi$, length at most six. Lemma~\ref{lem:cer:finite} gives $\Dtet\|\xi\|^2=0$, so $\Dtet=0$, and \coreref{prop:Dacal} gives $\Acal=\Dtet\prod o_{ab}=0$, a contradiction. (b) Let $y=p_A+p_B+p_C+p_D$. If $y$ were not invertible, an atom $r$ with $\ip ry=0$ would have $\ip r{p_i}=0$ for each $i$, so every $p_i$ would lie in $\J_0(r)$ (\coreref{lem:set:support}) and the atoms would generate a subalgebra of the proper subalgebra $\R r\oplus\J_0(r)$, against {\coreref[(iii)]{thm:fourpoint}}. So $\ip\rho y>0$ for every state, and some $j$ has $\ip\rho{p_j}>0$. Every word of $\mathcal F_j$ begins with ``yes'' on $j$, after which the octonionic state is $\ip\rho{p_j}p_j$; so on $\mathcal F_j$ the table from $\rho$ is $3\ip\rho{p_j}$ times the table from $e/3$, every combination used in (a) still vanishes, and the argument ends with $\Dtet\ip\rho{p_j}=0$. The four lists have different first letters, so $|\mathcal F|=4\cdot2{,}041$. (c) Appendix~\ref{app:cert:quantitative} retains the errors in the finite-vector proof and defines, by explicit symbolic substitutions on this same ledger, a functional supported on $\mathcal F\cup\{\varnothing\}$. It is nonnegative on every sequential projective representation and at most $-\gamma$ on the octonionic table from every state, where $\gamma=\Dtet\det(y)/8>0$ and $y=\sum_jp_j$. The construction gives a conservative coefficient-norm bound without conditioning on a preparation probability. \end{proof} \subsection{Soundness of sum-of-squares certificates}\label{sec:cer:sound} Write $\R[\mathbb Z_2^{*n}]$ for the group algebra of the free product of $n$ copies of $\mathbb Z_2$, with generators $r_i$, and $E(\Lambda)\in\R[\mathbb Z_2^{*n}]$ and $\theta(\Lambda)\in\J$ for the images of a combination $\Lambda=\sum_w\lambda_w[w]$ of words, as in \coreref{sec:cer:frame}. \provisionalref{Soundness}{lem:cer:sound}{Lean scope narrower}{\leandecl{AlgebraicProjectiveData.evaluatedSquareIdentity_sound}. The first assertion, with the identity required only after evaluation in the model; the version with a prefixed ``yes'' is not stated.} \begin{lemma}[Soundness]\label{lem:cer:sound} Suppose that \begin{equation}\label{eq:cer:sosform} E(\Lambda)=\sum_{k=1}^K \beta_k\,F_k^*F_k\qquad\text{in }\R[\mathbb Z_2^{*n}], \end{equation} with rational $\beta_k>0$ and $F_k\in\R[\mathbb Z_2^{*n}]$. Then $\sum_w\lambda_w\varphi(E_w)\ge0$ for every sequential projective representation, over $\R$, $\C$ or $\Ha$, in every dimension, with projections of any rank and any state $\varphi$. The same holds for the combination $\sum_w\lambda_w[(i,w)]$ obtained by prefixing a ``yes'' on test $i$ to every word. \end{lemma} \begin{proof} $r_i\mapsto \Sigma_i$ extends to a $*$-homomorphism from $\R[\mathbb Z_2^{*n}]$ to the bounded operators, so $\sum_w\lambda_w\varphi(E_w)=\sum_k\beta_k\varphi(F_k^*F_k)\ge0$. For the prefixed combination, $E_{(i,w)}=P_iE_wP_i$ and $P_iF_k^*F_kP_i=(F_kP_i)^*(F_kP_i)$. \end{proof} An identity \eqref{eq:cer:sosform} is a \emph{sum-of-squares certificate} for the inequality $\sum_w\lambda_w\varphi(E_w)\ge0$. This is the standard form of a certificate in noncommutative polynomial optimisation (Helton~\cite{Helton2002}, Helton and McCullough~\cite{HeltonMcCullough2004}, Pironio, Navascu\'es and Ac\'{\i}n~\cite{PNA2010}), and for sequential measurements it is dual to the moment matrix of \coreref{lem:cer:recon}~\cite{BMKG2013}. For the algebra of $n$ projections a separating inequality, once it exists, always has such a certificate of bounded degree: Klep, Levenson and McCullough~\cite[Thm.~1.6]{KLM2025} show that an element of the group algebra of $\mathbb Z_2^{*n}$ that is positive semidefinite in every unitary representation is a finite sum of hermitian squares (equivalently, $B^*B$ for a column-valued $B$), with a degree bound in terms of the element. This existence statement does not by itself supply a rational certificate. Exact rational certificates of this kind are standard (Peyrl and Parrilo~\cite{PeyrlParrilo2008}; Cafuta, Klep and Povh~\cite{CafutaKlepPovh2015}). \subsection{Explicit margins for \texorpdfstring{$\mathsf M$}{M} and \texorpdfstring{$\mathsf S$}{S}}\label{sec:cer:margins} For the two configurations $\mathsf M$ and $\mathsf S$ of \coreref{sec:cer}, \corename{thm:cer:seven} shows that no sequential projective representation reproduces from $e/3$ every word of length at most seven that begins with ``yes'' on $A$, by exhibiting one vector of negative squared norm. The same derivation, carried out in an arbitrary representation with every relation defect kept, gives an inequality linear in the word probabilities with explicit constants. \provisionalref{Robust constants}{thm:cer:sevenrobust}{Different form}{\leandecl{BlockB.thm_cer_seven_b_M_algebraic}, \leandecl{BlockB.thm_cer_seven_b_S_algebraic} and their Hilbert and quaternionic versions. Lean builds its own functional from encoded derivation data (\leandecl{BlockB.robustFunctionalM}, \leandecl{BlockB.robustFunctionalS}); that it is the functional of Appendix~\ref{app:cert:sevenrobust} is not a Lean theorem. Lean also proves $\|\lambda\|_1\le4.470\times10^9$ ($\mathsf M$) and $\le1.9519\times10^{10}$ ($\mathsf S$); the totals $T$, $W_\pi$: \leandecl{LengthSeven.Robust.counterT_M}, \leandecl{LengthSeven.Robust.counterT_S}.} \begin{theorem}[An explicit inequality of length seven]\label{thm:cer:sevenrobust} Let $p_A,p_B,p_C,p_D$ be the atoms of $\mathsf M$ or of $\mathsf S$. There are a finite set $\mathcal W^\ast$ of words of length at most seven, each beginning with ``yes'' on $A$, and rationals $\lambda_w$, $w\in\mathcal W^\ast$, such that $\sum_w\lambda_w\varphi(E_w)\ge0$ for every sequential projective representation, over $\R$, $\C$ or $\Ha$, of any dimension, with projections of any rank and any state, while $\sum_w\lambda_w\Pr_{e/3}(w)=-\gamma$ with \[ \gamma_{\mathsf M}=\frac{4{,}588{,}961}{169{,}869{,}312}\approx2.70\times10^{-2},\qquad \gamma_{\mathsf S}=\frac{412{,}734{,}271}{18{,}345{,}885{,}696}\approx2.25\times10^{-2}, \] and $\sum_w|\lambda_w|\le5.268\times10^{9}$ for $\mathsf M$, $\le2.892\times10^{10}$ for $\mathsf S$. \end{theorem} The proof is in Appendix~\ref{app:cert:sevenrobust}: it follows the derivation of the explicit vector of \coreref{thm:cer:seven} in an arbitrary representation, keeps each relation defect, and bounds it by the arithmetic--geometric mean inequality against one word probability. \Coreref{prop:robust} converts such an inequality into a margin: if $\Lambda=\sum_w\lambda_w[w]$ is supported on words of length at most $L$, is nonnegative on the table of every sequential projective representation and is at most $-\gamma<0$ on a target table $f$, then every sequential projective representation misses $f$ by at least $\gamma/\|\lambda\|_1$ on some word of the support, where $\|\lambda\|_1=\sum_w|\lambda_w|$. A reduced margin survives instruments that are close to projective. \provisionalref{Robust versions}{prop:robust:sqrt}{Same statement}{\leandecl{real_square_root_instrument_witness_bound}, \leandecl{complex_square_root_instrument_witness_bound}, \leandecl{quaternionic_square_root_instrument_witness_bound}.} \begin{proposition}[Generalised L\"uders instruments of nearly projective effects]\label{prop:robust:sqrt} Let $\Lambda$, $L$, $\gamma$ and $f$ be as in the preceding paragraph. Suppose a model uses for each test a fixed effect $0\le E\le1$ with $\|E^2-E\|\le \varepsilon<\frac14$ and the instrument with Kraus operators $\sqrt E$, $\sqrt{1-E}$ (the generalised L\"uders instrument of $E$), on a Hilbert space of any dimension with any state. Put $\tau=\frac12(1-\sqrt{1-4\varepsilon})$. If its probabilities are within $\delta$ of $f$ on the support, then $\delta+2L\sqrt\tau\ge\gamma/\|\lambda\|_1$. \end{proposition} \begin{proof} The spectrum of $E$ lies in $[0,\tau]\cup[1-\tau,1]$; let $P$ be its spectral projection onto $[\frac12,1]$. Then $\|\sqrt E-P\|\le\sqrt\tau$ and $\|\sqrt{1-E}-(1-P)\|\le\sqrt\tau$, and all these operators are contractions. For a word of length at most $L$ the two Kraus products $A$, $B$ differ in norm by at most $L\sqrt\tau$ (telescoping), so $\|A^*A-B^*B\|\le2L\sqrt\tau$ and the probabilities differ by at most $2L\sqrt\tau$ for every state. The projections $P$ form a sequential projective representation, to which \coreref{prop:robust} applies. \end{proof} For the inequalities of Theorem~\ref{thm:cer:sevenrobust}, \coreref{prop:robust} gives \[ \frac{\gamma}{\|\lambda\|_1}\ge \begin{cases} \dfrac{0.0270}{5.268\times10^{9}}>5.1\times10^{-12} & (\mathsf M),\\[2ex] \dfrac{0.0224}{2.892\times10^{10}}>7.7\times10^{-13} & (\mathsf S). \end{cases} \] These are the margins that the bookkeeping of Appendix~\ref{app:cert:sevenrobust} guarantees; write $m=\gamma/\|\lambda\|_1$. With $L=7$, Proposition~\ref{prop:robust:sqrt} excludes an exact match by such a model whenever $\varepsilon5.1\times10^{-12}/14>3.6\times10^{-13}$, so $s>1.29\times10^{-25}$, $s(1-s)>1.2\times10^{-25}$, and every $\varepsilon\le1.2\times10^{-25}$ is excluded. For $\mathsf S$, $\gamma/14\|\lambda\|_1>7.7\times10^{-13}/14=5.5\times10^{-14}$, so $s>3.02\times10^{-27}$, $s(1-s)>3.0\times10^{-27}$, and every $\varepsilon\le3.0\times10^{-27}$ is excluded. The two tolerances share one budget. Suppose the probabilities of the words of $\mathcal W^\ast$ are known to within $\delta$ each of the true table, which is either the octonionic table or that of a model with generalised L\"uders instruments as in Proposition~\ref{prop:robust:sqrt}. By that proposition the two tables differ by at least $m-2L\sqrt\tau$ on some word of $\mathcal W^\ast$, so the data decide between them whenever $2\delta+2L\sqrt\tau7.28\times10^{-13}$) and for $\beta\le1.1\times10^{-13}$ ($\mathsf S$, since $7.7\times10^{-13}/7=1.1\times10^{-13}$ and the margin is strictly larger). For $\mathsf S$ the construction behind Theorem~\ref{thm:eleven}(c) also gives an inequality from every state with a closed-form constant, and for a class of models wider than class~(I). An \emph{algebraic model} of four tests consists of a unital associative real $*$-algebra $\mathcal A$, self-adjoint idempotents $P_A,P_B,P_C,P_D\in\mathcal A$ and a real-linear functional $\varphi$ on $\mathcal A$ with $\varphi(1)=1$, $\varphi(X^*)=\varphi(X)$ and $\varphi(X^*X)\ge0$; its probability of a word $w$ is $\varphi(E_w)$, with $E_w$ formed from the $P_i$ as in \coreref{sec:set}. A sequential projective representation over $\R$, $\C$ or $\Ha$ gives one, with $\mathcal A$ the real $*$-algebra of bounded operators and $\varphi$ the real part of the state. \provisionalref{Length eleven for S}{prop:cer:Seleven}{Different form}{The functional: \leandecl{fixedSConstructed}\allowbreak\leandecl{Witness}; its length bound: \leandecl{fixedSConstructed}\allowbreak\leandecl{Witness_horizon}; (a) \leandecl{fixedSConstructed}\allowbreak\leandecl{Witness_model_nonnegative}; (b) \leandecl{fixedSConstructed}\allowbreak\leandecl{Witness_native_le}, with $y\ge\frac{14}{69}e$ in \leandecl{fixedSAtomSum_lower}; (c) \leandecl{fixedSConstructed}\allowbreak\leandecl{Witness_forces_error}. Lean builds the functional by the recursion of Appendix~\ref{app:cert:quantitative} from its own encoding of the occurrences; that these are the occurrences of the appendix, and that its words lie in $\mathcal F$, are not Lean theorems.} \begin{proposition}[An inequality of length eleven for $\mathsf S$ from every state]\label{prop:cer:Seleven} Let $p_A,p_B,p_C,p_D$ be the atoms of $\mathsf S$, so that $\Dtet=\frac1{12}$, and let \[ \Lambda_{\mathsf S}(f)=a\sum_{j,\alpha}R_{j\alpha}(f)-\frac1{12}\sum_jf(j)+\frac{7}{828}\,f(\varnothing), \qquad a=\frac{207\,(K_0+1)}{7}, \] be the functional $\Lambda$ of Appendix~\ref{app:cert:quantitative} with $\gamma=\frac7{828}$, where $R_{j\alpha}$ and $K_0$ are as defined there. Its coefficients are real and its words have length at most eleven. \begin{enumerate}[label=(\alph*),leftmargin=2em] \item $\Lambda_{\mathsf S}(w\mapsto\varphi(E_w))\ge0$ for every algebraic model. \item $\Lambda_{\mathsf S}(\Pr_\rho)\le-\frac7{828}$ for every state $\rho$ of $\J$. \item If an algebraic model and a state $\rho$ of $\J$ satisfy $|\varphi(E_w)-\Pr_\rho(w)|\le\varepsilon$ for every word $w$ of length at most eleven, then $\frac7{828}\le\|\Lambda_{\mathsf S}\|_1\,\varepsilon$, where $\|\Lambda_{\mathsf S}\|_1$ is the sum of the absolute values of the coefficients after merging coincident words. \end{enumerate} \end{proposition} \begin{proof} (a) On $\mathcal A$ the form $\ip XY=\varphi(X^*Y)$ is bilinear, symmetric (because $\varphi(X^*)=\varphi(X)$) and positive semidefinite. The identity \eqref{eq:cer:residualtrace} and \coreref{lem:purity} use only products of the projections, adjoints and a real inner product, so they hold in $\mathcal A$ with $\Omega=1$ and this form. A product $X$ of projections has $\ip XX\le1$, since $\varphi(Y^*PY)\le\varphi(Y^*Y)$ for every self-adjoint idempotent $P$ (the difference is $\varphi(((1-P)Y)^*(1-P)Y)$). The Cauchy--Schwarz and Young step of Appendix~\ref{app:cert:quantitative} then gives (a); it holds for every $\gamma>0$. (b) By \coreref{lem:purity} every $R_{j\alpha}$ vanishes on the octonionic table, so $\Lambda_{\mathsf S}(\Pr_\rho)=\frac7{828}-\frac1{12}\ip\rho y$ with $y=p_A+p_B+p_C+p_D$. Let $\lambda_1\le\lambda_2\le\lambda_3$ be the eigenvalues of $y$; they are nonnegative because $y$ is a sum of atoms. The traces of $\mathsf S$ listed in \coreref{sec:cer} give \[ \tr y=4,\qquad \tr y^2=4+2\sum_{i0$, while violating a purity relation. The following construction, called here the \emph{left-multiplication model}, explains where it comes from and works for every configuration with $\Acal>0$. \begin{proposition}[The left-multiplication model]\label{prop:cer:leftmult} Let $p_A,\dots,p_D$ be atoms with $\Acal>0$, in the normal form of {\coreref[(iv)]{thm:fourpoint}}, and write $y_3\,ij+y_4\ell=r\,e_3'$ with $r>0$ and the unit $e_3'=\cos\alpha\,ij+\sin\alpha\,\ell$. On $\R^{24}=\R^3\otimes\Oct$ let $P_v=\mathcal L_v\mathcal L_v^{\mathsf T}/|v|^2$, where $\mathcal L_vh=(v_1h,v_2h,v_3h)$ is left multiplication by the entries of $v$ (rank eight). Then: \begin{enumerate}[label=(\alph*),leftmargin=2em] \item $\R^{24}$ splits into two invariant subspaces of dimension twelve, on which the $P_v$ act as the real form (by left multiplication on $\Ha\cong\R^4$) of the standard representations of two quaternionic configurations $v^\pm$, obtained by substituting $e_3'\mapsto\pm\,ij$. \item With the state that gives weight $\frac12(1+\cos\alpha)$ to $v^+$ and $\frac12(1-\cos\alpha)$ to $v^-$ (maximally mixed within each block), the model reproduces every octonionic word of length at most two from $e/3$ and all thirteen traces. \item Its first deviation is at length three, where \[ \Pr\nolimits_{\rm model}(C,D,C)-\Pr\nolimits_{e/3}(C,D,C) =\frac{4}{3}\,\frac{\Acal}{o_{AB}(1-o_{AB})^2\,o_{AC}\,o_{AD}} , \] and this difference is the purity statistic of $z_C((D))$, which vanishes in $\J$. \end{enumerate} \end{proposition} The proof below gives the decisive steps; the expansion of a general word probability as a polynomial in $\cos\alpha$ is only described. \begin{proof}[Proof sketch] The units $i,j,e_3'$ are orthonormal and imaginary, so the left multiplications $L_i,L_j,L_{e_3'}$ on $\Oct$ anticommute and square to $-1$ (alternativity). Hence $\mathcal V=L_iL_jL_{e_3'}$ is symmetric, $\mathcal V^2=1$, and $\mathcal V$ commutes with all three; its trace is $0$. Each eigenspace $\Oct_\pm$ of $\mathcal V$ is a four-dimensional module of the Clifford algebra generated by $L_i,L_j,L_{e_3'}$, hence isomorphic to $\Ha$ with these operators acting by left multiplication by $i,j,\mp ij$, since $L_iL_j$ acts there as left multiplication by $ij$ and $(ij)^2=-1$. Every entry of every $v$ lies in $\R\oplus\spn\{i,j,e_3'\}$, so $P_v$ preserves $\R^3\otimes\Oct_\pm$ and acts there as the real form of the quaternionic projection onto $v^\mp\Ha$. Each block reproduces the Jordan table of its quaternionic configuration (the standard embedding preserves $U_ab=aba$), which gives (a) and the mixture formula. Fix all normal-form coordinates except $\alpha$. Every octonionic word probability $f$ is a polynomial in $\cos\alpha$; it is even in $\alpha$ by the automorphism fixing $\Ha$ and negating $\Ha\ell$. The model gives the chord $\frac12(1+c)f(1)+\frac12(1-c)f(-1)$, $c=\cos\alpha$. A probability that is linear in $p_D$ (no $U_{p_D}$ or $U_{e-p_D}$ except as the first or last update) is affine in $c$, because the $\sin\alpha$ terms pair $\ell$ with elements of $\Ha$ and have zero real part; affine functions agree with their chords. This covers all words of length at most two and the words that define the thirteen traces, which is (b). For $(C,D,C)$ one has $\Pr=\frac13o_{CD}^2$ with $o_{CD}=o_0+o_1c$, and the chord exceeds the value by $\frac13(1-c^2)o_1^2$; substituting the normal form gives the displayed formula. Since $\Pr(C,D)$ and $\Pr(C)$ are matched, the deviation equals $\Pr(C,D,C)-2o_{CD}\Pr(C,D)+o_{CD}^2\Pr(C)$, the purity statistic of $z_C((D))$. \end{proof} At the maximal configuration $\cos\alpha=0$ (the normal form has $Y=\ell$), the weights are equal and the model is tracial. For $\mathsf S$ the same holds, and the complex form of the construction, with the two irreducible complex modules of the Clifford algebra in place of $\Ha$, is the model of \coreref{prop:Dfails}; there the deviation is again $\frac4{243}$. The anticommuting matrices $C_1,C_2,C_3$ of that model are $L_i,L_j,L_{e_3'}$ in that complex module, and $\mathcal V$ becomes $1\otimes\sigma_x$. \Coreref{rem:cer:variance} explains why such a mixture is forced. The model is not optimal among those matching low-order statistics: the model of \coreref{prop:horizon} matches all words of length at most six for $\mathsf M$ and $\mathsf S$, and \coreref{prop:cer:universalsix} gives such a model for every generating quadruple. Representing octonions by left multiplications is classical, as in the left-multiplication graphs of Freedman, Shokrian-Zini and Wang for the octonionic line~\cite[Sec.~4.2]{FSW2018}. Each twelve-dimensional block of Proposition~\ref{prop:cer:leftmult}(a) realises quaternionic lines of $\Ha^3=\R^{12}$ as $4$-planes, the smallest case of Wolf's realisations of the quaternionic projective plane by isoclinic $n$-planes in $\R^{3n}$, $n$ divisible by four~\cite[Thm.~2]{Wolf1963elliptic}. \section{Linear equalities}\label{sec:eq} The setting is that of \coreref{sec:cer:equality}: $c$ is a projective equality if $\sum_wc_w\varphi(E_w)=0$ in every sequential projective representation; $FJ_n$ is the free unital Jordan algebra on idempotents $p_1,\dots,p_n$, $\varpi:FJ_n\to\R[\mathbb Z_2^{*n}]^+$ the homomorphism with $p_i\mapsto\frac12(1+r_i)$, $V_n\subseteq FJ_n$ the span of the elements $U_{q_1}\cdots U_{q_k}1$ with $q_t\in\{p_i,1-p_i\}$, and $f(c)=\sum_wc_wU_{q_1}\cdots U_{q_k}1$, so that $\varpi(f(c))=E(c)$. By \coreref{prop:cer:noequality}, $c$ is a projective equality exactly when $E(c)=0$, and its value on the octonionic table from $\rho$ is $\ip\rho{f(c)(p)}$. \subsection{No equality through five steps}\label{sec:eq:five} \provisionalref{Through five steps}{prop:eq:five}{Same statement}{\leandecl{HorizonFive.horizonFive_projectiveEquality}, assuming only that $c$ vanishes on real finite-dimensional representations with vector states; the form with $E(c)=0$: \leandecl{HorizonFive.horizonFiveClause}.} \begin{proposition}[No equality through five steps]\label{prop:eq:five} For any four atoms of $\J$, in particular those of $\mathsf S$, no projective equality supported on words of length at most five is violated from any state of $\J$. \end{proposition} This is \coreref{rem:cer:fivestep}, where only the idea of the proof is given. The proof is in Appendix~\ref{app:cert:h5}. \subsection{The second-order neighbourhood of a complex subalgebra}\label{sec:eq:second} Near a complex subalgebra every projective equality is weak: its residual vanishes to second order, and states in the subalgebra see it only at third order. The reason is that the second-order thickening of a complex subalgebra has a special quotient that retains everything except an $18$-dimensional piece. Let $\C_i^\perp\subset\Oct$ be the orthogonal complement of $\C_i=\R+\R i$ (dimension six), $W=H_3(\C_i)$, and $W^\perp\subset\J$ its trace-orthogonal complement: the matrices with zero diagonal and off-diagonal entries in $\C_i^\perp$ (dimension $18$). For $m\ge0$ put $D_m=\C_i\oplus t\Oct\oplus\dots\oplus t^m\Oct\subset\Oct[t]/(t^{m+1})$, with $t$ central and conjugation acting on the octonion coefficients, and $E_m=H_3(D_m)=W\oplus tJ\oplus\dots\oplus t^mJ$, a Jordan subalgebra of $\J[t]/(t^{m+1})$ with the matrix product $x\jp y=\frac12(xy+yx)$. \needspace{6\baselineskip} \provisionalref{Second-order kernel}{prop:so:kernel}{Lean scope narrower}{(a) \leandecl{AppF.red1_jordan}, \leandecl{AppF.red1_injective}; (b) \leandecl{AppF.E2_jordanHom_kills_K}, without Jacobson's theorem; (c) \leandecl{AppF.redq_jordan}, \leandecl{AppF.redq_unit}, \leandecl{AppF.redq_eq_zero_iff}, \leandecl{AppF.Dq.instRing}; non-speciality of $E_2$: \leandecl{AppF.E2_not_special}. In (a) and (c), $D_1$ and $D_2/I$ are explicit associative algebras and $E_1$ is identified with its image under the injective entrywise map. The dimensions $18$ and $45$ are not formalized.} \begin{proposition}[Second-order neighbourhood]\label{prop:so:kernel} \leavevmode \begin{enumerate}[label=(\alph*),leftmargin=2em] \item $E_1$ is special: it is a Jordan subalgebra of $M_3(D_1)^+$, and $D_1$ is associative. \item Every Jordan homomorphism $f:E_2\to A^+$, $A$ an associative real algebra, unital or not and $f$ unital or not, vanishes on $K=t^2W^\perp$. \item Entrywise reduction modulo $I=t^2\C_i^\perp$ is a Jordan homomorphism of $E_2$ into $M_3(D_2/I)^+$ with kernel exactly $K$, and $D_2/I$ is associative. \end{enumerate} Hence the intersection of the kernels of all Jordan homomorphisms from $E_2$ into associative algebras is $K$, of dimension $18$; it is attained by one of them; $E_2/K$ is special of dimension $45$; and $E_2$ is not special. \end{proposition} \begin{proof} \emph{Associators of $D_2$.} For $u,v\in\C_i^\perp$ the linearised alternative laws $[i,u,v]=-[u,i,v]=[u,v,i]$ and $ui=-iu$, $vi=-iv$ give $(iu)v=(uv)i$, so $[i,u,v]=-(i(uv)-(uv)i)\in\C_i^\perp$. The associator is trilinear and respects the degree in $t$; it vanishes when two arguments lie in $\C_i$ (alternativity) or the degrees add up to three or more. So the associators of $D_2$ are spanned by $t^2[i,x,y]=t^2[i,x_\perp,y_\perp]$, and lie in $I$. Conversely, from the Cayley--Dickson rule, $[i,j,q\ell]=2(q\,ij)\ell$ for $q\in\{1,i,j,ij\}$, $[i,\ell,j\ell]=-2ij$ and $[i,\ell,(ij)\ell]=2j$ span $\C_i^\perp$. So $I$ is the span of the associators; it is a two-sided ideal stable under conjugation, and $D_2/I$ is associative. For (a), an associator of $D_1$ needs two arguments of positive degree, hence vanishes. (b) The proof uses Jacobson's extension theorem~\cite[Thm.~7.3, p.~25]{Jacobson1954}: \emph{if $\mathfrak D$ is a unital algebra with involution over a field of characteristic $\ne2$, $n\ge3$, and $S$ is a homomorphism of the Hermitian $n\times n$ matrices over $\mathfrak D$ (canonical involution, product $xy+yx$) into $\mathfrak B_J$, an algebra $\mathfrak B$ with the product $xy+yx$, such that the images of the $E_{kk}$ and of the $E_{kl}+E_{lk}$ lie in the nucleus of the subalgebra generated by the image, then $S$ extends uniquely to a homomorphism $M_n(\mathfrak D)\to\mathfrak B$.} No associativity, simplicity or unitality of $S$ is assumed. Take $\mathfrak D=D_2$, $n=3$ and the conjugate transpose; $f$ preserves $x\jp y$ if and only if it preserves $2x\jp y$, and in the associative $A$ the nucleus condition is empty. So $f$ extends to an algebra homomorphism $\tilde f:M_3(D_2)\to A$, which kills associators. For $a,b,c\in D_2$ and $k\ne l$, $[aE_{kk},bE_{kk},cE_{kl}]=[a,b,c]E_{kl}$; hence $\tilde f(dE_{kl})=0$ for $d\in I$, and every element of $K$ is a sum of such terms. (c) Reduction modulo the conjugation-stable ideal $I$ commutes with the conjugate transpose and with matrix products, so it maps $E_2$ into the Hermitian part of $M_3(D_2/I)$, a Jordan subalgebra of $M_3(D_2/I)^+$. Its kernel is $E_2\cap M_3(I)=K$, since diagonal entries are real polynomials. The dimensions are $9+27+27=63$ for $E_2$ and $3\cdot6=18$ for $K$. An injective $f$ would contradict (b). \end{proof} \provisionalref{Residual near W}{cor:so:residual}{Same statement}{\leandecl{AppF.cor_so_residual}, \leandecl{AppF.cor_so_residual_of_atoms}, for idempotent curves of class $C^3$; a version stated with linear functionals only, \leandecl{AppF.cor_so_residual_linearFunctional}.} \begin{corollary}[Universal equalities near a complex subalgebra]\label{cor:so:residual} Let $c$ be a projective equality in $n$ tests ({\coreref[(a)]{prop:cer:noequality}}), and let $p_1(t),\dots,p_n(t)$ be atoms of $\J$ depending three times continuously differentiably (for instance, analytically) on $t\in\R$, with $p_k(0)\in W$ for all $k$. Then the residual $r(t)=f(c)(p_1(t),\dots,p_n(t))\in\J$ satisfies $r(t)=t^2Q+O(t^3)$ with $Q\in W^\perp$. Hence for every fixed state $\rho\in W$, $\sum_wc_w\Pr_\rho(w)=\ip\rho{r(t)}=O(t^3)$. \end{corollary} \begin{proof} Write $p_k(t)=p_k+tv_k+t^2a_k+O(t^3)$. Comparing coefficients in $p_k(t)^2=p_k(t)$, the truncations $\tilde p_k=p_k+tv_k+t^2a_k$ are idempotents of $E_2$, and $f(c)(\tilde p)=r(0)+tr'(0)+\frac12t^2r''(0)$ because $f(c)$ is a Jordan polynomial. The map of Proposition~\ref{prop:so:kernel}(c) is unital, so it sends the $\tilde p_k$ to idempotents $P_k$ of $M_3(D_2/I)^+$. Evaluation at them factors through $\varpi$ and the homomorphism $\R[\mathbb Z_2^{*n}]\to M_3(D_2/I)$, $r_k\mapsto2P_k-1$, so it kills $f(c)$, whose image under $\varpi$ is $E(c)=0$. Thus $f(c)(\tilde p)\in K=t^2W^\perp$. \end{proof} Only part (c) is used here, not Jacobson's theorem; with $C^2$ dependence the remainder is $o(t^2)$. Part (b) says that no special quotient of $E_2$ retains more of the second-order coefficient. The same argument, with Jacobson's Theorem~7.3~\cite{Jacobson1954} again for the intersection of the kernels, applies to the real and quaternionic subalgebras $H_3(\R)$ and $H_3(\Ha_{ij})$: with $D_m(B)=B\oplus t\Oct\oplus\dots\oplus t^m\Oct$, associators need three, two or one arguments of positive degree for $B=\R,\C_i,\Ha_{ij}$, so $E_m(B)=H_3(D_m(B))$ is special for $m<3,2,1$, and at $m=3,2,1$ the intersection of the kernels is $t^mH_3(B)^\perp$, of dimension $21$, $18$, $12$ (for $B=\R$ the associators also contain $[j,\ell,(ij)\ell]=-2i$; for $B=\Ha_{ij}$ they lie in $\Ha_{ij}\ell$, spanned by $[i,j,q\ell]$). \section{Five tests}\label{sec:five} \Coreref{thm:pentad} gives five atoms \[ \mathsf F=(p_0,p_1,p_2,p_3,p_4)=\bigl(E_{11},\ p(1,1,0),\ p(1,i,1),\ p(1,j,0),\ p(1,\ell,0)\bigr) \] of $\J$ such that every four of their tests have a sequential projective representation on $\C^6$ while the five together have none, and \coreref{cor:helly} shows that five tests always suffice to decide. Here $p(v)=vv^\dagger/\|v\|^2$ for a column $v$ whose entries lie in one associative subalgebra of $\Oct$, as in \coreref{sec:fiv:coord}. This section shows that words of length eleven already exclude the five tests, and that this length is the least possible; it then adds a general reflection-depth bound, the Helly number of the quaternionic subalgebras and a comparison with other local-to-global statements. \subsection{Atoms on one line}\label{sec:five:line} The atoms orthogonal to $E_{33}$ form the line $e-E_{33}$ of $\OP$ (the projective line of all atoms $p\le e-E_{33}$). Each has the form $(f+u)/2$, where $f=e-E_{33}$ and $u$ is a unit vector, its Bloch vector, in the nine-dimensional space of traceless elements of the Peirce space $J_0(E_{33})\cong H_2(\Oct)$. Two such atoms have transition probability $(1+u\cdot u')/2$. The algebra generated by $e$ and at least two distinct atoms on this line is $\R E_{33}\oplus\mathrm{JSpin}_r$, where $r$ is the dimension of the span of their Bloch vectors and $\mathrm{JSpin}_r=\R f\oplus\R^r$ is the spin factor. Indeed $u\jp v=(u\cdot v)f$ for Bloch vectors, so $\R f\oplus\spn\{u\}$ is closed, and it is generated by the atoms and $f=e-E_{33}$. For two distinct atoms $p,q$ with Bloch vectors $u,u'$, the square $(p-q)^2=\frac14|u-u'|^2f$ is a positive multiple of $f$, so $f$ lies in the generated algebra. A single atom generates with $e$ only the two-dimensional algebra $\spn\{e,p\}$. For example, \[ E_{11},\quad p(1,1,0),\quad p(1,j,0),\quad p(1,\ell,0) \] have Bloch vectors $(1,0)$, $(0,1)$, $(0,j)$, $(0,\ell)$ in $\R\oplus\Oct$. They are mutually orthogonal, so all six transition probabilities equal $\frac12$, and the four atoms generate $\R E_{33}\oplus\mathrm{JSpin}_4$, of dimension six. This is the four-element subfamily $\mathsf F\setminus\{p_2\}$ of the core paper. Atoms on one line are always special together, since spin factors of any dimension have finite-dimensional Clifford representations. To reach $\J$ one must leave the line. \subsection{Reflection transfer}\label{sec:fiv:list} {\Coreref[(b)]{thm:pentad}} concerns words of every length. The explicit certificates of Section~\ref{sec:cer} are for four atoms with $\Acal>0$, and every four-element subfamily of $\mathsf F$ has $\Acal=0$. The way around is to manufacture, from the five tests, four atoms with $\Acal>0$ whose word probabilities are fixed combinations of word probabilities of the five tests. Reflections do this: by \coreref{eq:bri:refl}, conjugating a sequential product by $2P-1$ gives a combination of three sequential products, two of them with one more test of $\mathsf F$. \Coreref{sec:bri:recon} used this to lift the whole reflection orbit of a family; here the argument uses it once and counts the cost. For a word $w$ in a family of tests some of which are marked \emph{reflected}, a \emph{run} is a maximal block of consecutive letters of $w$ that are outcomes of reflected tests. Put \[ \nu(w)=2\cdot\#\{\text{runs of }w\}-\begin{cases}1&\text{if the last letter of $w$ is reflected},\\0&\text{otherwise.}\end{cases} \] \provisionalref{Reflection transfer}{lem:five:reflect}{Lean scope narrower}{\leandecl{FiveTest.quad_lift}. Only for the reflected quadruple of Theorem~\ref{thm:five:refl}, with the transfer stated as membership in a span of words of bounded length; (a), (b) and (d) are not stated in general.} \begin{lemma}[Reflection transfer]\label{lem:five:reflect} Let $V$ be a Euclidean Jordan algebra, $p$ an idempotent of $V$ and $r=2p-e$. \begin{enumerate}[label=(\alph*),leftmargin=2em] \item $U_r=2U_p+2U_{e-p}-\id$ is a Jordan automorphism with $U_r^2=\id$ and $U_re=e$; it maps idempotents to idempotents and atoms to atoms, and for every idempotent $q$ \[ U_{U_rq}=U_rU_qU_r,\qquad e-U_rq=U_r(e-q). \] \item Let $P$ be an orthogonal projection on a real, complex or quaternionic Hilbert space and $\Sigma=2P-1$. Then $\Sigma X\Sigma=2PXP+2(1-P)X(1-P)-X$ for every bounded $X$; for every projection $Q$, $\Sigma Q\Sigma$ is a projection and $1-\Sigma Q\Sigma=\Sigma(1-Q)\Sigma$. \item Let $p_1,\dots,p_n$ be idempotents of $V$ (the tests), $p=p_c$ one of them, $F\subseteq \{1,\dots,n\}$, and $p'_s=U_rp_s$ for $s\in F$ (the reflected tests). Let $w$ be a word in the tests $p_1,\dots,p_n$ and $p'_s$, $s\in F$. Then there are $N\le3^{\nu(w)}$ words $w_1,\dots,w_N$ in $p_1,\dots,p_n$ alone and integers $d_k$, depending only on $w$ and $c$, such that \begin{equation}\label{eq:fiv:transfer} \theta(w)=\sum_{k=1}^Nd_k\,\theta(w_k),\qquad E_w=\sum_{k=1}^Nd_k\,E_{w_k} \end{equation} for all projections $P_1,\dots,P_n$ on a real, complex or quaternionic Hilbert space, where on the left $p'_s$ is represented by $\Sigma P_s\Sigma$ with $\Sigma=2P_c-1$. Each $w_k$ arises from $w$ by replacing every outcome of $p'_s$ by the same outcome of $p_s$ and inserting, at each of $\nu(w)$ fixed places, nothing or one outcome of $p_c$; so $|w_k|\le|w|+\nu(w)$. \item Consequently, if $(\Hil,P_1,\dots,P_n,\varphi)$ is a sequential projective representation of the tests $p_1,\dots,p_n$ from a state $\rho$, then $\Hil$ with the projections $P_i$ and $\Sigma P_s\Sigma$ ($s\in F$) and the same $\varphi$ is one of the enlarged family from $\rho$. For a finite set $\mathcal L$ of words in the enlarged family, it suffices that the words $w_k$ arising from the $w\in\mathcal L$ are reproduced. \end{enumerate} \end{lemma} \begin{proof} \emph{(a)} The formula for $U_r$ is \coreref{eq:bri:refl}, and $U_r$ is a Jordan automorphism by \coreref{sec:set:eja}; $U_r^2=\id$ and $U_re=e$ are clear. An automorphism $\beta$ preserves squares, hence order, spectral decompositions and primitivity, and $U_{\beta x}=\beta U_x\beta^{-1}$ follows from $U_xy=2x\jp(x\jp y)-(x\jp x)\jp y$. Take $\beta=U_r=\beta^{-1}$. \emph{(b)} The formula is \coreref{eq:bri:refl}; expand $\Sigma=P-(1-P)$. Since $\Sigma$ is a self-adjoint unitary, $\Sigma Q\Sigma$ is a projection and $\Sigma(1-Q)\Sigma=1-\Sigma Q\Sigma$. \emph{(c)} The proof is in Appendix~\ref{app:sub}. \emph{(d)} By (b) the $\Sigma P_s\Sigma$ are projections, and by \eqref{eq:fiv:transfer} $\varphi(E_w)=\sum_kd_k\varphi(E_{w_k})=\sum_kd_k\ip\rho{\theta(w_k)}=\ip\rho{\theta(w)}$. \end{proof} For $\mathsf F$ take $p_2$ as the reflecting test, $r=2p_2-e$, and reflect $p_3$ and $p_4$. The vectors $(1,i,1)$ and $(1,j,0)$ lie in $\Ha_{ij}^3$, where $U_rx=rxr$, and a short calculation (using $i(ij)=-j$) gives \[ U_rp_3=p(-1-2ij,\ 2i-j,\ 2-2ij),\qquad U_rp_4=p(-1-2i\ell,\ 2i-\ell,\ 2-2i\ell). \] Write $(A,B,C,D)=(p_0,p_1,U_rp_3,U_rp_4)$; the tests $C$ and $D$ are the reflected ones, and $\nu$ is counted with respect to them. Since $U_r$ is an automorphism, $o_{CD}=\ip{p_3}{p_4}=\frac12$. Appendix~\ref{app:cert:refl} computes the other traces by hand from the entries of $U_rp_3$ and $U_rp_4$: \[ \begin{gathered} o_{AB}=o_{CD}=\tfrac12,\qquad o_{AC}=o_{AD}=o_{BC}=o_{BD}=\tfrac5{18},\\ R_{ABC}=R_{ABD}=R_{ACD}=R_{BCD}=\tfrac5{36}, \end{gathered} \] \[ (t_1,t_2,t_3)=\bigl(\tfrac5{72},\tfrac{43}{648},\tfrac{43}{648}\bigr),\qquad (\omega_1,\omega_2,\omega_3)=\bigl(\tfrac{41}{648},\tfrac5{72},\tfrac5{72}\bigr), \] for the double-pair traces $t_r$ and the combinations $\omega_r=t_1+t_2+t_3-2t_r$ (\coreref{def:traces}). Substituting in \coreref{eq:A:Acal52} gives $\Acal=\frac1{11664}>0$ (the eight groups of terms are listed there), so by \coreref{thm:fourpoint} the quadruple generates $\J$. The uniform argument of \coreref{sec:cer:horizon} applies to it, with one change forced by the cost of reflections, and Lemma~\ref{lem:five:reflect} carries the result to the five tests. \subsection{The reflected quadruple and its budget}\label{sec:five:budget} \paragraph{\Coreref[(a)]{lem:cer:gram7} by a cheaper route.} The words $(v,\check v,A)$ used by (R3$_0$) can contain three runs of reflected letters; for middle letter $B$ the words $(D,B,C,B,D,A)$ and $(C,B,D,B,C,A)$ have $\nu=6$ and $|u|+\nu(u)=12$ (for middle letters $C$ and $D$ a suitable order stays within the budget, but one route for all three is simpler). The vector identity of {\coreref[(a)]{lem:cer:gram7}} is therefore proved by a pair identity. For an ordering $(x,y,z)$ of $B,C,D$, signing the four purity instances at $A$ on $\xi$ with the words $(\pm x,\pm y,z)$ by $(-1)^{\text{number of ``yes'' among the outcomes on $x$ and $y$}}$ gives \[ \text{(P3)}\qquad P_AS_{xyz}P_A\xi=c_{xyz}\,\xi, \] where \[ c_{xyz}=\ip{p_z}{(\id-2L_{p_y})(\id-2L_{p_x})p_A}=o_{Az}-2R_{Axz}-2R_{Ayz}+4\ip{p_A\jp p_x}{p_y\jp p_z} \] with $S_{xyz}=P_z-(P_xP_z+P_zP_x)-(P_yP_z+P_zP_y)+(P_xP_yP_z+P_zP_yP_x)+(P_xP_zP_y+P_yP_zP_x)$; here $U_{e-p}-U_p=\id-2L_p$ (from \coreref{eq:fp:access}) is used twice. With (R1$_0$) and (R2$_0$) at $A$, (P3) reads $\Pi_y+\Pi_z=4\ip{p_A\jp p_x}{p_y\jp p_z}\,\xi$, where $\Pi_u=\eta^C_s+\eta^C_{s^{\mathrm{rev}}}$ for the orderings $s,s^{\mathrm{rev}}$ with middle letter $u$. The three such equations give $\Pi_u=2\omega_u\xi$, which is {\coreref[(a)]{lem:cer:gram7}}; the pair $\Pi_y+\Pi_z$ is also obtained from the ordering $(x,z,y)$, and the choice between the two is made below. \paragraph{The budget.} \Coreref{lem:cer:gram28} applies with $\mathcal S$ the set of words $u$ from $p_0$ with $|u|+\nu(u)\le10$: \begin{enumerate}[label=(\arabic*),leftmargin=2em] \item Hypothesis (i): a word of length at most five is within the budget, since $\nu$ is twice the number of runs, less one if the last letter is reflected, and three runs in five letters force reflected letters in places $1,3,5$, so $\nu\le5$; four letters give $\nu\le4$. \item Hypothesis (ii): in an instance of (R2$_t$) with $|t|=1$ the longest words are $(t,q,\bar v,w,\bar v,q)$ and $(t,q,v,w,v,q)$, of length six. Choose as $v$ an inner letter reflected if and only if $q$ is, whenever one exists. If $q$ is reflected, the reflected letters lie in $t$, $q$, the block $(\bar v,w,\bar v)$ adjacent to $q$, and the final $q$, which form at most two runs; the word ends reflected and $\nu\le3$. This includes the case where both inner letters are unreflected: then the reflected letters lie in $t\cup\{q\}$ and in the final $q$, at most two runs, $\nu\le3$, whichever letter is repeated. If $q$ is not reflected and $v$ is chosen unreflected, or both inner letters are reflected, the runs come only from $t$ and the block $(\bar v,w,\bar v)$, at most two in all, the word ends unreflected, and $\nu\le4$. The word $(t,q,v,w,v,q)$ has the same tests as $(t,q,\bar v,w,\bar v,q)$, hence the same $\nu$; the other words of the instance have length at most five. \item Hypothesis (iii): (P3) with the ordering $(x,y,z)$ chosen with $y$ reflected whenever $x$ is: $(C,D,B)$ for $x=C$, $(D,C,B)$ for $x=D$, either for $x=B$. Then $\nu\le4$ and the length is six. \end{enumerate} \provisionalref{Reflected quadruple}{thm:five:refl}{Same statement, different route}{\leandecl{FiveTest.thm_five_refl}, by the first argument of the proof, through \leandecl{LengthSeven.LemmaA.lemmaA_budget}; the explicit vector $\Phi$ is not formalized.} \begin{theorem}[The reflected quadruple within the budget]\label{thm:five:refl} No sequential projective representation of the four tests $(A,B,C,D)=(p_0,p_1,U_rp_3,U_rp_4)$ reproduces from $p_0$ the probability of every word $u$ with $|u|+\nu(u)\le10$. \end{theorem} \begin{proof} Suppose one does, and pass to its real cyclic representation; since $\Pr(A)=1$ from $p_0$, $\xi=\Omega$ satisfies $P_A\xi=\xi$. By the budget, \coreref{lem:cer:gram28} applies, and since $\Acal=\frac1{11664}>0$ the proof of {\coreref[(a)]{thm:cer:uniform}} gives a contradiction. For this quadruple a single explicit vector also suffices. The formulas of \coreref{lem:cer:gram7} hold under the same hypotheses; with $\eta^J_{uv}=P_vP_u\xi$ put \[ \begin{aligned} \Phi={}&2\eta^C_{BCD}-2\eta^C_{BDC}-3\eta^C_{CBD}+2\eta^C_{CDB}+3\eta^C_{DBC}-2\eta^C_{DCB}\\ &+4\eta^R_{BCD}-4\eta^R_{BDC}-6\eta^R_{CBD}+6\eta^R_{CDB}+6\eta^R_{DBC}-6\eta^R_{DCB}\\ &-4\eta^H_{BCD}+4\eta^H_{BDC}+7\eta^H_{CBD}-4\eta^H_{CDB}-7\eta^H_{DBC}+4\eta^H_{DCB}\\ &-\tfrac{10}9\eta^J_{CD}+\tfrac12\eta^J_{CB}+\tfrac{10}9\eta^J_{DC}-\tfrac12\eta^J_{DB}. \end{aligned} \] The automorphism of $\Oct$ that fixes $i$ and exchanges $j$ and $\ell$ fixes $p_0$, $p_1$, $p_2$ and exchanges $U_rp_3$ and $U_rp_4$; so the traces, and with them all values of the formulas, are unchanged by exchanging the labels $C$ and $D$, while $\Phi$ changes sign. By Appendix~\ref{app:cert:refl}, the part of $\|\Phi\|^2$ from the $\eta^C,\eta^R,\eta^H$ terms is $\frac{1873}{11664}$, the cross terms with the $\eta^J$ give $-\frac{335}{972}$, and the $\eta^J$ terms among themselves give $\frac{2081}{11664}$. Hence \[ \|\Phi\|^2=\frac{1873-4020+2081}{11664}=-\frac{66}{11664}=-\frac{11}{1944}<0. \qedhere \] \end{proof} \provisionalref{Five-test witness}{prop:five:short}{Same statement}{(a) \leandecl{FiveTest.five_short_p0}, \leandecl{FiveTest.five_short_e3}; (b) \leandecl{Ends.five_short_b_exists}; also a margin from $p_0$ at length ten, \leandecl{Ends.five_short_b_p0_exists}. The explicit value $9\times10^{-21}$ after the proof is not formalized.} \begin{proposition}[Words of length eleven exclude the five tests]\label{prop:five:short} Let $\mathsf F=(p_0,\dots,p_4)$ be the configuration of \coreref{thm:pentad}; the model class is~(I). \begin{enumerate}[label=(\alph*),leftmargin=2em] \item No sequential projective representation of the five tests reproduces from $p_0$ the probability of every word of length at most ten, and none reproduces from $e/3$ the probability of every word of length at most eleven that begins with ``yes'' on $p_0$. \item There is $\delta_{\mathsf F}>0$ such that every sequential projective representation of the five tests differs from the statistics of $\mathsf F$ from $e/3$ by more than $\delta_{\mathsf F}$ on some word of length at most eleven that begins with ``yes'' on $p_0$. \end{enumerate} \end{proposition} \begin{proof} (a) A representation reproducing the words from $e/3$ gives, after conditioning on ``yes'' on $p_0$ as in \coreref{sec:cer:horizon}, one reproducing from $p_0$ every word of length at most ten. By Lemma~\ref{lem:five:reflect}(d), with $p_2$ reflecting and $p_3,p_4$ reflected, the projections $P_0$, $P_1$, $\Sigma P_3\Sigma$, $\Sigma P_4\Sigma$ ($\Sigma=2P_2-1$) then reproduce every word $u$ of the reflected quadruple whose expansion words are reproduced, and these have length at most $|u|+\nu(u)$ by Lemma~\ref{lem:five:reflect}(c). So they reproduce every word with $|u|+\nu(u)\le10$, against Theorem~\ref{thm:five:refl}. (b) As in the proof of {\coreref[(b)]{thm:cer:uniform}}: the tables, on the finite set of words of length at most eleven beginning with ``yes'' on $p_0$, of sequential projective representations of the five tests form a compact set by \coreref{lem:cer:recon}, which by (a) does not contain the table of $\mathsf F$; so the largest deviation on these words has a positive minimum, and $\delta_{\mathsf F}$ may be taken as half of it. \end{proof} An explicit value follows from the construction of Theorem~\ref{thm:cer:sevenrobust}. Build $\Lambda''$ as in its proof (Appendix~\ref{app:cert:sevenrobust}, whose notation $T$, $y_k$, $X_k$, $[A,\alpha_k]$ is used here) for the vector $\Phi$ of Theorem~\ref{thm:five:refl} (coefficients of total absolute value $\frac{713}9$, $\|\Phi\|^2=-\frac{11}{1944}$), from $\zeta=P_A\Omega$, with one $\epsilon$ for all defects. Each step has a fixed defect weight ($1$ for (R1$_t$), $3$ for (R2$_t$) or a use of (W3), $\frac32$ for (R2$_t$) paired and halved, $\frac{33}2$ and $\frac{15}2$ for a vector and a scalar use of (P3)), and every multiplier is one of $o_{ab},2R_{abc},2\omega_u\in[0,1]$. Going through the derivations of \coreref{lem:cer:gram7}, which are printed there only in outline, every entry has table weight (its counter $T_e$) at most $\frac{17}2$ and defect weight at most $\frac{97}2$ (the largest, $\mathrm{RC}(zxy)$, is $\frac{33}2+\frac32+\frac{61}2$), and every word of $\Lambda''$ satisfies $|u|+\nu(u)\le10$. Then $T\le(\frac{713}9)^2\frac{17}2$ and $\sum_k|y_k|\le\bar Y=(\frac{713}9)^2\frac{97}2$. Every purity constant lies in $[0,1]$, so $\|X_k\|_1\le4$, and each $[A,\alpha_k]$ is at most $\frac13$ on the target. With $\epsilon=\frac{11}{1944}/\bar Y$ this gives $\gamma\ge\frac{11}{11664}$ and $\|\lambda''\|_1\le T+\frac{11}{3888}+\frac{2\cdot1944}{11}\bar Y^2<3.275\times10^{13}$. Replacing each word $u$ by its expansion of Lemma~\ref{lem:five:reflect}(c), whose coefficients have absolute sum at most $5^{\nu(u)}\le5^5$ (each place contributes $2,2,-1$), gives a combination of five-test words of length at most eleven beginning with $p_0$, nonnegative on every representation of the five tests (Lemma~\ref{lem:five:reflect}(d)), with the same octonionic value and $\ell^1$ norm at most $5^5\|\lambda''\|_1$. \Coreref{prop:robust} then gives $\delta_{\mathsf F}=\frac{11}{11664}/(5^5\cdot3.275\times10^{13})>9\times10^{-21}$. This value is not formalized, and it is crude: it bounds every entry by the worst one and every purity constant by one. \provisionalref{Five-test horizons}{cor:five:minimum}{Lean scope narrower}{\leandecl{MatchingModels.cor_five_minimum_p0_isLeast_classI}, \leandecl{MatchingModels.cor_five_minimum_e3_isLeast_classI}. The minimum is taken over Hilbert spaces in Lean's lowest universe; the exclusions hold in every universe.} \begin{corollary}[Minimum separating word length for the five tests]\label{cor:five:minimum} For the configuration $\mathsf F$ and model class~(I), let $L_{\min}(\rho)$ be the least $L$ such that no sequential projective representation reproduces all word probabilities from $\rho$ through length $L$. Then \[ L_{\min}(p_0)=10,\qquad L_{\min}(e/3)=11. \] \end{corollary} \begin{proof} The upper bounds are Proposition~\ref{prop:five:short}(a). For the lower bounds replace $p_4=p(1,\ell,0)$ by $p(1,\pm ij,0)$, leaving $p_0,p_1,p_2,p_3$ unchanged. These are two quaternionic configurations $\mathsf F_\pm$. Their standard complex representations on $\C^6$ give a direct-sum representation on $\C^{12}$, with one rank-four projection $P_k$ per test. Use the state $P_0/4$ for initial state $p_0$, and $I_{12}/12$ for initial state $e/3$. Its word probabilities are the average of those of $\mathsf F_+$ and $\mathsf F_-$ from the corresponding Jordan state. The Peirce--Clifford and reset-interval argument in Appendix~\ref{app:cert:five-minimum} proves equality with $\mathsf F$ for every word through length nine from $p_0$ and through length ten from $e/3$. This representation belongs to class~(I), so its agreement proves the claimed lower bounds against that class. \end{proof} \subsection{Reflection depth}\label{sec:five:depth} The route of Section~\ref{sec:five:budget} works for any finite family of atoms whose reflection orbit contains four atoms with $\Acal>0$. Use the orbit labels $\alpha=(g,i)$ of \coreref{sec:bri:recon}: $q_\alpha=U_{r_{g_1}}\cdots U_{r_{g_k}}p_i$ with $r_g=2p_g-e$, and $k$ is called the \emph{depth} of the label. \begin{corollary}[Reflection depth and length]\label{cor:five:depth} Let $p_1,\dots,p_n$ be atoms of $\J$, and suppose that four orbit points $q_{\alpha_1},\dots,q_{\alpha_4}$ with labels of depth at most $d$ have $\Acal>0$. The model class is (I). Then there is an explicit finite list of words in the tests $p_1,\dots,p_n$, each of length at most $11(2d+1)$, whose probabilities from $e/3$ no sequential projective representation of these tests reproduces; and an explicit list of words of the same maximal length whose probabilities from $\rho$ none reproduces, for every state $\rho$. \end{corollary} \begin{proof} For a label $\alpha=(g,i)$ of depth $k$, Lemma~\ref{lem:five:reflect}(a) applied $k$ times gives $U_{q_\alpha}=U_{r_{g_1}}\cdots U_{r_{g_k}}U_{p_i}U_{r_{g_k}}\cdots U_{r_{g_1}}$, and the same with $e-q_\alpha$, $e-p_i$; on the Hilbert side $P_\alpha=\Sigma_{g_1}\cdots \Sigma_{g_k}P_i\Sigma_{g_k}\cdots \Sigma_{g_1}$, with $\Sigma_g=2P_g-1$, has the matching factorisation. Insert these into the sequential products of a word of length $m\le11$ of Theorem~\ref{thm:eleven} for the four orbit points, as in the proof of Lemma~\ref{lem:five:reflect}(c). This produces at most $2dm$ reflections. Expanding each one as $2U_{p_g}+2U_{e-p_g}-\id$, resp.\ $2\mathcal C_{P_g}+2\mathcal C_{1-P_g}-\id$ with $\mathcal C_A(X)=AXA$, gives identities of the form \eqref{eq:fiv:transfer} with words of length at most $m(2d+1)\le11(2d+1)$ in the original tests. Let $\mathcal L$ be the set of all words so obtained from the list $\mathcal F_A$ of Theorem~\ref{thm:eleven}. If a sequential projective representation of $p_1,\dots,p_n$ reproduced $\Pr_{e/3}$ on $\mathcal L$, then the projections $P_{\alpha_1},\dots, P_{\alpha_4}$ with the same state would reproduce $\Pr_{e/3}$ on $\mathcal F_A$ for the four orbit points, against Theorem~\ref{thm:eleven}(a). The list obtained in the same way from the list $\mathcal F$ of Theorem~\ref{thm:eleven}(b) works from every state. \end{proof} For $\mathsf F$ one has $d=1$, and the bound is $33$; Proposition~\ref{prop:five:short} uses an explicit vector for the reflected quadruple instead of the list, and reaches eleven. \begin{remark}[Does every generating family reach $\Acal>0$?]\label{rem:five:orbit} It is not known whether the hypothesis of Corollary~\ref{cor:five:depth} always holds: \begin{quote} \emph{Question.} Let $S$ be a finite family of atoms that generates $\J$. Does its reflection orbit $\{q_\alpha\}$ always contain four atoms with $\Acal>0$? \end{quote} The orbit spans $\J$ (\coreref{lem:bri:orbit}), so it generates $\J$, and by {\coreref[(iii)]{thm:fourpoint}} the question asks whether some four of its points already generate $\J$. Generation by the family alone does not suffice: $\mathsf F$ generates $\J$ while all its four-element subfamilies have $\Acal=0$. For $\mathsf F$ the answer is yes at depth one (Section~\ref{sec:five:budget}). A sufficient condition is that the subgroup $\mathrm{Ref}(S)$ of $\Ffour=\Aut(\J)$ generated by the reflections $U_{r_p}$, $p\in S$, be dense in $\Ffour$. Indeed, the orbit is $\mathrm{Ref}(S)\cdot S$, since $U_{r_{U_rq}}=U_rU_{r_q}U_r$ by Lemma~\ref{lem:five:reflect}(a). If $\Acal$ vanished on all quadruples from $\mathrm{Ref}(S)\cdot S$, then for fixed $s_1,\dots,s_4\in S$ the continuous function $(g_1,\dots,g_4)\mapsto\Acal(g_1s_1,\dots,g_4s_4)$ would vanish on $\mathrm{Ref}(S)^4$, hence on $\Ffour^4$; since $\Ffour$ is transitive on atoms, $\Acal$ would vanish on all quadruples of atoms, which it does not (\coreref{prop:max}). A positive value at some $(g_1s_1,\dots,g_4s_4)$ with $g_t\in \mathrm{Ref}(S)$ involves only finitely many reflections, so a finite depth $d$ exists. When $\mathrm{Ref}(S)$ is dense is not known; it can be finite for special families, and whether a finite $\mathrm{Ref}(S)$ can occur for a family that generates $\J$ is open. \end{remark} \subsection{Quaternionic subalgebras: Helly number six}\label{sec:c:five:quat} \Coreref{cor:helly} says that five is the Helly number of the family of special (proper) unital subalgebras of $\J$: a set of atoms lies in one of them as soon as each of its subsets of at most five elements does. The analogous number for quaternionic subalgebras is six. \begin{proposition}[Six atoms on a line]\label{cor:c:hellyquat} The six atoms on the line $e-E_{33}$ \[ E_{11},\ p(1,1,0),\ p(1,i,0),\ p(1,j,0),\ p(1,ij,0),\ p(1,\ell,0) \] do not lie in a common unital subalgebra isomorphic to $\HH$, but each five of them do. Hence the Helly number of the family of quaternionic subalgebras of $\J$ (the least $h$ such that a set of atoms lies in a quaternionic subalgebra whenever each of its subsets of at most $h$ elements does), if it is finite, is at least six. \end{proposition} \begin{proof} The Bloch vectors $(1,0)$, $(0,1)$, $(0,i)$, $(0,j)$, $(0,ij)$, $(0,\ell)$ in $\R\oplus\Oct$ are linearly independent. By Section~\ref{sec:five:line} the six atoms generate $\R E_{33}\oplus\mathrm{JSpin}_6$, of dimension $8$, and each five of them generate $\R E_{33}\oplus\mathrm{JSpin}_5$, of dimension $7$. \emph{Six do not.} Suppose a unital subalgebra $V\cong\HH$ contains the six atoms. It contains the algebra they generate, hence $E_{33}$, which is then a primitive idempotent of $V$ (a Jordan frame of $V$ is one of $\J$). Its Peirce space $V\cap J_0(E_{33})$ is isomorphic to $H_2(\Ha)$, a spin factor $\R f\oplus\R^5$ of dimension $6$. But it would contain $\mathrm{JSpin}_6$, of dimension $7$. \emph{Five do.} The first five atoms have entries in the quaternion algebra $\spn\{1,i,j,ij\}$, so they lie in the standard copy of $\HH$. For any other five, the span of their Bloch vectors is a $5$-plane $\Pi$ in the nine-dimensional space of Bloch vectors. The stabiliser of $E_{33}$ in $\Ffour$ is $\Spin(9)$, which acts on this space through $SO(9)$~\cite[\S4.2]{Baez2002}, and $SO(9)$ is transitive on $5$-planes. An element mapping $\Pi$ to the Bloch span $\R\oplus\Ha$ of the first five maps the five atoms into the standard copy of $\HH$ (an atom on the line is determined by its Bloch vector), and its inverse maps that copy to a quaternionic subalgebra containing the five atoms. \end{proof} \begin{corollary}[Six suffices]\label{cor:c:hellysix} A set $S$ of atoms of $\J$, finite or infinite, lies in a unital subalgebra isomorphic to $\HH$ whenever each of its subsets of at most six elements does. Hence the Helly number of the family of quaternionic subalgebras of $\J$ is six. In particular, for model class~(I), if every subfamily of at most six tests of a finite family of atoms lies in a quaternionic subalgebra, the whole family has a sequential projective representation on $\C^6$ from every state of $\J$. \end{corollary} \begin{proof} Every subset of $S$ of at most five elements lies in a copy of $\HH$, a proper unital subalgebra, so by \coreref{cor:helly} $S$ lies in a proper unital subalgebra $B$. By \coreref{lem:five:sub}, $B$ lies in a copy of $\HH$, and then the proof is complete, or in a line subalgebra $\R q\oplus J_0(q)$ for an atom $q$. Assume the second. An atom $x$ of $\J$ in $\R q\oplus J_0(q)$ is $\lambda q+y$ with $y\in J_0(q)$; since $q\jp J_0(q)=0$ and $J_0(q)$ is a subalgebra, $x^2=x$ gives $\lambda\in\{0,1\}$ and $y^2=y$, and $\tr x=1$ leaves $x=q$ or $x=y$, an atom on the line $f=e-q$. Let $S'=S\setminus\{q\}$, a set of atoms on that line, and let $\Pi$ be the span of their Bloch vectors in the nine-dimensional traceless part of $J_0(q)\cong H_2(\Oct)$ (Section~\ref{sec:five:line}, with $q$ in place of $E_{33}$). \emph{$\dim\Pi\le5$.} Otherwise six members of $S'$ have independent Bloch vectors $u_1,\dots,u_6$. By hypothesis they lie in a unital $V\cong\HH$. By Section~\ref{sec:five:line} the algebra they generate with $e$ is $\R q\oplus\R f\oplus\spn\{u_k\}$, so $V$ contains $q$, $f$ and the $u_k$. The idempotent $q$ is primitive in $V$, since a decomposition into two nonzero orthogonal idempotents of $V$ would split the trace one of $q$ into two positive integers. The Peirce projections of $q$ are polynomials in the multiplication operator of $q$ and preserve $V$, so the Peirce-$0$ space of $q$ in $V$ is $V\cap J_0(q)$, which for a primitive idempotent of $\HH$ is a copy of $H_2(\Ha)$, of dimension six. It contains the seven-dimensional space $\R f\oplus\spn\{u_k\}$, a contradiction. \emph{Placing $S$.} Choose $g\in\Ffour$ with $gq=E_{33}$; $\Ffour$ is transitive on atoms. Then the members of $gS'$ lie on the line $e-E_{33}$, with Bloch span $g\Pi$ of dimension at most five. Enlarge it to a $5$-plane. The stabiliser $\Spin(9)$ of $E_{33}$ acts on the Bloch vectors through $SO(9)$~\cite[\S4.2]{Baez2002}, which is transitive on $5$-planes; pick $h$ in it mapping the plane onto $\R\oplus\Ha$, the Bloch span of the first five atoms of Proposition~\ref{cor:c:hellyquat}. An automorphism fixing $E_{33}$ fixes $f$ and preserves $J_0(E_{33})$ and the trace form, so it maps the atom $\frac12(f+u)$ to $\frac12(f+hu)$. Every member of $hgS'$ is therefore an atom on the line whose off-diagonal entry lies in $\Ha$, hence lies in the standard copy of $\HH$, and so does $hgq=E_{33}$. So $S$ lies in $(hg)^{-1}\HH$, a unital subalgebra isomorphic to $\HH$. With Proposition~\ref{cor:c:hellyquat} the Helly number is six. For the last sentence, the family lies in a unital $Q\cong\HH$, and the construction in the proof of {\coreref[(a)]{thm:pentad}} (an isomorphism onto $\HH$ followed by the symplectic embedding $\iota_{\Ha}:\HH\to H_6(\C)$, with the state restricted to $Q$, transported and extended to a density matrix) gives the representation. \end{proof} The operational statement is one-directional: a sequential projective representation on $\C^6$ does not by itself place a family in a quaternionic subalgebra. \paragraph{Maximal subalgebras.} For comparison with \coreref{lem:five:sub}, which proves the containment statement needed for \coreref{cor:helly} directly: the maximal proper unital subalgebras of the simple Euclidean Jordan algebras of rank three are, with $q$ an atom and $\R q\oplus H_2(\mathbb K)$ the line subalgebra $\R q\oplus J_0(q)$ of $H_3(\mathbb K)$, \begin{center} \begin{tabular}{ll} \toprule Ambient algebra & Maximal proper unital subalgebras\\ \midrule $\HR$ & $\R q\oplus H_2(\R)$\\ $\HC$ & copies of $\HR$; $\ \R q\oplus H_2(\C)$\\ $\HH$ & copies of $\HC$; $\ \R q\oplus H_2(\Ha)$\\ $\HO$ & copies of $\HH$; $\ \R q\oplus H_2(\Oct)$\\ \bottomrule \end{tabular} \end{center} For $\J$ this is Racine's classification of maximal subalgebras of exceptional Jordan algebras~\cite{Racine1977}, as used in the compact setting by Correia, Pacheco and Svensson~\cite{CorreiaPachecoSvensson2021}; for $H_3(\mathbb K)$ with $\mathbb K$ associative it is the standard classification of compact Jordan subalgebras. One tempting shortcut fails: an orthogonal frame generates only $\R^3$, so not every non-collinear real triple of atoms generates $\HR$. \subsection{Local and global}\label{sec:c:five:local} The configuration $\mathsf F$ is an exact failure of local algebraic compatibility to extend globally; \coreref{sec:fiv:local} compares it with marginal problems and joint measurability. \paragraph{It is not a cohomology class.} It is tempting to read $\mathsf F$ as a nonzero obstruction class built from local data. No such construction is known, and the most natural candidate fails. Regard $\Acal$ as a function on ordered quadruples of atoms, a $3$-cochain, and ask whether it is a cocycle. The \emph{facets} of an ordered five-atom family are its five four-element subfamilies. Take \[ \mathsf C=\bigl(E_{11},\ p(1,1,0),\ p(1,i,1),\ p(1,j,\ell),\ E_{33}\bigr). \] Every facet containing $E_{33}$ also contains $E_{11}$ or $p(1,1,0)$, both orthogonal to $E_{33}$, so $\Acal$ vanishes there ({\coreref[(ii)]{thm:fourpoint}}). The facet omitting $E_{33}$ is the configuration $\mathsf S$, with $\Acal=\frac1{5832}$. The facet values are $(0,0,0,0,\frac1{5832})$, and the alternating coboundary is \[ (\delta\Acal)(\mathsf C)=\sum_{k=0}^{4}(-1)^k\Acal(\mathsf C^{\hat k})=\tfrac{1}{5832}\neq0 , \] where $\mathsf C^{\hat k}$ omits the member of index $k$. So $\Acal$ is not a cocycle. The set-valued presheaf that assigns to each subfamily its set of quaternionic subalgebras containing it, with restriction maps given by inclusion, is nonempty on the whole family exactly when a common quaternionic subalgebra exists. That is a reformulation, not an invariant: the full simplex on the family is contractible, so ordinary cohomology with constant coefficients vanishes. \paragraph{Joint measurability.} Beyond the comparison in \coreref{sec:fiv:local}, every joint-measurability hypergraph is realised by quantum observables~\cite{KunjwalHeunenFritz2014}. Such statements concern the existence of a joint observable inside a fixed Hilbert space; the present one concerns the existence of any sequential projective representation at all, for a family whose members already have representations in groups of four. \paragraph{A projective look-alike.} Four points in general position carry no projective invariant (Remark~\ref{rem:fp:metric}). Placing a quadrangle at $(1,0,0)$, $(0,1,0)$, $(0,0,1)$, $(1,1,1)$, a fifth point $(1,x,y)$ lies, by Artin's theorem, in the subplane over the associative algebra generated by $x$ and $y$, while a sixth generic point does not. This concerns the projective group, of dimension $78$, and subplanes. The invariant and Helly numbers here concern the metric group $\Ffour$ and Jordan subalgebras. \section{Three and four atoms}\label{sec:c:four} This section collects remarks on three and four atoms of $\J$ that the core paper does not need: what published results on three idempotents give, the law of a random atom seen in a complex subalgebra, and what kind of invariant $\Acal$ is. The notation is that of \coreref{sec:four}: the thirteen traces of \coreref{def:traces}, the four-point invariant $\Acal$, and a complex subalgebra $W$ containing three of the atoms (for instance $W=\Jord\langle e,p_1,p_2,p_3\rangle$ when $\Delta_W\ne0$, \coreref{prop:windowdet}), with its trace-orthogonal projection $P_W$. \subsection{What the published results on three idempotents give}\label{sec:c:four:three} \Coreref{prop:three} puts any three atoms of $\J$ in a complex subalgebra. Several papers bound or identify the algebra generated by three idempotents. The proof of \coreref{prop:three} uses none of them; they come close, and stop as follows. Gorshkov and Staroletov~\cite{GorshkovStaroletov2020} consider an algebra over a field of characteristic not two generated by three idempotents $a,b,c$ each of whose $1$-eigenspaces under multiplication is one-dimensional; atoms of $\J$ are of this kind. For Jordan algebras their Theorem~1 gives that the algebra is spanned by $a,b,c,ab,bc,ac,(ab)c,(ac)b,(bc)a$, and their Theorem~2 that, if it is nine-dimensional and simple, it is $M_3(F)^+$ or a Hermitian form $H(M_3(F(z)),j)$. Over $\R$, inside a formally real algebra, only the real form $\HC$ of the second type survives; that inference is not made there, and Theorem~2 does not treat dimensions below nine. Roelands and Tiersma~\cite{RoelandsTiersma2026} (Theorem~6.3) prove that a JB-algebra generated by two projections, one further element and the unit is a JC-algebra, that is, special. In particular three projections generate a special algebra. Their proof of the related Theorem~6.2 shows that after an automorphism a third atom has entries in a proper associative subalgebra of $\Oct$ (they quote van Gaans, Kalauch and Roelands~\cite{vanGaansKalauchRoelands2024}). That coordinate algebra may be quaternionic and is not visibly complex. So these theorems give that three projections generate a special algebra, and that a nine-dimensional simple algebra generated by three such idempotents is a form of $M_3^+$. Containment of every triple of atoms, degenerate ones included, in one complex subalgebra is \coreref{prop:three}. \subsection{What one complex subalgebra sees}\label{sec:fp:law} \Coreref{prop:windowdet} views one atom through one complex subalgebra; here the atom is random. A random atom is \emph{uniformly distributed} if its law is the $\Ffour$-invariant probability measure on $\OP$ (unique, as $\Ffour$ is transitive on atoms). Write $\nu_{3,k}$ for the law of $ZZ^\dagger/\Tr(ZZ^\dagger)$, where $Z$ is a $3\times k$ complex matrix with independent standard complex Gaussian entries (density $\pi^{-1}e^{-|z|^2}$). It is the law of the reduced state on $\C^3$ of a uniformly random pure state of $\C^3\otimes\C^k$: the induced measure of \.Zyczkowski and Sommers~\cite{ZyczkowskiSommers2001}. \begin{proposition}[Law of a pure state seen in a complex subalgebra]\label{prop:windowlaw} Let $W\subseteq\J$ be a complex subalgebra, $\psi:W\to H_3(\C)$ a unital Jordan isomorphism and $P_W$ the trace-orthogonal projection onto $W$. If $p$ is a uniformly distributed atom of $\J$, then $\psi(P_Wp)$ has the law $\nu_{3,4}$. More generally, if $V$ is a unital subalgebra with $W\subseteq V\subseteq\J$ and $V\cong H_3(\mathbb K)$, $\mathbb K\in\{\C,\Ha,\Oct\}$, and $p$ is an atom of $V$ distributed according to the $\Aut(V)$-invariant probability measure, then $\psi(P_Wp)$ has the law $\nu_{3,k}$ with $k=\frac12\dim_\R\mathbb K\in\{1,2,4\}$; so $P_Wp$ has rank $\min(3,k)$ almost surely, and for $V=\J$ its eigenvalues have density proportional to $\lambda_1\lambda_2\lambda_3\prod_{a0$, \eqref{eq:fp:riesz} gives \[ \int e^{-\Tr(\psi(T)\,\psi(P_Wx))}\,\mu(dx)=N(e+T)^{-4}=\det\bigl(I+\psi(T)\bigr)^{-4}, \] which is the Laplace transform of $ZZ^\dagger$ for a $3\times4$ standard complex Gaussian matrix $Z$; normalising the trace gives $\nu_{3,4}$. The number $k=4=\frac12\dim_\R\Oct$ is the first nonzero point $d/2$ of the Wallach set of the cone of $\J$, the set of exponents $s$ for which $N(y)^{-s}$ is the Laplace transform of a positive measure. \subsection{What kind of invariant this is}\label{sec:fp:kind} \begin{remark}[The invariant is metric, not projective]\label{rem:fp:metric} One might expect the three-versus-four threshold to be a fact of projective geometry, since the octonionic plane famously fails Desargues' theorem. It is not. The collineation group of $\OP$ (the octonionic projective plane, realised in $\J$ by the rank-one elements up to scale~\cite{Springer1960}), which acts on $\J$ through the group $\Esix$ of linear maps preserving the cubic norm, is transitive on quadrangles, that is, on ordered quadruples of points no three of which are collinear (Pickert~\cite{Pickert1955}, 7.3.14; see Knarr and Stroppel~\cite{KnarrStroppel2011}, Corollary~1.8). So every collineation-invariant function of four points is constant on quadrangles. Quadrangles form a dense open set, so every continuous one is constant, and there is no nonconstant continuous projective four-point invariant. The invariant $\Acal$ is invariant only under the smaller group $\Ffour$, the stabiliser of $e$ in $\Esix$, whose elements preserve the trace form, that is, transition probabilities. It is not constant on quadrangles: it vanishes on quaternionic quadrangles and equals $\frac4{19683}$ at the maximiser (\coreref{eq:fp:maximiser}). The obstruction is therefore a statement about transition probabilities, which is what sequential tests measure. \end{remark} \begin{remark}[Relation to invariant theory]\label{rem:fp:invariants} For three generic elements of the (complexified) Albert algebra, Iltyakov~\cite{Iltyakov1997} (Theorem~1, with Propositions~1 and~2, the first credited there to I.~P. Shestakov) shows that the $\Ffour$-invariants are generated by $29$ traces of Jordan monomials together with one invariant $\omega$ of degree nine satisfying a quadratic equation over them. The registry entry~\cite{MacGregorPhi2026} uses a related signed degree-nine invariant \[ \Phi(x,y,z)=\ip{[U_x,U_y]z}{U_z(x\jp y)}, \] defined for any three elements of $\J$. It has degree three in each argument, changes sign when $x$ and $y$ are exchanged, and $\Phi\ne0$ implies that $e,x,y,z$ generate $\J$. Two differences from the present situation are worth stating. First, $\Phi$ vanishes on every triple of atoms, so it cannot see primitive-idempotent data. Indeed, three atoms lie in a complex subalgebra $W$ (\coreref{prop:three}), which is closed under $\jp$ and every $U_a$, $a\in W$. Under a unital Jordan isomorphism $\psi:W\to H_3(\C)$, $U_ab$ becomes $ABA$ (capital letters for images under $\psi$), and $\ip ab=\tau(AB)$ with the real-linear cyclic trace $\tau=\operatorname{Re}\Tr$ on $M_3(\C)$. With the matrix products $P=XYZ$ and $Q=YXZ$, the two arguments of $\Phi$ become $XYZYX-YXZXY$ and $\frac12Z(XY+YX)Z$, so \[ 2\Phi(x,y,z)=\tau\bigl(PQP+PQQ-QPP-QPQ\bigr)=0, \] because $\tau(PQP)=\tau(QPP)$ and $\tau(QPQ)=\tau(PQQ)$ by cyclicity. This cancellation is valid on any Jordan subalgebra of an associative algebra with a cyclic trace. Second, $\Phi$ is a signed quantity, while $\Acal$ is a complete squared norm whose zero set is exactly the special locus. Both are built from octonion associators; no precise identity between them is known. Iltyakov~\cite[Thm.~1, Cor.~1]{Iltyakov1997laplace} shows that for any number of elements the invariants are generated by trace polynomials and their images under iterated Laplace operators, in degrees at most $85$; for four elements this follows by setting the other arguments to zero. Trace polynomials generate the invariant field, and the ring is integral over them (Elduque and Iltyakov~\cite{ElduqueIltyakov1999}), but from five elements on they do not generate the ring~\cite[Prop.~8]{Iltyakov1997laplace}. The thirteen traces answer a different question: for four atoms they determine $\Acal$, and $\Acal$ decides generation. For the smaller groups $G_2=\Aut(\Oct)$ and $\Spin(7)$ acting on octonions, generators and relations of the invariants are given by Schwarz~\cite{Schwarz1988}. \end{remark} \section{Instruments without the projection postulate}\label{sec:c:sharp} \Coreref{sec:ins} shows that the statistics of ideal tests on $\J$ have a finite completely positive realization with non-L\"uders instruments (\coreref{thm:cp69}), on $\C^{17n+1}$ for $n$ tests, and that the instruments can be taken to have sharp effects, one fixed projection per test, with arbitrary channels after the outcome (\coreref{prop:ins:sharpdilation}, \coreref{cor:ins:sharpalbert}). The model class there is class~(II), \coreref{def:memorymodel}; its state on $\C^d$ is called the \emph{memory state}. This section adds four things: on the states reachable from $e/3$ the Albert tests are ideal; the existence of a realization of fixed dimension is decided by finite menus; repeatability can always be imposed; and the dimension of the sharp realization can be lowered to $d+\max_i\operatorname{rank}(A_i-A_i^2)$, which is $26n-7$ for $n\ge1$ tests (Corollary~\ref{cor:c:sharp26}). \subsection{Ideality on reachable states}\label{sec:ins:reachable} \provisionalref{Reachable ideality}{prop:ins:reachable}{Same statement}{\leandecl{native_reachable_sharpness}, \leandecl{native_reachable_certainty_ideality}; all assertions together, including positivity on the cone and the trace-preserving sum: \leandecl{nativeReachable}\allowbreak\leandecl{IdealityObligations}.} \begin{proposition}[Ideality on reachable states]\label{prop:ins:reachable} Fix a family of idempotent tests on $\J$ and start at $e/3$. Let $C$ be the closed convex hull of the normalized nonzero states reached by finite outcome words, including the empty word. On the cone generated by $C$, the restricted pairs $(U_p,U_{e-p})$ are positive instruments satisfying (Id) and (Sh) of \coreref{def:ideal}, with effects taken to be \emph{all} affine functions $f:C\to[0,1]$. This is a statement about the restricted state space $C$, not the full state space of a completely positive realization. \end{proposition} \begin{proof} Write $G_w$ for the composite update along $w$. The states in $C$ are positive in $\J$ and have trace one. Each branch preserves their cone: appending a letter preserves the generating rays, and linearity and continuity preserve their conic hull and closure. The complementary maps have trace-preserving sum. If a state in $C$ has outcome probability one or zero, the support lemma (\coreref{lem:set:support}) gives $U_p\rho=\rho$ or $U_{e-p}\rho=\rho$, respectively. This proves (Id). For a fixed finite sequence of tests, summing its unnormalized branch states gives \[ \sum_{w\text{ over these tests}}G_w(e/3)=e/3, \] because every nonselective map $U_p+U_{e-p}$ fixes $e$. Thus $e/3$ is a convex combination, with strictly positive weights, of that sequence's normalized nonzero branch states. Any nonnegative affine effect vanishing at $e/3$ vanishes on every such state. It therefore vanishes on $C$: bounded affine functions on a finite-dimensional convex set extend continuously to its affine hull, so convex combinations and closure preserve the equality. For a nontrivial test $p$ of rank $r$, the reachable states $p/r$ and $(e-p)/(3-r)$ lie in its two certainty faces and \[ e/3=(r/3)(p/r)+((3-r)/3)((e-p)/(3-r)). \] An affine effect zero on both faces is consequently zero at $e/3$, hence on all of $C$. It cannot take value one at another state of $C$, which proves (Sh). For $p=0$ or $p=e$, one certainty face is already all of $C$. \end{proof} This restricted ordered space need not be a Euclidean Jordan algebra with its full positive cone, so the ambient hypotheses of \coreref{thm:ideal} remain essential. The realization of \coreref{thm:cp69} reproduces the original history table; that does not provide an affine encoding of $C$ or implement every affine effect on $C$. Different histories representing the same Albert state may have different memory states. \subsection{A fixed memory bound across menus}\label{sec:ins:compact} The dependence on the menu can be separated from the choice of a realization by a standard compactness argument. The preparation labels in the following statement need not form an Albert state space, and no affine encoding of those labels is assumed. \provisionalref{Fixed dimension}{prop:ins:compactness}{Same statement}{\leandecl{stationaryKraus_global_iff_local_finite_menus}; the error bound for realizations of the finite menu: \leandecl{stationaryKraus_finite_menu_error_gap}; both in Kraus form.} \begin{proposition}[Fixed-dimensional compactness]\label{prop:ins:compactness} Fix $D\ge1$, sets of preparation labels $\mathcal P$ and test labels $\mathcal X$, and a nonempty finite outcome set $A_x$ for each $x\in\mathcal X$. Specify target joint probabilities $t(p,w)$ for every $p\in\mathcal P$ and every finite outcome-labelled word $w$. A \emph{stationary} (time-homogeneous) $D$-dimensional CP realization consists of a density matrix $\sigma_p$ on $\C^D$ for each $p$, and one CP instrument $(\Phi_{x,a})_{a\in A_x}$ per test, with trace-preserving sum, such that \[ \Tr\bigl(\Phi_{x_m,a_m}\cdots\Phi_{x_1,a_1}(\sigma_p)\bigr)=t(p,w), \qquad w=((x_1,a_1),\ldots,(x_m,a_m)). \] The same maps are used at every occurrence of a test. Such a realization exists for all labels if and only if it exists for every finite set of preparation and test labels, matching every word over that finite menu. The finite-menu realizations need not be chosen compatibly. If no global realization exists at this fixed $D$, there are finitely many pairs $(p_i,w_i)$ and a number $\delta>0$ such that every $D$-dimensional stationary CP realization of their finite menu has \[ \max_i\left|\Pr(p_i,w_i)-t(p_i,w_i)\right|\ge\delta. \] The conclusion supplies neither a quantitative value of $\delta$ nor necessarily a linear inequality in the word probabilities. \end{proposition} \begin{proof} The set $S_D$ of density matrices is compact. Use the unnormalized Choi matrix $C(\Phi)=\sum_{j,k}|j\rangle\langle k|\otimes\Phi(|j\rangle\langle k|)$. The instrument set $I_x$ consists of tuples $C_a\ge0$ satisfying $\Tr_{\mathrm{out}}\sum_a C_a=I_D$. It is closed and bounded, since $\sum_a\Tr C_a=D$, and is therefore compact. The product \[ K=\prod_{p\in\mathcal P}S_D\ \times\!\prod_{x\in\mathcal X}I_x \] is compact. Each word probability is a continuous polynomial in finitely many coordinates of $K$, so each prescribed equality defines a closed subset. Any finite collection of these equalities uses finitely many labels and is satisfied by the corresponding finite-menu realization, extended arbitrarily to the unused coordinates. The finite intersection property proves that all equalities hold at one point of $K$. The converse is restriction. If the intersection is empty, a finite subcollection already has empty intersection. On the compact model space for its finite set of labels, the continuous maximum absolute error attains a minimum. That minimum is positive, since a zero would satisfy the subcollection. This proves the second assertion. \end{proof} A uniform upper bound $D$ on the finite-menu dimensions is sufficient: embed each smaller model into a $D$-dimensional block, initialize in that block, and extend each instrument to the unused block by a channel (a completely positive trace-preserving map) assigned to one outcome, discarding cross-block coherences. This extension is CP and preserves every word probability. The proposition imposes no continuity or symmetry in the labels. It does not supply a uniform $D$, and the projective no-go theorem does not supply its premise of nonexistence in the larger CP class. \subsection{Repeatability and a smaller sharp dilation}\label{sec:ins:repair} The notation is that of \coreref{sec:ins:dilation}: a binary instrument $(\mathcal E_+,\mathcal E_-)$ on $M_d(\C)$ with effect $A=\mathcal E_+^*(1)$, and $\iota x=(x,0)$. \begin{lemma}[Repeatability can be imposed]\label{lem:ins:repair} Let $(\sigma,\mathcal E_{i,\pm})$ be a $d$-dimensional completely positive realization (\coreref{def:memorymodel}) of a process with $\Pr(w\,(i,+)\,(i,-))=\Pr(w\,(i,-)\,(i,+))=0$ for all words $w$ and tests $i$. Then there is a $d$-dimensional completely positive realization of the same process whose instruments are repeatable on all states. Its effects are the $A_i$, except that a test whose outcome $+$ (respectively $-$) has probability zero after every word may be replaced by the trivial test with effect $0$ (respectively $1$). \end{lemma} \begin{proof} Call $\sigma_w=\mathcal E_w\sigma$ the reachable states. Fix $i$, and let $E_1$, $E_0$ be the eigenvalue-$1$ space and the kernel of $A=A_i$. For a reachable $\sigma_w$, the matrix $Y=\mathcal E_{i,+}(\sigma_w)\ge0$ satisfies $\Tr((1-A)Y)=\Pr(w\,(i,+)\,(i,-))=0$, so $(1-A)^{1/2}Y^{1/2}=0$ and $Y$ is supported in $E_1$; likewise $\mathcal E_{i,-}(\sigma_w)$ is supported in $E_0$. If $E_1\ne0$, let $\Pi$ be the projection onto $E_1$, $s\in E_1$ a unit vector and $\mathcal C(Y)=\Pi Y\Pi+\Tr((1-\Pi)Y)\,|s\rangle\langle s|$, a channel that is the identity on matrices supported in $E_1$. Replace $\mathcal E_{i,+}$ by $\mathcal C\circ\mathcal E_{i,+}$: its effect is still $A$, since $\mathcal C$ is trace preserving; its outputs lie in $E_1$; and it agrees with $\mathcal E_{i,+}$ on reachable states. If $E_1=0$, then $\mathcal E_{i,+}(\sigma_w)=0$ for every $w$, and the instrument is replaced by $(0,\mathcal E_{i,+}+\mathcal E_{i,-})$, which agrees with it on reachable states. Treat the outcome $-$ in the same way. ($E_1$ and $E_0$ cannot both be zero, since $\Tr\mathcal E_{i,+}(\sigma)+\Tr\mathcal E_{i,-}(\sigma)=1$.) By induction on $w$ the new realization has the same reachable states, hence the same word probabilities. \end{proof} For ideal tests the hypothesis holds, because $U_{e-p}U_p=U_pU_{e-p}=0$. An instrument that is repeatable on all states sends positive matrices into the eigenvalue-$1$ space of its effect on the outcome $+$ and into the kernel on the outcome $-$, which is the hypothesis of {\coreref[(d)]{prop:ins:sharpdilation}}. So after this repair (d) applies to every finite realization of the statistics of ideal tests, not only to the one of \coreref{thm:cp69}, where the hypothesis holds by construction. \paragraph{Failure of ideality on a reachable state.} \Coreref{rem:ins:sharpnotideal} exhibits a state on which ``yes'' is certain but which the dilated instrument moves; that state is not reachable from the initial state. Ideality also fails on a reachable state when the initial state is chosen suitably. For two atoms $p_i\ne p_j$ with $o_{ij}>0$, let $r=e-(p_i\vee p_j)$, the atom orthogonal to both, and start in $\rho=r$. In the realization of \coreref{thm:cp69}, after ``no'' on $p_i$ the memory state lies in the block of the Peirce space $J_0(p_i)$ (\coreref{app:mem:cp69}). On it ``no'' on $p_j$ is certain, since $\tr(p_j\jp r)=0$, but this outcome moves the memory state to the block of $J_0(p_j)$. \begin{proposition}[A smaller sharp dilation]\label{prop:c:sharpdim} In the situation of \coreref{prop:ins:sharpdilation}, with $n\ge1$, put $m=\max_i\operatorname{rank}(A_i-A_i^2)$. The conclusions (a)--(d) of that proposition hold with $\mathcal K'=\C^d\oplus\C^m$ in place of $\mathcal K_2=\C^d\oplus\C^d$. Conversely, a projection $P$ on $\C^d\oplus\C^{m'}$ with $\iota^*P\iota=A$ requires $m'\ge\operatorname{rank}(A-A^2)$. If $\sigma$ and all Kraus operators are real matrices, all operators constructed can be chosen real, as in \coreref{prop:ins:sharpdilation}. \end{proposition} \begin{proof} Fix $i$ and drop it; the trivial cases $A\in\{0,1\}$ are treated as in the core. Let $E_0$, $E_1$ and $E_{\rm mid}$ be the spectral subspaces of $A$ for $\{0\}$, $\{1\}$ and $(0,1)$, so $\operatorname{rank}(A-A^2)=\dim E_{\rm mid}\le m$. Let $T:\C^d\to\C^m$ be a partial isometry with initial space $E_{\rm mid}$, so $T^*T=\Pi_{\rm mid}$, the projection onto $E_{\rm mid}$. Put \[ \iota_+=\begin{pmatrix}\sqrt A\\ T\sqrt{1-A}\end{pmatrix},\qquad \iota_-=\begin{pmatrix}\sqrt{1-A}\\ -T\sqrt A\end{pmatrix}:\ \C^d\to\C^d\oplus\C^m . \] All of $A$, $\sqrt A$, $\sqrt{1-A}$ and $\Pi_{\rm mid}$ commute. On $E_1$, $E_0$, $E_{\rm mid}$ one reads off $\iota_+^*\iota_+=A+(1-A)\Pi_{\rm mid}=\Pi_{01}$, the projection onto $E_1\oplus E_{\rm mid}$ (the support of $A$); $\iota_-^*\iota_-=(1-A)+A\Pi_{\rm mid}$, the projection onto the support of $1-A$; and $\iota_+^*\iota_-=\sqrt{A(1-A)}\,(1-\Pi_{\rm mid})=0$. So $\iota_+$, $\iota_-$ are partial isometries with orthogonal ranges; $P=\iota_+\iota_+^*$ and $\iota_-\iota_-^*$ are orthogonal projections with $P\iota_-=0$, and $Q_0=1-P-\iota_-\iota_-^*$ is a projection, of rank $m-\dim E_{\rm mid}$. Moreover $\iota^*P\iota =\sqrt A\sqrt A=A$, $\iota_+^*\iota =\sqrt A$ and $\iota_-^*\iota =\sqrt{1-A}$, so \[ P\iota =\iota_+\sqrt A,\qquad (1-P)\iota =\iota -\iota_+\sqrt A=\iota_-\sqrt{1-A}, \] the second by comparing the two components. Choose $\mathcal F_\pm$ by \coreref{lem:ins:factor} as in the core, a density matrix $\tau_+$ on $\mathcal K'$, and a density matrix $\tau_0$ on $\C^d$, supported in $E_0$ whenever $E_0\ne0$ (under the hypothesis of (d) this is always the case), and put \begin{align*} \Lambda_+(Z)&=\iota \mathcal F_+(\iota_+^*Z\iota_+)\iota^*+\Tr\bigl((1-P)Z\bigr)\tau_+,\\ \Lambda_-(Z)&=\iota \mathcal F_-(\iota_-^*Z\iota_-)\iota^*+\Tr\bigl((1-\iota_-\iota_-^*)Z\bigr)\iota \tau_0\iota^*. \end{align*} Since $\iota_+\iota_+^*=P$ and $\mathcal F_\pm$ are trace preserving, both are channels. Now $\Lambda_+(PZP)=\iota \mathcal F_+(\iota_+^*Z\iota_+)\iota^*$, because $\iota_+^*P=\iota_+^*$ and $(1-P)P=0$; and, since $\iota_-^*(1-P)=\iota_-^*$ and $(1-\iota_-\iota_-^*)(1-P)=Q_0$, $\Lambda_-((1-P)Z(1-P))=\iota \mathcal F_-(\iota_-^*Z\iota_-)\iota^*+ \Tr(Q_0Z)\,\iota \tau_0\iota^*$. Every output lies in $\iota M_d(\C)\iota^*$, which is (b). For $X\in M_d(\C)$, \[ \iota_+^*P\iota X\iota^*P\iota_+=\iota_+^*\iota_+\sqrt AX\sqrt A\iota_+^*\iota_+=\sqrt AX\sqrt A \] because $\sqrt A$ is supported in the support of $A$; likewise $\iota_-^*(1-P)\iota X\iota^*(1-P)\iota_-=\sqrt{1-A}X \sqrt{1-A}$, and $Q_0(1-P)\iota =Q_0\iota_-\sqrt{1-A}=0$. Hence $\widetilde{\mathcal E}_\pm(\iota X\iota^*)=\iota \mathcal E_\pm(X)\iota^*$ by \coreref{lem:ins:factor}, which is (a); (c) follows by induction on words as in the core. For (d), take the auxiliary states of $\mathcal F_+$ and $\mathcal F_-$ in $E_1$ and $E_0$ as in the core, and note that $\tau_0$ is supported in $E_0$. For $x\in E_1$, $(1-P)\iota x=\iota_-\sqrt{1-A}\,x=0$, and for $x\in E_0$, $P\iota x=\iota_+\sqrt A\,x=0$; so $\iota E_1\subseteq\operatorname{ran}P$ and $\iota E_0\subseteq\operatorname{ran}(1-P)$, and all outputs of $\widetilde{\mathcal E}_+$, resp.\ $\widetilde{\mathcal E}_-$, lie in these ranges. For the converse, write $P=\left(\begin{smallmatrix}A&B\\B^*&C\end{smallmatrix}\right)$. The upper left corner of $P^2=P$ gives $A^2+BB^*=A$, so $BB^*=A-A^2$ and $m'\ge\operatorname{rank}B\ge\operatorname{rank}(A-A^2)$. For the last sentence, every operator above is obtained from the given matrices by real matrix operations and the spectral theorem for real symmetric matrices; choose $T$, $\tau_\pm$, $\tau_0$ real, and $\mathcal F_\pm$ real as in the core. \end{proof} \begin{corollary}[Sharp effects in dimension $26n-7$]\label{cor:c:sharp26} Let $p_1,\dots,p_n$, $n\ge1$, be atoms of $\J$ in any configuration and $\rho$ any state. There are a density matrix on $\C^{26n-7}$ and, for each test $j$, one orthogonal projection $P_j$ and two channels $\Lambda_{j,\pm}$, all real matrices, such that the instruments $Z\mapsto\Lambda_{j,+}(P_jZP_j)$ and $Z\mapsto\Lambda_{j,-}((1-P_j)Z(1-P_j))$ reproduce $\Pr_\rho(w)$ for every word and are repeatable on all states. For four tests the dimension is $97$. \end{corollary} \begin{proof} Apply Proposition~\ref{prop:c:sharpdim} to the realization on $\C^{17n+1}$ given by \coreref{thm:cp69}, whose data are real and whose instruments satisfy the hypothesis of {\coreref[(d)]{prop:ins:sharpdilation}} by construction, as in the proof of \coreref{cor:ins:sharpalbert}. There the effect $A_j$ of ``yes'' on $p_j$ is block diagonal (\coreref{app:mem:cp69} and \coreref{rem:ins:notideal}): it is the scalar $\tr(p_j\jp\rho)$ on the one-dimensional initial block, $o_{jk}$ on the reset block of $p_k$, $1$ on that of $p_j$, $0$ on the block of $J_0(p_j)$, a projection on the block of $J_0(p_i)$ if $o_{ij}=0$, and it has spectrum $\{0,1-o_{ij}\}$, each eigenvalue eight times, on that block if $o_{ij}>0$. So $\operatorname{rank}(A_j-A_j^2)\le1+(n-1)+8(n-1)=9n-8$. The proof of Proposition~\ref{prop:c:sharpdim} uses only $m\ge\operatorname{rank}(A_i-A_i^2)$ for every $i$, so it applies with $m=9n-8$ and gives $\C^{17n+1}\oplus\C^{9n-8}=\C^{26n-7}$, with real data, repeatable instruments and the same word probabilities from the same initial state. \end{proof} \begin{corollary}[A smaller realization for initial states without a Peirce-half component] \label{cor:c:sharpnohalf} Under the initial-state hypothesis of \coreref{cor:ins:nohalf}, the ideal-test statistics have a real sharp-effect realization, repeatable on all states, in dimension $26n-9$. For four tests from $e/3$ this is $95$. \end{corollary} \begin{proof} Use the $17n$-dimensional construction of \coreref{cor:ins:nohalf}, which deletes the initial ray and keeps every reset and spinor block. The effect spectra listed in the proof of Corollary~\ref{cor:c:sharp26} then give \[ \operatorname{rank}(A_j-A_j^2)\le(n-1)+8(n-1)=9(n-1). \] Indeed, the reset ray for $p_j$ contributes zero, the other reset rays contribute at most one each, the spinor block for $p_j$ contributes zero, and every other spinor block contributes at most eight; a zero overlap gives a projection and can only reduce the count. Proposition~\ref{prop:c:sharpdim} therefore gives dimension at most $17n+9(n-1)=26n-9$ (with $m=9(n-1)$, as in that proof), with the same real-data and repeatability properties. The initial state and all word probabilities are preserved by the dilation. \end{proof} These are upper bounds for the stated initial states, with no assertion of optimality. The $69$, $138$ and $97$ bounds above remain valid for arbitrary initial states; $68$, $136$ and $95$ use the hypothesis of \coreref{cor:ins:nohalf}. Since a sharp realization is in particular a completely positive realization, \coreref{prop:ins:lower} gives the lower bound six for the quadruple and state treated there. Whether sharp effects force a larger dimension than unrestricted instruments is open. \clearpage \section{Related work}\label{sec:prior} The core paper surveys reconstructions, ideal measurements, temporal correlations and octonionic quantum mechanics. This section covers the literature closest to the results of the present paper. \paragraph{Certificates and self-testing.} Sum-of-squares certificates in noncommutative variables (Helton~\cite{Helton2002}; Helton and McCullough~\cite{HeltonMcCullough2004}), the degree bound of Klep, Levenson and McCullough~\cite[Thm.~1.6]{KLM2025} for the algebra of $n$ projections, and exact rounding (Peyrl and Parrilo~\cite{PeyrlParrilo2008}; Cafuta, Klep and Povh~\cite{CafutaKlepPovh2015}) are discussed in Section~\ref{sec:cer:sound}. Saha, Xu and Cabello~\cite{SahaXuCabello2026} self-test ideal quantum instruments using Bell correlations with an intermediate measurement device. Their certification includes the outcome operations, up to local extraction isometries and auxiliary systems. This addresses calibration within quantum theory; the present criterion concerns the existence of a sequential projective representation. Their protocol does not by itself make the present single-system exclusion device-independent. Kleinmann's dichotomic models~\cite{Kleinmann2014} and the density cubes of Daki\'c, Paterek and Brukner~\cite{DakicPaterekBrukner2014} exceed temporal quantum bounds; neither is Jordan-algebraic. \paragraph{Quaternionic Gram matrices and free algebras.} For tuples of points in compact quaternionic projective space, Brehm and Et-Taoui describe congruence using quaternionic Gram matrices. When no two points are orthogonal, their criterion uses the triangle congruence classes and an oriented configuration on the ordinary two-sphere~\cite{BrehmEtTaoui2001}. Roelands and Tiersma~\cite{RoelandsTiersma2026} show that the free JB-algebra on $n$ projections is special if and only if $n\le3$ (Thm.~6.6), and De Medts, Rowen and Segev~\cite{DeMedtsRowenSegev2023} bound four-generated axial algebras of Jordan type. Gogi\'c and Toma\v{s}evi\'c~\cite{GogicTomasevic2026} determine which elements of finite-dimensional simple Jordan algebras, the split Albert algebra included, are left-nested Jordan products of idempotents, and how long such words must be; the words of Section~\ref{sec:eq} are instead the elements $U_{q_1}\cdots U_{q_k}1$. \paragraph{Random states, Riesz measures and induced measures.} Proposition~\ref{prop:windowlaw} assembles published ingredients. The Riesz measures of a symmetric cone, the Wallach set and the Laplace transform $N(y)^{-s}$ are in Faraut and Kor\'anyi~\cite{FarautKoranyi1994}, Chapter~VII, as restated by Sokal~\cite{Sokal2011} (Theorem~1.1, Proposition~3.3(a): at $s=kd/2$ the Riesz measure is positive and concentrated on elements of rank $k$; Lemma~3.5 identifies it with an invariant measure on that stratum; equation~(3.7) for the Laplace transform) and by Hassairi and Lajmi~\cite{HassairiLajmi2018} (Theorem~2.4, the Gindikin theorem). Clerc~\cite{Clerc2022} restates the positivity and rank-$k$ support in his introduction. The induced measures on density matrices, with their Gaussian construction and eigenvalue densities, are due to \.Zyczkowski and Sommers~\cite{ZyczkowskiSommers2001} (Sections~III--IV). Hassairi and Lajmi project Riesz measures onto Peirce corners of lower rank, through the principal minors $\Delta_k(x)=\det_{(k)}(P_kx)$~\cite[Sec.~2]{HassairiLajmi2018}; this has the same shape as the factor $N(P_Wp_4)$ of \coreref{prop:windowdet}, but a corner has lower rank, whereas a complex subalgebra contains a frame. Results on projections of orbital measures of Duistermaat--Heckman type do not apply, since $\OP$ is not a coadjoint orbit: its second cohomology vanishes, so it carries no symplectic form. \paragraph{Completely positive realizations.} \Coreref{sec:ins} and Section~\ref{sec:c:sharp} leave class~(I) and allow arbitrary instruments; the frame of reference there is realization theory, recalled in the core paper. The dilation literature behind Section~\ref{sec:c:sharp} is compared with \coreref{prop:ins:sharpdilation} in \coreref{sec:ins:dilation}. \paragraph{Three frames.} Glennie's identity also enters for three Jordan frames, where a contraction of Glennie's identity with three frames reads a sign that the pair and triple traces leave open~\cite{MacGregorPhi2026}; four atoms never see that sign, since any four of the nine atoms of three frames lie in a quaternionic subalgebra (\coreref{lem:A:degenerate}). \clearpage \appendix \addtocontents{toc}{\protect\setcounter{tocdepth}{1}} \section{Certificate details}\label{app:cert} This appendix gives the proofs deferred from Sections~\ref{sec:cer}--\ref{sec:five}. The notation for the length-seven derivation (the vectors $\eta^X_s$, the steps $[qv\,|\,t]$ and $(q;vw\,|\,t)$, the relations (R1$_t$)--(R3$_0$) and (W3)) is that of \coreref{app:cert:seven}, whose formulas \coreref{lem:cer:gram7} are used throughout. \subsection{The finite Gram computation}\label{app:cert:gram} The proof of \eqref{eq:cer:gram} proceeds using the relations only on the vectors listed in Lemma~\ref{lem:cer:finite}. Write $\xi=P_j\Omega$, $\beta=\|\xi\|^2$, $z=N_1N_2N_3\xi$, and \[ m_1=M_{11},\quad m_2=M_{22},\quad m_3=M_{33},\quad u=M_{12},\quad v=M_{13},\quad w=M_{23}. \] The argument uses the square relations $N_a^2\eta=-M_{aa}\eta$ for $\eta\in\mathcal K_2$ and the Clifford relations $(N_aN_b+N_bN_a)\eta=-2M_{ab}\eta$ for $\eta\in\mathcal K_1$. Each $N_a$ is skew-adjoint and maps $\mathcal K_d$ into $\mathcal K_{d+1}$, and in a real Hilbert space $\ip\eta{K\eta}=0$ for skew $K$. \emph{Degree one.} $[N_a,N_b]$ is skew, so $\ip\xi{N_aN_b\xi}=\frac12\ip\xi{(N_aN_b+N_bN_a)\xi} =-M_{ab}\beta$ (for $a=b$ by the square relation). Hence $\ip{N_a\xi}{N_b\xi}=M_{ab}\beta$. \emph{The norm of $z$.} Moving $N_1$ across and using its square relation on $N_2N_3\xi$, then that of $N_2$ on $N_3\xi$, then that of $N_3$ on $\xi$: $\|z\|^2=m_1m_2m_3\,\beta$. \emph{Cross terms.} First, \[ \ip z{N_1\xi}=-\ip{N_2N_3\xi}{N_1^2\xi}=m_1\ip{N_2N_3\xi}\xi=-m_1\ip{N_3\xi}{N_2\xi}=-m_1w\beta . \] Second, rewriting $N_1N_2\xi=-N_2N_1\xi-2u\xi$ and using the square relation of $N_2$ on $N_1\xi$, \[ \ip z{N_2\xi}=-\ip{N_2N_3\xi}{N_1N_2\xi}=m_2\ip{N_3\xi}{N_1\xi}-2u\ip{N_3\xi}{N_2\xi} =(m_2v-2uw)\beta . \] Third, the Clifford relations on $\xi$ and on $N_2\xi$ give $z=N_3N_1N_2\xi+2vN_2\xi-2wN_1\xi$; pairing with $N_3\xi$, the last two terms contribute $2vw\beta-2wv\beta=0$, and the square relation of $N_3$ on $\xi$ leaves $\ip z{N_3\xi}=m_3\ip{N_1N_2\xi}\xi=-m_3u\beta$. \emph{The volume element.} Reversing $N_3N_2N_1\xi$ with the Clifford relations on vectors of degree at most one gives $\Gamma\xi=z+wN_1\xi-vN_2\xi+uN_3\xi$. With $q=(w,-v,u)$, \begin{align*} \|\Gamma\xi\|^2&=\bigl(m_1m_2m_3+2\,q\cdot(-m_1w,\;m_2v-2uw,\;-m_3u)+q^{\mathsf T}Mq\bigr)\beta\\ &=(m_1m_2m_3+2uvw-m_1w^2-m_2v^2-m_3u^2)\beta=\det M\,\beta . \end{align*} The number $\ip\xi z$, which the relations do not determine, never enters. \subsection{The robust form at length seven}\label{app:cert:sevenrobust} \paragraph{Proof of Theorem~\ref{thm:cer:sevenrobust}.} Take any sequential projective representation, its real cyclic representation $(\mathcal K,P_i,\Omega)$, and the table $g(w)=\|P_w\Omega\|^2$; nothing is assumed about $g$. Put $\zeta=P_A\Omega$, so that $P_A\zeta=\zeta$ and nothing is divided by $\Pr(A)$, and build $\tilde\Phi$ from $\zeta$ exactly as $\Phi$ (or $\Phi_{\mathsf S}^{\rm rob}$ of Appendix~\ref{app:cert:sevenrobust} for $\mathsf S$) was built from $\xi$. Follow the derivation of \coreref{lem:cer:gram7} without assuming that any word is matched. Each use of \coreref{lem:cer:short} on a vector $\eta=P_qP_t\zeta$, paired with a left vector $a$, leaves a defect: (R1$_t$) leaves $\ip a{z_q(v)P_t\zeta}$, (R2$_t$) three such terms, and (R3$_0$) seventeen with coefficients $\pm\frac12$. Every other step (all-``yes'' words and the polarisations) is exact. Two facts make the bookkeeping finite. \begin{enumerate}[label=(F\arabic*),leftmargin=2.6em] \item Every left vector is a single product $a=P_\alpha\zeta$ of projections, so $\|a\|^2=g(A,\alpha)$ is one word probability. \item Each defect vector has $\|z_q(v)P_t\zeta\|^2=X(g)$ with $X=[A,t,q,v,\check v,q]-2c[A,t,q,v]+c^2[A,t,q]$ (\coreref{lem:purity}), which vanishes on the octonionic table and has $\|X\|_1=(1+c)^2$. \end{enumerate} So, in every representation, \[ \|\tilde\Phi\|^2=\Lambda_{\rm tab}(g)+\sum_ky_k\ip{a_k}{d_k}, \] where $\Lambda_{\rm tab}$ is the combination of word probabilities that replaces each Gram entry of the derivation, $a_k=P_{\alpha_k}\zeta$, $d_k=z_{q_k}(v_k)P_{t_k}\zeta$, and the real $y_k$ are the products of the coefficients along the derivation. On the octonionic table from $e/3$, $\Lambda_{\rm tab}$ takes the value $\frac13\|\Phi\|^2$, that is $-\frac4{81}$ for $\mathsf M$ and $-\frac{205}{4374}$ for $\mathsf S$, because every entry is then $\frac13$ times its value in \coreref{lem:cer:gram7}. For any $\epsilon_k>0$, $2|\ip{a_k}{d_k}|\le\epsilon_k\|a_k\|^2+\|d_k\|^2/\epsilon_k$, so \[ \Lambda''=\Lambda_{\rm tab}+\sum_k|y_k|\Bigl(\frac{\epsilon_k}2[A,\alpha_k]+\frac1{2\epsilon_k}X_k\Bigr) \] satisfies $\Lambda''(g)\ge\|\tilde\Phi\|^2\ge0$ in every representation. Take for $\mathcal W^\ast$ in Theorem~\ref{thm:cer:sevenrobust} the set of words of $\Lambda''$ with nonzero coefficient and for the $\lambda_w$ those coefficients; every such word is one of the words listed below, so it begins with $A$ and has length at most seven. On the octonionic table every $X_k$ vanishes and $[A,\alpha_k]$ has the value $\frac13\pi_k$, with $\pi_k=\Pr_{p_A}(\alpha_k)$. Write $\|\Phi\|^2$ for $-\frac4{27}$ ($\mathsf M$) and $-\frac{205}{1458}$ ($\mathsf S$, the vector $\Phi_{\mathsf S}^{\rm rob}$). Group the defects by $\pi$ and take $\epsilon_k=\theta r_\pi$, with $r_\pi$ the largest power of $2$ not exceeding $\pi^{-1/2}$, and $\theta=3/2^{19}$ for $\mathsf M$, $3/2^{20}$ for $\mathsf S$. Let $W_\pi=\sum_{k:\pi_k=\pi}|y_k|$ and let $T$ be the sum of the absolute coefficients of $\Lambda_{\rm tab}$. Then \begin{align*} \gamma&=-\Lambda''(\Pr\nolimits_{e/3})=\tfrac13\Bigl(-\|\Phi\|^2-\tfrac\theta2\textstyle\sum_\pi W_\pi r_\pi\pi\Bigr),\\ \|\lambda''\|_1&\le T+\textstyle\sum_\pi W_\pi\Bigl(\frac{\theta r_\pi}2+\frac{(1+c_{\max})^2}{2\theta r_\pi}\Bigr), \end{align*} where $c_{\max}=\frac13$ for $\mathsf M$ and $\frac12$ for $\mathsf S$ bounds every purity constant used. The numbers $W_\pi$ and $T$ are sums of per-entry counters along the derivation, defined in Appendix~\ref{app:cert:sevenrobust}. For $\mathsf M$ they are $T=\frac{1738400}{243}$ and $W_\pi=\frac{231377}{18},\frac{17863}3,\frac{45542}3,14008,32064$ for $\pi=1,\frac13,\frac19, \frac1{27},\frac1{81}$, with $r_\pi=1,1,2,4,8$. Then $\sum_\pi W_\pi r_\pi\pi=\frac{422183}{18}$, so \[ \gamma_{\mathsf M}=\frac13\Bigl(\frac4{27}-\frac{422183}{6\cdot2^{20}}\Bigr) =\frac13\cdot\frac{8388608-3799647}{27\cdot2^{21}}=\frac{4588961}{169869312}. \] In the bound for $\|\lambda''\|_1$ the dominant term is $\frac{(4/3)^2}{2\theta}\sum_\pi W_\pi/r_\pi=\frac{2^{22}}{27}\cdot\frac{610361}{18}<5.2676\times10^{9}$; the term $T$ is below $7.2\times10^3$ and the term with $\theta r_\pi/2$ is below $2$. So $\|\lambda''\|_1\le5.268\times10^9$. The constants for $\mathsf S$ are obtained in the same way in Appendix~\ref{app:cert:sevenrobust}. \qed \paragraph{The words.} The support $\mathcal W^\ast$ of $\Lambda''$ is contained in the following set of words, for every assignment of $B,C,D$ to the letters $x,y,z$: \begin{enumerate}[label=(\roman*),leftmargin=2em] \item the word $(A)$; \item the all-``yes'' words $(A,u)$ with $|u|\le4$ and the polarisation words $(A,x,y,A)$, $(A,x,y,z,A)$, $(A,x,y,\bar z,A)$, $(A,x,y,z,x,A)$, $(A,x,y,z,\bar x,A)$; \item for every step $[qv\,|\,t]$ of the proof of \coreref{lem:cer:gram7}, the words $(A,t,q,v,q)$, $(A,t,q,v)$, $(A,t,q)$; for every step $(q;vw\,|\,t)$, the words $(A,t,q,w,q)$, $(A,t,q,\bar v,w,\bar v,q)$, $(A,t,q,v,w,v,q)$, $(A,t,q,w)$, $(A,t,q,\bar v,w)$, $(A,t,q,v,w)$, $(A,t,q)$; for (R3$_0$), the words $(A,v,\check v,A)$ and $(A,v)$ for the seventeen words $v$ of \coreref{lem:cer:short}; \item the words $(A,\alpha)$ for the left words $\alpha$ of these steps (of length at most four). \end{enumerate} Equal adjacent ``yes'' outcomes are merged ($P_q^2=P_q$), among them the leading letters $A,A$. Every word has length at most seven and begins with ``yes'' on $A$. \paragraph{Counters.} For each entry $e$ of \coreref{lem:cer:gram7} in a given row, $T_e$ is the sum of the absolute coefficients of the word probabilities that replace it in $\Lambda_{\rm tab}$, and $Y_e[\pi]$ is the sum of the absolute coefficients of its defects whose left vector $a=P_\alpha\zeta$ has $\pi=\Pr_{p_A}(\alpha)$. Each step contributes, times the absolute coefficient with which its output enters the entry: one unit for $[qv\,|\,t]$, three for $(q;vw\,|\,t)$, three for a use of (W3), $\frac{17}2$ for a vector use of (W2) (seventeen purity terms with coefficient $\frac12$), and $\frac{17}4$ for the scalar $L(xyz)=\omega_y$, which is (R3$_0$) paired with $\xi$ and halved. Entries quoted from other entries carry those entries' counters. The multiplicity $m_e$ is the sum of $|c_ic_j|$ over the pairs of coefficients of the certificate vector whose inner product uses the entry in its row; the computing row is that of the $\eta^R$ vector for the pairs RC and RH, and that of the $\eta^C$ vector for the pairs CH. Since a defect vector $z_q(v)P_t\zeta$ lies in the range of $P_q$, its left word may be taken with or without a final $q$; the totals below take it with the final $q$ in two places, $\alpha=(x)$ for $L(xyzx)$ in (W1) and $\alpha=(x,y,z,A)$ for the use of (R2$_0$) at $A$ in $\mathrm{RC}(yxz)$, and without it elsewhere. Other admissible choices change the totals. Then $T=\sum_em_eT_e$ and $W_\pi=\sum_em_eY_e[\pi]$, the sums over all rows. \paragraph{The counters for $\mathsf M$.} All rows have equal counters, because all traces of $\mathsf M$ are equal; the multiplicities $m_e$ are $486$ ($\mathrm{HH}$), $528$ ($\mathrm{CC}$), $600$ ($\mathrm{RR}$), $1080$ ($\mathrm{RH}$) and $480$ ($\mathrm{RC}$) after summing over the six rows. The per-entry counters $T_e$, $Y_e[\pi]$ are read off the derivations of \coreref{lem:cer:gram7} by the rule above; their tabulation is carried out in the accompanying formalization. Summing gives the values of $T$ and $W_\pi$ used in the proof above; for instance $W_{1/27}=528\cdot\frac83+600\cdot4+1080\cdot5+480\cdot10=14008$, where $\frac83$, $4$, $5$, $10$ are the row sums of $Y_e[\frac1{27}]$ over the entries $\mathrm{CC}$, $\mathrm{RR}$, $\mathrm{RH}$, $\mathrm{RC}$ (those of $\mathrm{HH}$ vanish). \paragraph{The configuration $\mathsf S$.} The robust form uses \[ \Phi_{\mathsf S}^{\rm rob}=\sum_s\operatorname{sgn}(s)\bigl(w_C(s)\eta^C_s+w_R(s)\eta^R_s+w_H(s)\eta^H_s\bigr), \] with $(w_C,w_R,w_H)=(3,14,-13)$ if $B$ is the first letter of $s$, $(4,15,-13)$ if it is the middle letter and $(6,17,-15)$ if it is the last. From the formulas of \coreref{lem:cer:gram7} and the formula for $\Psi$, $\Psi(BCD)=\frac{869}{972}$, $\Psi(CBD)=-\frac{22}{27}$ and $\Psi(CDB)=-\frac{109}{729}$, so $\|\Phi_{\mathsf S}^{\rm rob}\|^2=2\bigl(\frac{2607-2376-436}{2916}\bigr)=-\frac{205}{1458}$; this also excludes the representation. Here every purity constant lies in $[\frac1{27},\frac12]$, so $(1+c)^2\le\frac94$, and the buckets are $\pi=1,\frac12,\frac13,\frac16,\frac19,\frac1{18},\frac1{27}, \frac1{54},\frac1{81}$, with $r_\pi=1,1,1,2,2,4,4,4,8$. The counters of the rows $BCD$, $CBD$, $CDB$ (their images under the exchange of $C$ and $D$ have the same counters) are obtained by the same rule and are tabulated in the accompanying formalization. Summing, $T=\frac{21466321}{1458}$ and \[ (W_\pi)_\pi=\Bigl(\tfrac{3099331}{108},\ 4065,\ \tfrac{73880}9,\ \tfrac{28534}3,\ \tfrac{191914}9,\ 9674,\ \tfrac{56086}3,\ 25290,\ 29772\Bigr). \] With $\theta=3/2^{20}$, $\sum_\pi W_\pi r_\pi\pi=\frac{16559187}{324}$, so \[ \gamma_{\mathsf S}=\frac13\Bigl(\frac{205}{1458}-\frac{16559187}{108\cdot2^{21}}\Bigr) =\frac13\cdot\frac{859832320-447098049}{729\cdot2^{23}}=\frac{412734271}{18345885696}. \] The dominant term of the bound for $\|\lambda''\|_1$ is $\frac{9/4}{2\theta}\sum_\pi W_\pi/r_\pi=3\cdot2^{17}\sum_\pi W_\pi/r_\pi<3\cdot2^{17}\cdot73526 <2.8912\times10^{10}$; with $T<1.5\times10^4$ and a $\theta$-term below $1$, $\|\lambda''\|_1\le 2.892\times10^{10}$. \subsection{The reflected quadruple}\label{app:cert:refl} \paragraph{Traces.} The vector $v=(-1-2ij,\,2i-j,\,2-2ij)$ has $|v|^2=5+5+8=18$, and the entries $v_a\bar v_b$, computed in the quaternion algebra $\Ha_{ij}$ (with $(ij)i=j$, $(ij)j=-i$), give \[ 18\,U_rp_3=\begin{pmatrix}5&4i+3j&2-6ij\\-4i-3j&5&2i-6j\\2+6ij&-2i+6j&8\end{pmatrix}; \] $18\,U_rp_4$ is the same with $j,ij$ replaced by $\ell,i\ell$. Also $p_0=E_{11}$, $p_1=\frac12(E_{11}+E_{22}+F_{12}(1))$ and $p_0\jp p_1=\frac14(2E_{11}+F_{12}(1))$. Write $q_3=U_rp_3$, $q_4=U_rp_4$, and pair by $\ip XY=\sum_{a,b}\operatorname{Re}(X_{ab}Y_{ba})$. Reading entries gives $o_{AB}=o_{CD}=\frac12$, $o_{AC}=o_{AD}=o_{BC}=o_{BD}=\frac5{18}$, all four shape invariants $R=\frac5{36}$ and $t_1=\ip{p_0\jp p_1}{q_3\jp q_4}=\frac5{72}$. For instance $t_2=\ip{p_0\jp q_3}{p_1\jp q_4}$: the matrix $p_0\jp q_3$ vanishes outside its first row and column, $36\,(p_0\jp q_3)$ having first row $(10,\,4i+3j,\,2-6ij)$, and $72\,(p_1\jp q_4)_{1\cdot}=(10,\,10+8i+6\ell,\,2+2i-6\ell-6i\ell)$, so $t_2=\frac{100+2\cdot32+2\cdot4}{2592}=\frac{43}{648}$. Also $t_3=t_2$: the automorphism of $\Oct$ fixing $i$ and exchanging $j$ and $\ell$ exists because $(i,\ell,j)$ is again a basic triple, and it fixes $p_0,p_1,p_2$ and exchanges $q_3,q_4$. Hence $\omega_1=t_2+t_3-t_1=\frac{41}{648}$ and $\omega_2=\omega_3=t_1=\frac5{72}$. \paragraph{The four-point invariant.} In \coreref{eq:A:Acal52} the eight groups of terms are $\frac{625}{419904}$, $-\frac{625}{209952}$, $-\frac{625}{419904}$, $-\frac{2931}{1679616}$, $\frac{1025}{419904}$, $\frac{625}{419904}$, $\frac{625}{419904}$ and $-\frac{1025}{1679616}$. Over the denominator $1679616$ they are $2500-5000-2500-2931+4100+2500+2500-1025=144$, so $\Acal=\frac{144}{1679616}=\frac1{11664}$. For instance the fourth group is $-\bigl(\frac14\omega_1^2+2\cdot\frac{25}{324}\omega_2^2\bigr)=-\frac{1681+1250}{1679616}$. The uniform argument needs only $\Acal>0$; the explicit vector in the proof of Theorem~\ref{thm:five:refl} does not use the value. \paragraph{Assembly.} With the traces above, the formulas of \coreref{lem:cer:gram7} give the values entering $\|\Phi\|^2$ for the vector $\Phi$ of Theorem~\ref{thm:five:refl}; by the exchange of $C$ and $D$ only the rows $BCD$, $CBD$, $CDB$ and the two-step vectors $\eta^J_{CD}$, $\eta^J_{CB}$ need to be evaluated. With $\Psi$ as in \coreref{app:cert:seven}, $\Psi(BCD)=\Psi(BDC)=\frac{469}{3888}$, $\Psi(CBD)=\Psi(DBC)=\frac{5437}{23328}$ and $\Psi(CDB)=\Psi(DCB)=-\frac{1063}{3888}$, so the $\eta^C,\eta^R,\eta^H$ part of $\|\Phi\|^2$ is $2\cdot\frac{2814+5437-6378}{23328}=\frac{1873}{11664}$. For the cross terms put $S_J=\sum_\rho\bigl(c^H_\rho\ip{\eta^J}{\eta^H_\rho}+c^C_\rho\ip{\eta^J}{\eta^C_\rho} +c^R_\rho\ip{\eta^J}{\eta^R_\rho}\bigr)$; then $S_{CD}=\frac2{27}$, $S_{CB}=-\frac5{648}$, and the exchange of $C$ and $D$ gives $S_{DC}=-\frac2{27}$, $S_{DB}=\frac5{648}$. The cross terms are \[ 2\Bigl(-\tfrac{10}9\cdot\tfrac2{27}+\tfrac12\cdot\bigl(-\tfrac5{648}\bigr)+\tfrac{10}9\cdot\bigl(-\tfrac2{27}\bigr) -\tfrac12\cdot\tfrac5{648}\Bigr)=-\tfrac{80}{243}-\tfrac{20}{1296}=-\tfrac{335}{972}. \] Among the two-step vectors, $\ip{\eta^J_{CD}}{\eta^J_{CD}}=\frac5{36}$, $\ip{\eta^J_{CD}}{\eta^J_{DC}}=\frac5{72}$, $\ip{\eta^J_{CB}}{\eta^J_{CB}}=\frac{25}{324}$ and $\ip{\eta^J_{CB}}{\eta^J_{DB}}=\frac{41}{648}$, so they give $\frac{100}{81}\bigl(\frac5{36}+\frac5{36}-\frac5{36}\bigr)+\frac14\bigl(\frac{25}{324}+\frac{25}{324}\bigr) -\frac12\cdot\frac{41}{648}=\frac{125}{729}+\frac1{144}=\frac{2081}{11664}$, the four products of a coefficient $\pm\frac{10}9$ with a coefficient $\pm\frac12$ cancelling in pairs. \subsection{A constructive eleven-step separator}\label{app:cert:quantitative} This gives a finite symbolic definition of the functional in Theorem~\ref{thm:eleven}(c). The configuration is fixed throughout. All constants below depend only on its atoms, not on the tested representation or its state. In a real cyclic representation with unit vector $\Omega$ put \[ r_{j\alpha}=z_q(v)P_t\Omega,\qquad R_{j\alpha}(f)=f(t,q,v,\check v,q)-2c_q(v)f(t,q,v)+c_q(v)^2f(t,q). \] The index $\alpha$ records an occurrence of a purity relation in the ledger for base $j$. \Coreref{lem:purity} gives $R_{j\alpha}(f)=\|r_{j\alpha}\|^2$ for every sequential projective representation. On the octonionic table from every state it is zero. The following identities define finite projection-word polynomials $U_{j\alpha}$ such that \begin{equation}\label{eq:cer:residualtrace} \Dtet\|P_j\Omega\|^2 =\sum_\alpha\langle U_{j\alpha}(P)\Omega,r_{j\alpha}\rangle. \end{equation} Here $\Dtet$ is computed at the base $j$; for atoms of $\J$ its value does not depend on the base, by \coreref{prop:Dacal}. Repeated occurrences are retained; their number need not equal the $1{,}673$ distinct applications. \paragraph{The scalar telescope.} Use the notation of Appendix~\ref{app:cert:gram}. Put $a_i=N_i\xi$, $z=N_1N_2N_3\xi$, $\beta=\|\xi\|^2$, $q=(w,-v,u)$ and $d=(-m_1w,m_2v-2uw,-m_3u)$. The errors are vectors, not assumed global relations: \[ S_i(\eta)=(N_i^2+m_i)\eta,\qquad C_{ij}(\eta)=(N_iN_j+N_jN_i+2M_{ij})\eta. \] Let $E_{ij}=\langle a_i,a_j\rangle-M_{ij}\beta$ and $Z_i=\langle z,a_i\rangle-d_i\beta$. Exact skew-adjointness gives \[ E_{ii}=-\langle\xi,S_i(\xi)\rangle,\qquad E_{ij}=-\tfrac12\langle\xi,C_{ij}(\xi)\rangle\quad(i\ne j). \] Moving the $N_i$ across pairings, without discarding their defects, gives \begin{align*} Z_1={}&-m_1E_{23}-\langle N_2N_3\xi,S_1(\xi)\rangle,\\ Z_2={}&m_2E_{13}-2uE_{23}-\langle N_3\xi,S_2(N_1\xi)\rangle -\langle N_2N_3\xi,C_{12}(\xi)\rangle,\\ Z_3={}&-m_3E_{12}+2vE_{23}-2wE_{13} -\langle N_1N_2\xi,S_3(\xi)\rangle+\langle V,a_3\rangle, \end{align*} where $V=N_1C_{23}(\xi)-C_{13}(N_2\xi)$. Similarly, \begin{align*} Z_0:=\|z\|^2-m_1m_2m_3\beta ={}&-\langle N_2N_3\xi,S_1(N_2N_3\xi)\rangle\\ &-m_1\langle N_3\xi,S_2(N_3\xi)\rangle -m_1m_2\langle\xi,S_3(\xi)\rangle. \end{align*} Put $g=z+wa_1-va_2+ua_3$ and $d_0=c_4-S$. Reversing the triple product in the order $(32),(31),(21)$ yields \[ E_{\rm vol}:=\Gamma\xi-g =-\tfrac12C_{23}(N_1\xi)+\tfrac12N_2C_{13}(\xi) -\tfrac12C_{12}(N_3\xi). \] With $E_{\rm data}=(\Gamma-d_0)\xi$, the exact terminal identity is \begin{align}\label{eq:cer:scalartelescope} \Dtet\,\beta={}&\langle E_{\rm data},\Gamma\xi+d_0\xi\rangle -\langle E_{\rm vol},\Gamma\xi+g\rangle-Z_0\notag\\ &-2\sum_iq_iZ_i-\sum_{i,j}q_iq_jE_{ij}. \end{align} Indeed $m_1m_2m_3+2q\cdot d+q^{\mathsf T}Mq=\det M$, the six-term scalar expansion already used in the Gram proof. No division by $\beta$, $d_0$ or $\det M$ occurs. \paragraph{Reducing the vector errors.} On the $j$ corner define the operator defects \begin{align*} Z_{ab}&=P_aP_bP_a-o_{ab}P_a,\\ B_{a;bc}&=P_a(P_bP_c+P_cP_b)P_a-2R_{abc}P_a,\\ D_{j;klm}&=P_j(P_kP_lP_m+P_mP_lP_k)P_j-2Q_{j;klm}P_j. \end{align*} For $h=P_jP_kP_lP_j/\nu$, $\nu^2=o_{jk}o_{kl}o_{jl}$, direct subtraction gives \[ \nu^2(h^*h-P_j)=P_jP_lZ_{kj}P_lP_j+o_{jk}P_jZ_{lk}P_j+o_{jk}o_{kl}Z_{jl}. \] Interchanging $k,l$ gives $hh^*-P_j$. Also $H_1-c_1P_j=B_{j;kl}/(2\nu_1)$, with cyclic variants, and \[ S_i(\eta)=(H_i+c_i)(H_i-c_i)\eta -\tfrac12\bigl[(h_ih_i^*-P_j)+(h_i^*h_i-P_j)\bigr]\eta. \] For $h_1=P_jP_kP_lP_j/\nu_1$, $h_2=P_jP_lP_mP_j/\nu_2$, put $T_{12}=(h_1h_2+h_2^*h_1^*)/2-K_{12}P_j$. Then \[ 2\nu_1\nu_2T_{12} =P_jP_kZ_{lj}P_mP_j+P_jP_mZ_{lj}P_kP_j+o_{jl}D_{j;klm}. \] The scalar cancellation is $K_{12}=o_{jl}Q_{j;klm}/(\nu_1\nu_2)$. Put $A_i=H_i-c_iP_j$ and $b_i=c_iP_j+N_i$. For each cyclic pair, \[ C_{ij}(\eta)=2T_{ij}\eta -(A_ih_j+b_iA_j+A_jh_i^*+b_j^*A_i)\eta. \] Thus a Clifford defect on degree at most one uses only $H$ defects on degree at most two and the displayed contraction and R3 defects. For the fourth face set $n=\nu_1\nu_2\nu_3$ and $F=(h_1h_2h_3+(h_1h_2h_3)^*)/2-c_4P_j$. The exact identity is \begin{align*} 2nF={}&P_jP_kP_lP_jP_lZ_{mj}P_kP_j+o_{jm}P_jP_kZ_{lj}P_mP_kP_j\\ &+P_jP_kP_mP_jP_mZ_{lj}P_kP_j+o_{jl}P_jP_kZ_{mj}P_lP_kP_j\\ &+o_{jm}o_{jl}P_jB_{k;lm}P_j+2o_{jm}o_{jl}R_{klm}Z_{jk}. \end{align*} To obtain these right histories, contract the rightmost $P_mP_jP_m$ first in $h_1h_2h_3$, then $P_lP_jP_l$; in its adjoint contract the rightmost $P_lP_jP_l$ first, then $P_mP_jP_m$. The scalar equality is $nc_4=o_{jm}o_{jl}o_{jk}R_{klm}$. Here $B_{k;lm}=B_{k;lm}P_k$, so $B_{k;lm}P_j\xi=B_{k;lm}P_kP_j\xi$. The R1 right histories are $P_kP_j\xi,P_mP_kP_j\xi,P_kP_j\xi,P_lP_kP_j\xi$; the R2 history is $P_kP_j\xi$, and the last R1 acts on $\xi$. They are precisely row (iv) of the ledger. The remaining $H$ telescope is \begin{align*} E_H=\tfrac12[&A_1h_2h_3+b_1A_2h_3+b_1b_2A_3\\ &+A_3h_2^*h_1^*+b_3^*A_2h_1^*+b_3^*b_2^*A_1]\xi,\\ E_{\rm data}={}&F\xi-E_H -\tfrac12\bigl[c_1C_{23}(\xi)+c_2C_{13}(\xi)+c_3C_{12}(\xi)\bigr]. \end{align*} Every $A_i$ here acts on degree at most two. \paragraph{The primitive leaves and their support.} Write $z_a(v)=P_aE_vP_a-c_a(v)P_a$, with chronological words $v$. The exact access identities give \[ B_{a;bc}=z_a(c)+z_a(b,c)-z_a(\mathrm{no}\,b,c). \] For R3 let $\sigma,\tau\in\{\mathrm{skip},\mathrm{yes},\mathrm{no}\}$, with respective signs $+1,+1,-1$, and remove every skip from a word. Then \begin{align*} D_{j;klm}=\tfrac12\sum_{\sigma,\tau}\epsilon_\sigma\epsilon_\tau [&z_j(\sigma_k,\tau_l,m)+z_j(\sigma_m,\tau_k,l)\\ &-z_j(\sigma_l,\tau_k,m)]. \end{align*} This follows by inserting $L_s=(I+U_{P_s}-U_{I-P_s})/2$ into $KLM'+M'LK=2[L_kL_l(M')+L_mL_k(L)-L_lL_k(M')]$. The fixed scalar terms cancel by the same Jordan access identity. R1 is already $Z_{ab}=z_a(b)$. These formulas specify a finite recursion for \eqref{eq:cer:residualtrace}: substitute the vector identities into \eqref{eq:cer:scalartelescope}, expand $H,N$ and holonomies in their written order, insert the displayed primitive formulas, and move left coefficients across each real inner product by the adjoint. Keep separate occurrences if desired. Every right history is one in the original ledger: $H$ errors of degree two and unitarity extensions cost eleven letters, Clifford errors of degree one cost ten, and the fourth face costs six. Longer left coefficient words do not introduce new measured histories. This is an exact symbolic definition, not an inference of a residual identity from a zero-residual implication. \paragraph{A uniform margin.} Expand each $U_{j\alpha}$ as a finite sum of projection words and let $B_{j\alpha}$ be the sum of absolute values of its coefficients. Then $\|U_{j\alpha}(P)\Omega\|\le B_{j\alpha}$ for every normalized model. Put $K_0=\sum_{j,\alpha}B_{j\alpha}^2$, over actual occurrences. Let $y=\sum_jp_j$ and \[ \gamma=\frac{\Dtet}{8}\det y>0. \] The determinant is the rank-three Jordan determinant with $\det e=1$. The positive eigenvalues of $y$ sum to four; the product of the largest two is at most four. Hence $y\ge(\det y/4)e$ and $\Dtet\,y\ge2\gamma e$. Set $a=(K_0+1)/(4\gamma)>0$ and define \[ \Lambda(f)=a\sum_{j,\alpha}R_{j\alpha}(f) -\Dtet\sum_jf(j)+\gamma f(\varnothing). \] By \eqref{eq:cer:residualtrace}, Cauchy--Schwarz and Young's inequality, \[ \Dtet\sum_jf(j)\le\sqrt{K_0\sum_{j,\alpha}R_{j\alpha}(f)} \le a\sum_{j,\alpha}R_{j\alpha}(f)+\gamma \] on every sequential projective representation; the Young remainder is $K_0/(4a)=\gamma K_0/(K_0+1)\le\gamma$, including $K_0=0$. Thus $\Lambda\ge0$ there, whereas $\Lambda(\Pr_\rho)\le-\gamma$ for every state $\rho$ of $\J$. Realification gives the stated comparison class. No conditioning on a positive preparation probability is used. After merging coincident words a conservative bound is \[ \|\Lambda\|_1\le a\sum_{j,\alpha}(1+2c_q(v)+c_q(v)^2) +4\Dtet+\gamma. \] This gives a symbolic quantitative ratio on the prescribed $\mathcal F\cup\{\varnothing\}$. It need not be well-conditioned as transition probabilities approach zero, and gives neither an optimal ratio nor the constants of the separate fixed seven-step or five-test inequalities. \subsection{The eleven-step list}\label{app:cert:count} Write $j=A$ for the base, $(k,l,m)=(B,C,D)$ and $X=\{k,l,m\}$, and put $E=\{(k,l),(l,m),(m,k)\}$ and $E^{\rm op}=\{(l,k),(m,l),(k,m)\}$. Words are chronological; a letter $x$ is the outcome ``yes'' on $x$ and $\bar x$ the outcome ``no''. In the tails below $x,y,z$ are distinct letters of $X$, subject only to the stated restriction; $\sigma x$ is $x$ or $\bar x$, and $\tau y$ is $y$ or $\bar y$, with the same choice at each occurrence. The holonomy prefixes are $H_d=\{jx_1y_1j\cdots x_dy_dj:x_i,y_i\in X,\ x_i\ne y_i\}$, of cardinality $6^d$ and length $1+3d$. The list of Theorem~\ref{thm:eleven} is \begin{equation}\label{eq:cer:elevenlist} \mathcal F_A=\{Tb:T\in H_0\cup H_1\cup H_2,\ b\in\mathcal B\}\cup \{Td:T\in H_0\cup H_1,\ d\in\mathcal D\}\cup\mathcal F_{\rm far}, \end{equation} with the forty tails $\mathcal B$ and the fifty-seven tails $\mathcal D$ of Table~\ref{tab:cer:tails} and the six far-face words \[ \mathcal F_{\rm far}=\{jkmlj,\ jklmj,\ jklmlk,\ jk\bar lm\bar lk,\ jkmljl,\ jklmjm\}. \] \begin{table}[htbp]\centering\footnotesize \caption{The tails of the eleven-step list.}\label{tab:cer:tails} \begin{tabular}{@{}c c l r@{}}\toprule &length&tails&number\\\midrule $\mathcal B$&$0$&$\varnothing$&$1$\\ &$1$&$x$&$3$\\ &$2$&$xj$;\ $xy$;\ $\bar xy$ with $(x,y)\in E$&$12$\\ &$3$&$xyx$;\ $xyj$&$12$\\ &$4$&$xyjy$;\ $xyxj$ and $\bar xy\bar xj$ with $(x,y)\in E$&$12$\\\midrule $\mathcal D$&$2$&$\bar xy$ with $(x,y)\in E^{\rm op}$&$3$\\ &$3$&$\sigma x\,\tau y\,z$&$24$\\ &$4$&$\sigma x\,y\,\sigma x\,j$ with $(x,y)\in E^{\rm op}$&$6$\\ &$6$&$\sigma x\,\tau y\,z\,\tau y\,\sigma x\,j$&$24$\\\bottomrule \end{tabular} \end{table} These are the words that the relation applications of Table~\ref{tab:cer:ledger} require. After a prefix $T$, the three unitarity contractions for an ordered pair $a\ne b$ of $X$ need the tails $ba,baj,baja$; $b,ba,bab$; and $\varnothing,b,bj$, and the scalar-part relation for $(x,y)\in E$ needs $y,yj,xy,xyxj,\bar xy,\bar xy\bar xj$ and $\varnothing$; together these give $\mathcal B$. The anticommutator tests after $T\in H_0\cup H_1$ apply purity with the suffixes $V=x$, $\sigma x\,y$ and $\sigma x\,\tau y\,z$ and need $T$, $TV$ and $TV\check Vj$, where $\check V$ is $V$ without its last letter, reversed; the tails not already in $\mathcal B$ form $\mathcal D$. The far-face tests (row (iv)) give $\mathcal F_{\rm far}$. At a fixed prefix length the words $Tb$ are distinct. Between successive prefix lengths the only collisions are the tails $xyj$ and $xyjy$, which are the empty tail and the tail $y$ after the longer prefix $Txyj$, six of each kind per shorter prefix. So the first family has $40(1+6+36)-12(1+6)=1{,}636$ words. The tails of $\mathcal D$ contain no internal $j$, which makes all $57\cdot7=399$ words $Td$ new, and the six far-face words are new as well. Hence $|\mathcal F_A|=1{,}636+399+6=2{,}041$, with length polynomial \[ z+3z^2+15z^3+36z^4+32z^5+94z^6+240z^7+180z^8+432z^9+576z^{10}+432z^{11}. \] The four relabelled lists are disjoint because their first letters differ, so their union has $4\cdot2{,}041=8{,}164$ words. Separately, the relation applications number $43\cdot27+7\cdot3\cdot24+8=1{,}673$. \subsection{Equalities through five steps}\label{app:cert:h5} This subsection proves Proposition~\ref{prop:eq:five}, which is \coreref{rem:cer:fivestep}. By parts (a) and (c) of \corename{prop:cer:noequality} and the proof of (c), the assertion says: if $c$ is supported on words of length at most five and $E(c)=0$, then $f(c)(p)=0$ in $\J$. Here $c$ has real coefficients. (A nonzero element of $\J$ has nonzero pairing with some state, since the states span $\J$; so the conclusion says that no state of $\J$ violates the equality. Words need not be reduced: by part~(b) of that proposition an unreduced word has the same $E_w$ and $f(w)$ as a reduced word, or both vanish.) The argument compares two evaluations of the same formal expressions, one in the group algebra and one in $\J$, and shows that the first has the smaller kernel. On the way it uses auxiliary Jordan polynomials that are not themselves measured words; what is controlled is their degree, not a measurement length. In outline: words of length at most five become auxiliary polynomials of degree at most nine. Those of degree nine are palindromes whose coefficients can be read off one by one. Those of degree eight are products of two reflections, which cancel class by class or contract to lower degree (Lemma~\ref{lem:cer:h5:refl}). Through degree seven, a component using all four letters uses one of them only once, and the complex subalgebra of the other three carries the relation from the group algebra to $\J$ (Lemmas~\ref{lem:cer:h5:lift} and~\ref{lem:cer:h5:seven}). At degree eight the balanced multidegree $(2,2,2,2)$ uses no letter once, and that is where the argument stops. \paragraph{Two evaluations.} Let $\mathcal P$ be the free unital Jordan algebra over $\R$ on four letters $x_1,\dots,x_4$; its elements are \emph{Jordan polynomials}, with $U_xy=2x\jp(x\jp y)-(x\jp x)\jp y$. It is graded by \emph{multidegree}, the number of occurrences of each letter, and the \emph{degree} of a polynomial is the largest total degree of its homogeneous components. A polynomial has \emph{degree one in $x_j$} if every component has exactly one occurrence of $x_j$. Let $G=\mathbb Z_2^{*4}$ with generators $r_1,\dots,r_4$, as in \coreref{sec:cer:frame}, and let $s_i=2p_i-e\in\J$, so $s_i\jp s_i=e$. The \emph{special value} $\hat\varpi:\mathcal P\to\R[G]^+$ and the \emph{Albert value} $\hat\nu:\mathcal P\to\J$ are the unital Jordan homomorphisms with $x_i\mapsto r_i$ and $x_i\mapsto s_i$; here $\R[G]^+$ is the whole group algebra with the product $X\jp Y=\frac12(XY+YX)$. All scalars in this subsection are real. The special value factors through the free associative algebra $\R\langle x_1,\dots,x_4\rangle$: a polynomial first has its \emph{associative expansion} there, then each word is reduced by $r_i^2=1$. A component of multidegree $\mu$ expands into words of multidegree $\mu$, and reduction deletes letters in equal pairs, so \begin{equation}\label{eq:cer:h5:filter} \hat\varpi(F_\mu)\ \text{is a combination of reduced words of multidegree }\mu-2\kappa,\ \kappa\in\mathbb N^4 . \end{equation} The special value is fixed by the reversal $u\mapsto u^{\mathrm{rev}}$, as every Jordan product of reversal-fixed elements is. \paragraph{From words to symbols.} For a symmetry $s$ ($s\jp s=e$) of a unital Jordan algebra put $A_s=2L_s$ and $C_s=U_s$. Polarizing $U$ at the unit gives $U_{(e\pm s)/2}=\frac14(\id\pm A_s+C_s)$. On the Peirce spaces $V_1,V_{1/2},V_0$ of the idempotent $(e+s)/2$ (\coreref{sec:set:eja}) the operator $L_s$ acts by $1,0,-1$ and $U_s$ by $1,-1,1$, so \begin{equation}\label{eq:cer:h5:canon} A_s^2=2(\id+C_s),\qquad A_sC_s=C_sA_s=A_s,\qquad C_s^2=\id,\qquad C_se=e,\qquad A_se=2s . \end{equation} In $\R[G]^+$ these are $A_rX=rX+Xr$ and $C_rX=rXr$, and \eqref{eq:cer:h5:canon} can be checked directly. A \emph{symbol} is an expression $X_1\cdots X_m1$ with each $X_t$ one of $A_{x_i},C_{x_i}$, adjacent operators carrying different letters, and $X_m$ an $A$ unless $m=0$. Read with $A_x=2L_x$, $C_x=U_x$, a symbol with $a$ operators $A$ and $b$ operators $C$ is a homogeneous Jordan polynomial of degree $a+2b$. Let $\kappa:\mathcal P\to FJ_4$ be the unital homomorphism with $x_i\mapsto2p_i-1$. Then $\varpi\circ\kappa=\hat\varpi$, and evaluation at $p$ composed with $\kappa$ is $\hat\nu$. Since $p_i=\frac12(1+(2p_i-1))$ and $1-p_i=\frac12(1-(2p_i-1))$, expanding every $U_{q_t}$ in $f(w)=U_{q_1}\cdots U_{q_k}1$ by the first formula writes $f(c)$, for $c$ supported on words of length at most five, as $\kappa$ of a combination of operator words of length at most five in the $A_{x_i},C_{x_i}$, applied to $1$. Both evaluations send $x_i$ to a symmetry, so \eqref{eq:cer:h5:canon} may be used to shorten these words without changing either value; each use removes an operator. The result is a real combination $F_c$ of symbols with at most five operators, with $\hat\varpi(F_c)=\varpi(f(c))=E(c)$ and $\hat\nu(F_c)=f(c)(p)$. It therefore suffices to prove: \begin{equation}\label{eq:cer:h5:goal} \parbox{0.85\linewidth}{if $F$ is a combination of symbols with at most five operators and $\hat\varpi(F)=0$, then $\hat\nu(F)=0$.} \end{equation} A symbol with $a\ge1$ operators $A$ and $b$ operators $C$, $a+b\le5$, has degree $a+2b\le10-a$. So degree nine occurs only for $(a,b)=(1,4)$, degree eight only for $(a,b)=(2,3)$, and every other symbol has degree at most seven. \paragraph{Reflections.} Each $U_{s_i}$ is an involutive Jordan automorphism of $\J$ (\coreref{sec:set:eja}), so $r_i\mapsto U_{s_i}$ defines an action $\rho$ of $G$ on $\J$. A \emph{reflection} of $G$ is a conjugate $t=vr_iv^{-1}$ of a generator. Its reduced word is $vr_iv^{-1}$ with $v$ reduced and not ending in $r_i$; this \emph{canonical form} is unique, and $|t|=2|v|+1$. Put \[ \varsigma(t)=\rho(v)s_i . \] This does not depend on $v$: the generator $r_i$ is not conjugate to another generator (their images in $G/[G,G]=\mathbb Z_2^4$ differ), its centralizer is $\{1,r_i\}$, and $U_{s_i}s_i=s_i$. Since $\rho(v)$ is an automorphism, $\varsigma(t)=2q-e$ for the atom $q=\rho(v)p_i$, and \begin{equation}\label{eq:cer:h5:cov} \varsigma(utu^{-1})=\rho(u)\varsigma(t),\qquad \rho(t)=U_{\varsigma(t)} . \end{equation} If $v=r_{v_1}\cdots r_{v_k}$, the \emph{reflection polynomial} $R_t=U_{x_{v_1}}\cdots U_{x_{v_k}}x_i$ has degree $|t|$, special value $t$ and Albert value $\varsigma(t)$. Pushing the operators of a symbol to the outside with \eqref{eq:cer:h5:cov} gives \begin{equation}\label{eq:cer:h5:twoA} C_uA_x1\ \longmapsto\ 2\,t,\ 2\,\varsigma(t);\qquad C_uA_xC_vA_y1\ \longmapsto\ 4\,[tt'],\ 4\,\varsigma(t)\jp\varsigma(t'), \end{equation} special and Albert values respectively, where $C_u$ stands for a product of operators $C$, $t=uxu^{-1}$, $t'=uvyv^{-1}u^{-1}$ and $[g]=\frac12(g+g^{-1})$. In particular a symbol with two operators $A$ has as special value a single reversal class, with no lower-order terms. \provisionalref{Reflection products}{lem:cer:h5:refl}{Same statement}{(a) \leandecl{HorizonFive.reflectionClass}\allowbreak\leandecl{Consistency}; (b) \leandecl{HorizonFive.reflectionContraction}.} \begin{lemma}[Products of two reflections]\label{lem:cer:h5:refl} Let $t,u,t',u'$ be reflections of $G$. \begin{enumerate}[label=(\alph*),leftmargin=2em] \item If $t'u'=tu$ or $t'u'=(tu)^{-1}$, then $\varsigma(t)\jp\varsigma(u)=\varsigma(t')\jp\varsigma(u')$. \item There is a Jordan polynomial $K$ of degree equal to the reduced length $|tu|$ with $\hat\varpi(K)=[tu]$ and $\hat\nu(K)=\varsigma(t)\jp\varsigma(u)$. \end{enumerate} \end{lemma} \begin{proof} (a) If $tu=1$ then $t=u$, $t'=u'$, and both sides are $e$. Otherwise use the Cayley graph $\Gamma$ of $G$: its vertices are the elements of $G$ and its edges the pairs $\{g,gr_i\}$. It is a tree, because reduced words are unique, and $G$ acts on it by left multiplication, freely on vertices. The reflection $vr_iv^{-1}$ exchanges the two ends of the edge $\{v,vr_i\}$. Conversely, an involution $\beta\ne1$ of $G$ is a reflection. For a vertex $g$, $\beta$ maps the path from $g$ to $\beta g$ onto itself with the ends exchanged, so it fixes the midpoint of that path. Since $\beta$ fixes no vertex, the path has odd length and its midpoint is the midpoint of an edge $\{g',g'r_i\}$; $\beta$ exchanges the ends of that edge, so $\beta g'=g'r_i$ and $\beta=g'r_ig'^{-1}$. For $k\in G$ let $\operatorname{Min}(k)$ be the set of vertices $P$ at which $d(P,kP)$ is smallest. Stripping matching outer letters from the reduced word of $h=tu\ne1$, which has even length, gives $h=wcw^{-1}$, reduced as written, with $|c|\ge2$ even and the first and last letters of $c$ different. The vertices $wc^nc'$ ($n\in\mathbb Z$, $c'$ a prefix of $c$) form a line $\mathcal X$ that $h$ translates by $|c|$. For a vertex $P$ with nearest vertex $P'$ on $\mathcal X$, the vertex of $\mathcal X$ nearest to $hP$ is $hP'\ne P'$. The paths from $P$ to $P'$ and from $hP'$ to $hP$ meet $\mathcal X$ only at their ends, so with the segment of $\mathcal X$ between them they form a path without backtracking, the geodesic from $P$ to $hP$; hence $d(P,hP)=2d(P,\mathcal X)+|c|$. Hence $\operatorname{Min}(h)=\mathcal X$, the \emph{axis} of $h$~\cite[Ch.~I, \S6.4]{Serre1980}. Since $d(gP,ghg^{-1}\cdot gP)=d(P,hP)$ and $d(P,h^{-1}P)=d(hP,P)$, one has $\operatorname{Min}(ghg^{-1})=g\operatorname{Min}(h)$ and $\operatorname{Min}(h^{-1})=\operatorname{Min}(h)$. Each of $t,u,t',u'$ conjugates $h$ to $h^{-1}$ (for instance $tht=ut=h^{-1}$, and $t'ht'=u't'=h^{-1}$ when $t'u'=h$), so each maps $\mathcal X$ onto itself. Let $D$ be the stabilizer of $\mathcal X$. It acts on the line faithfully, since an element fixing the line pointwise fixes a vertex. So $D$ is a group of automorphisms of a line that contains a nonzero translation and the reversal $t$; its translations are generated by one of them, $\tau$, and $D$ is generated by $\alpha=t$ and $\beta=\tau t$. Both are involutions, hence reflections, and $\alpha\beta=\tau^{-1}$ has infinite order, so $D$ is the free product of $\langle\alpha\rangle$ and $\langle\beta\rangle$. By \coreref{prop:three}, applied to the atoms $(e+\varsigma(\alpha))/2$ and $(e+\varsigma(\beta))/2$ with one of them repeated, both lie in a complex subalgebra $W$. Let $\psi:W\to\HC$ be a unital Jordan isomorphism. Since $U_ab=2a\jp(a\jp b)-(a\jp a)\jp b$ and $U_XY=XYX$ in $\HC$, $\psi(U_ab)=\psi(a)\psi(b)\psi(a)$; and $\psi\varsigma(\alpha)$, $\psi\varsigma(\beta)$ are Hermitian matrices of square one. Let $\pi:D\to GL_3(\C)$ be the homomorphism with $\alpha\mapsto \psi\varsigma(\alpha)$, $\beta\mapsto\psi\varsigma(\beta)$. For $d\in D$, $\rho(d)$ maps $W$ to itself and $\psi\rho(d)\psi^{-1}$ is conjugation by $\pi(d)$: by \eqref{eq:cer:h5:cov} this holds for $\alpha$ and $\beta$, and both sides are multiplicative. Every involution of $D$ is $d\alpha d^{-1}$ or $d\beta d^{-1}$ with $d\in D$: the elements of $D$ are $\tau^k$ and $\tau^kt$, the $\tau^k$ with $k\ne0$ have infinite order, and $\tau^{2m}t=\tau^m\alpha\tau^{-m}$, $\tau^{2m+1}t=\tau^m\beta\tau^{-m}$ because $t\tau t=\tau^{-1}$. So \eqref{eq:cer:h5:cov} gives $\psi\varsigma=\pi$ on the involutions of $D$, among them $t,u,t',u'$. Hence \[ \psi\bigl(\varsigma(t)\jp\varsigma(u)\bigr)=\tfrac12\bigl(\pi(t)\pi(u)+\pi(u)\pi(t)\bigr) =\tfrac12\bigl(\pi(h)+\pi(h)^{-1}\bigr), \] which depends only on the pair $\{h,h^{-1}\}$. The same computation applies to $t',u'$. (b) If $t=u$ take $K=1$. Otherwise $h=tu$ has the axis $\mathcal X$ of part (a). Let $q$ be the vertex of $\mathcal X$ nearest to the vertex $1$, and give $\mathcal X$ integer coordinates with $q$ at $0$, oriented so that $h$ translates by $2m$ with $m\ge1$ ($h$ preserves the parity of vertices). The path from $1$ to $h$ runs through $q$ and $hq$, so $|tu|=2|q|+2m$. On $\mathcal X$ the reflection $t$ reverses direction about a half-integer $a$, since it fixes no vertex. For an integer $n$ the involution $s=h^nt$ reverses $\mathcal X$ about $a+nm$; choose $n$ with $00$, so $\mu(\{0\})=0$. Hence $\mu$-almost every $x$ is positive of rank one with $\tr x>0$, and $x/\tr x$ is an atom. Its law is an $\Ffour$-invariant probability measure on $\OP$, hence the uniform one. \emph{Projection.} $P_W$ preserves traces because $e\in W$; as in the proof of \coreref{lem:A:restrict}, $\ip xy=\Tr(\psi(x)\psi(y))$ on $W$, and $N|_W=\det\circ\psi$, since a frame of $W$ is a Jordan frame of $\J$. For $T\in W$ with $I+\psi(T)>0$, $e+T\in\Omega$ and $\ip{P_Wx}T=\ip xT$, so \[ \int e^{-\Tr(\psi(T)\,\psi(P_Wx))}\,\mu(dx)=N(e+T)^{-4}=\det\bigl(I+\psi(T)\bigr)^{-4}. \] \emph{The Gaussian side.} For $S\in H_3(\C)$ with $I+S>0$ and a standard complex Gaussian column $c\in\C^3$, diagonalising $S$ gives $\mathbb E\,e^{-c^\dagger Sc}=\det(I+S)^{-1}$; the four columns of $Z$ are independent, so $\mathbb E\,e^{-\Tr(SZZ^\dagger)}=\det(I+S)^{-4}$. \emph{Conclusion.} The random matrices $\psi(P_Wx)$ ($x\sim\mu$) and $ZZ^\dagger$ have the same Laplace transform on the open set $\{S:I+S>0\}$, hence the same law. Since $P_W$ preserves traces, $\psi(P_W(x/\tr x))=\psi(P_Wx)/\Tr\psi(P_Wx)$, the same function of $\psi(P_Wx)$ that takes $ZZ^\dagger$ to $ZZ^\dagger/\Tr(ZZ^\dagger)$. For $3\times k$ Gaussian $Z$, $ZZ^\dagger$ has rank $\min(3,k)$ almost surely; for $k\ge3$ the eigenvalue density of $\nu_{3,k}$ is proportional to $\prod_a\lambda_a^{k-3}\prod_{a