\documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage{lmodern,amsmath,amssymb,amsthm,mathtools} \usepackage[margin=1in]{geometry} \usepackage{microtype} \usepackage[hidelinks]{hyperref} \usepackage{xurl} \usepackage{enumitem} \usepackage{needspace} \hypersetup{pdftitle={The Osserman (1990) conjecture is fully resolved, and other miscellanea}, pdfauthor={Dave Goel},pdfsubject={Version 3.60}, pdfkeywords={Osserman conjecture, Jacobi operator, algebraic curvature tensor, Clifford structure, Cayley plane, rank-one symmetric space, isoparametric hypersurface}} \newcommand{\R}{\mathbb R} \newcommand{\OCT}{\mathbb O} \newcommand{\Cl}{\operatorname{Cl}} \newcommand{\tr}{\operatorname{tr}} \newcommand{\Span}{\operatorname{span}} \newcommand{\Id}{I} \newcommand{\ip}[2]{\langle #1,#2\rangle} \newcommand{\norm}[1]{\lVert #1\rVert} \newcommand{\Jac}{\mathcal J} \newcommand{\rank}{\operatorname{rank}} \newtheorem{mainthm}{Theorem} \renewcommand{\themainthm}{\Alph{mainthm}} \newtheorem{maincor}[mainthm]{Corollary} \newtheorem{lemma}{Lemma}[section] \newtheorem{proposition}[lemma]{Proposition} \theoremstyle{remark} \newtheorem{remark}[lemma]{Remark} \numberwithin{equation}{section} \newcommand{\classcor}{Corollary~\ref{cor:C}} \newcommand{\classAtheorem}{Theorem~\ref{thm:A}} \title{The Osserman (1990) conjecture is fully resolved, and other miscellanea} \author{Dave Goel\\[2pt] {\small \href{mailto:david44hello@gmail.com}{david44hello@gmail.com}}} \date{October 28, 2026\\[3pt] {\small Manuscript version 3.60, 2026-10-28.}} \begin{document} \maketitle \begin{abstract} The Osserman conjecture says that a Riemannian manifold whose Jacobi operators have the same eigenvalues in every unit direction is flat or locally rank-one symmetric. Chi and Nikolayevsky proved it in every dimension except one remaining case: dimension sixteen, when a Jacobi eigenvalue has multiplicity seven or eight. Nikolayevsky reduced this remaining case to a purely algebraic statement, the Nikolayevsky (2009) conjecture, his Conjecture~A. We prove the Nikolayevsky (2009) conjecture. Hence the Osserman conjecture is true in every dimension, including its pointwise form in dimension sixteen. The same result completes the conformal Osserman conjecture, which is stated for dimensions greater than four. Conjecture~A says that an Osserman algebraic curvature tensor on $\R^{16}$ whose reduced Jacobi operator has an eigenvalue of multiplicity seven or eight has a Clifford structure or is a linear combination of the constant-curvature tensor and an orthogonal pullback of the Cayley-plane curvature tensor. For the two-eigenvalue case, we construct a Cartan--M\"unzner quartic and recover the original Jacobi eigenspaces from its isoparametric classification. For the remaining multiplicity patterns, we study a rational coherent factor of the Jacobi family. Its transpose pencil, polar divisors and residue images reduce the possible poles to cases excluded by degree bounds, real positivity and homogeneous factorization. A frame-independent closed-kernel identity completes the passage from local pole reductions to the algebraic classification. \medskip\noindent \textit{2020 Mathematics Subject Classification.} Primary 53C25; Secondary 53B20, 53C35, 53C40. \noindent \textit{Key words and phrases.} Osserman conjecture, Jacobi operator, algebraic curvature tensor, Clifford structure, Cayley plane, rank-one symmetric space, isoparametric hypersurface. \end{abstract} \section{Introduction and statements} \paragraph{The remaining case.} The Osserman conjecture, posed by Osserman in 1990 \cite{Oss1990}, says that a Riemannian manifold whose Jacobi operators have the same eigenvalues in every unit direction is flat or locally rank-one symmetric. Chi proved it in dimensions not divisible by four and in dimension four \cite{Chi1988}. Nikolayevsky proved it, also in the pointwise form for $n\ne2,4$, in every other dimension except one remaining case: dimension sixteen, when a reduced Jacobi eigenvalue has multiplicity seven or eight \cite{Nik2004,Nik2005,Nik2006}, \cite[Corollary~1.1]{Nik2006}. He then reduced this remaining case to a purely algebraic statement, the Nikolayevsky (2009) conjecture, his Conjecture~A \cite[Theorem~1 and p.~679]{Nik2012}. For surveys of Osserman manifolds see \cite{GKV2002,Gilkey2001}. We prove the Nikolayevsky (2009) conjecture (Theorem~\ref{thm:conjA}). Hence the Osserman conjecture is true in every dimension (Corollary~\ref{cor:osserman16} and Remark~\ref{rem:osserman-complete}). The same result completes the conformal Osserman conjecture, which is stated for dimensions greater than four \cite{Nik2010}. \paragraph{Main results.} The main new results of this paper are the following. \begin{itemize}[leftmargin=1.5em] \item Theorem~\ref{thm:conjA}: every Osserman algebraic curvature tensor on $\R^{16}$ with a reduced Jacobi eigenvalue of multiplicity seven or eight is Clifford or of Cayley type. \emph{This proves the Nikolayevsky (2009) conjecture (his Conjecture~A), the algebraic statement to which he reduced the Osserman conjectures in dimension sixteen \cite{Nik2012}.} \item Corollary~\ref{cor:osserman16}: every pointwise Osserman $16$-manifold is flat or locally rank-one symmetric, and the conformal analogue holds. \emph{Dimension sixteen was the only open case \cite[Corollary~1.1]{Nik2006}; this completes the Osserman conjecture \cite{Oss1990} in every dimension and, with \cite{Nik2010}, the conformal Osserman conjecture, which is stated for dimensions $n>4$.} \item Theorem~\ref{thm:A} and Corollary~\ref{cor:B}: the two-eigenvalue case with multiplicities $7$ and $8$ is $\Cl(7)$, $\Cl(8)$ or Cayley, and the corresponding manifolds are locally the Cayley projective plane or its dual. \emph{This proves the theorem announced by Nikolayevsky in 2006 \cite[Theorem~1.2]{Nik2006}, whose proof did not appear \cite[Section~4]{And2022}.} \item Corollary~\ref{cor:C}: Jacobi-proportional Osserman tensors on $\R^{16}$ are Clifford or of Cayley type. \emph{This settles the two patterns $(8,7)$ and $(7,7,1)$ to which Andreji\'c reduced this question \cite[Theorem~5]{And2022}.} \item Theorem~\ref{thm:D}: four three-eigenvalue patterns are Clifford. \emph{This extends Nikolayevsky's polynomial eigenfield method \cite[Lemmas~3.1--3.2]{Nik2006} to multiplicities seven and eight.} \end{itemize} \paragraph{Notation.} Let $V$ be a Euclidean vector space. For an algebraic curvature tensor $R$ we use the Jacobi operator \[ \Jac_x y=R(y,x)x. \] The tensor is \emph{Osserman} if the spectrum of $\Jac_x$ on $x^\perp$ is independent of the unit vector $x$. All multiplicities below refer to this reduced operator. This is also Nikolayevsky's convention \cite[p.~379]{Nik2006}, \cite[proof of Lemma~4]{Nik2012}, so Theorem~\ref{thm:conjA} below is his Conjecture~A. Set \[ R_1(u,v)w=\ip vw u-\ip uw v,\qquad \Pi(x)=\norm{x}^2 I-xx^T . \] Thus $\Jac_x(R_1)=\Pi(x)$. Here $uv^T$ denotes the map $z\mapsto\ip vz u$. Our Jacobi ordering is opposite to that in \cite{Nik2005,Nik2006,Nik2012}; the resulting overall sign changes will be specified where their formulas are used. For an orthogonal map $T$ we write $(T^*R)(u,v)w=T^{-1}R(Tu,Tv)Tw$. A \emph{Clifford structure} consists of skew-symmetric endomorphisms $J_1,\ldots,J_\nu$ with \[ J_iJ_j+J_jJ_i=-2\delta_{ij}I. \] We call $R$ Clifford if its Jacobi family is $\lambda_0\Pi(x)+\sum_i\eta_i(J_ix)(J_ix)^T$ for constants $\lambda_0,\eta_i$. A \emph{symmetric Clifford system} is a family $H_0,\ldots,H_m$ of symmetric endomorphisms satisfying $H_iH_j+H_jH_i=2\delta_{ij}I$. Our main result is the Nikolayevsky (2009) conjecture, Conjecture~A of the preprint version of \cite{Nik2012} (arXiv:0910.1645v1, p.~2); the published version states it without a number \cite[p.~679]{Nik2012}. \begin{mainthm}[Nikolayevsky (2009) conjecture]\label{thm:conjA} Let $R$ be an Osserman algebraic curvature tensor on $\R^{16}$ one of whose reduced Jacobi eigenvalues has multiplicity $7$ or $8$. Then $R$ is Clifford, or $R$ has exactly two reduced eigenvalues and $R=aR_1+b\,T^*R_{\rm Cay}$ for constants $a,b$ and some $T\in O(16)$, as in Theorem~\ref{thm:A}. \end{mainthm} \noindent Nikolayevsky showed that this algebraic statement is the only missing input for the Osserman and the conformal Osserman conjectures in dimension sixteen \cite[Theorem~1 and p.~679]{Nik2012}. \begin{maincor}\label{cor:osserman16} Let $M$ be a connected smooth Riemannian manifold of dimension $16$. \begin{enumerate} \item If $M$ is pointwise Osserman, it is flat or locally isometric to a rank-one symmetric space. \item If $M$ is conformally Osserman, it is locally conformally equivalent to a Euclidean space or to a rank-one symmetric space. \end{enumerate} \end{maincor} \begin{proof} By \cite[Theorem~1 and the paragraph following it, p.~679]{Nik2012}, both statements hold once the curvature tensor, respectively the Weyl tensor, has a Clifford structure or is a combination of $R_1$ and an orthogonal conjugate of $R_{\rm Cay}$ at every point. (In the preprint arXiv:0910.1645v1 this is Theorem~1, ``Assuming Conjecture~A, \dots''; the dimension-sixteen argument ends in Section~3.4.) At points where some reduced eigenvalue has multiplicity $7$ or $8$, this is Theorem~\ref{thm:conjA}. At the other points, the largest multiplicity of a reduced eigenvalue is at least~$7$, by Adams' theorem on vector fields on $S^{15}$ \cite[proof of Theorems~1 and~2, pp.~3--4 of arXiv:math/0204258v2]{Nik2005}; hence it is at least~$9$, and the tensor is Clifford by \cite[Proposition~2.1]{Nik2006} if it is~$9$ and by \cite[Proposition~1]{Nik2005} if it is at least~$10$, as summarized in \cite[pp.~678 and 680]{Nik2012}. \end{proof} \begin{remark}\label{rem:osserman-complete} Together with \cite[Corollary~1.1]{Nik2006}, part~(1) completes the proof of the Osserman conjecture \cite{Oss1990}: every globally Osserman manifold, and every pointwise Osserman manifold of dimension $n\ne2,4$, is flat or locally rank-one symmetric. Together with \cite[Theorem~1]{Nik2010}, part~(2) proves the conformal Osserman conjecture in every dimension $n>4$. \end{remark} \paragraph{Organization of the proof.} We first classify the two-eigenvalue case using its associated quartic. Polynomial eigenfield reductions then leave sixteen multiplicity patterns. The coherent-factor and transpose-pencil arguments treat these by excluding all possible pole branches. Section~\ref{sec:assembly} combines the reductions on a real open set and proves Theorem~\ref{thm:conjA}. The two-eigenvalue statement below also fixes the normalization of $R_{\rm Cay}$ used in the main theorem. \begin{mainthm}\label{thm:A} Let $R$ be an Osserman algebraic curvature tensor on $\R^{16}$ with exactly two reduced eigenvalues, of multiplicities $7$ and $8$. There are constants $\mu$ and $\kappa\ne0$ such that \[ \Jac_x=\mu\Pi(x)+\kappa A(x), \] where, in suitable orthogonal coordinates, $A$ has one of the forms \begin{align} A(x)&=\sum_{i=1}^7(J_ix)(J_ix)^T, &&\text{for a $\Cl(7)$-structure},\label{eq:Cl7}\\ A(x)&=\Pi(x)-\sum_{i=1}^8(J_ix)(J_ix)^T, &&\text{for a $\Cl(8)$-structure},\label{eq:Cl8}\\ A(x)&=\frac12\left(\norm{x}^2I+ \sum_{i=0}^8\ip{H_ix}xH_i\right)-xx^T, &&\text{for a symmetric Clifford system on $\R^{16}$}. \label{eq:Cayley} \end{align} The first two cases are Clifford. In the third, \[ R=(\mu-\kappa/3)R_1+(\kappa/3)\,T^*R_{\rm Cay} \quad\text{for some }T\in O(16), \] where $R_{\rm Cay}$ is the curvature tensor of the Cayley projective plane, normalized to have sectional curvatures between $1$ and $4$. \end{mainthm} \noindent We have not located this algebraic classification in the literature: \cite[Lemma~4]{Nik2012} takes the trichotomy from Conjecture~A, and \cite{AndTwoRoot} classifies two-root tensors only in dimensions twice an odd number. \begin{maincor}\label{cor:B} Let $M$ be a connected smooth pointwise Osserman Riemannian $16$-manifold. Suppose that, at every point, its reduced Jacobi operator has exactly two eigenvalues of multiplicities $7$ and $8$. Then, up to a constant scaling, $M$ is locally isometric to the Cayley projective plane or its noncompact dual. \end{maincor} \noindent Nikolayevsky announced this statement in 2006 \cite[Theorem~1.2, p.~380]{Nik2006}, with its proof deferred; it is described as an announcement without proof in \cite[Section~4]{And2022} and \cite[Section~1, p.~2]{AndTwoRoot}. For an Osserman tensor with reduced eigenspace projections $\pi_i(x)$, Jacobi-proportionality means \begin{equation}\label{eq:JP} \norm{x}^2\ip{\pi_i(x)y}y =\norm{y}^2\ip{\pi_i(y)x}x \quad(x,y\ne0,\ \text{every }i). \end{equation} The radial projection is separated from the reduced zero eigenspace, if that eigenvalue occurs. \begin{maincor}\label{cor:C} Every Jacobi-proportional Riemannian Osserman algebraic curvature tensor on $\R^{16}$ is Clifford or a linear combination of $R_1$ and an orthogonal conjugate of $R_{\rm Cay}$. \end{maincor} \noindent Andreji\'c showed that a Jacobi-proportional tensor which is not Clifford has multiplicities $(8,7)$ or $(7,7,1)$ \cite[Theorem~5]{And2022}, \cite[Theorem~3]{AL2023}; the corollary settles both patterns. Both contain an eigenvalue of multiplicity seven, so the corollary also follows from Theorem~\ref{thm:conjA}; the proof given below, through Theorems~\ref{thm:A} and~\ref{thm:D}, does not use the long argument for Theorem~\ref{thm:conjA}. \begin{mainthm}\label{thm:D} Let $R$ be an Osserman algebraic curvature tensor on $\R^{16}$ with exactly three reduced eigenvalues. If their multiplicities are \[ (7,7,1),\quad(8,6,1),\quad(8,5,2),\quad\text{or }(7,6,2), \] then $R$ is Clifford. \end{mainthm} \noindent Theorem~\ref{thm:D} extends the polynomial eigenfield argument used by Nikolayevsky for an eigenvalue of multiplicity nine \cite[Lemmas~3.1--3.2 and Proposition~2.1]{Nik2006}; the pattern $(7,7,1)$ is one of Andreji\'c's two patterns \cite[Theorem~5]{And2022}. \paragraph{Relation to earlier work.} Osserman \cite{Oss1990} asked whether a manifold whose Jacobi operators have constant eigenvalues on the unit tangent bundle is flat or locally rank-one symmetric. Chi proved this in dimensions not divisible by four and in dimension four \cite{Chi1988}. Nikolayevsky proved it, also in the pointwise form for $n\ne2,4$, except in dimension sixteen when a reduced Jacobi eigenvalue has multiplicity seven or eight \cite{Nik2004,Nik2005,Nik2006}; see \cite[Corollary~1.1, p.~380]{Nik2006}. The algebraic classification behind this leaves exactly these exceptional algebraic cases \cite[Proposition~1]{Nik2005}, \cite[Proposition~2.1]{Nik2006}, \cite[p.~680]{Nik2012}. In \cite{Nik2012} Nikolayevsky reduced the remaining case, and the dimension-sixteen case of the conformal Osserman conjecture, to Conjecture~A. The proof of Theorem~\ref{thm:A} uses M\"unzner's converse to the Cartan--M\"unzner equations, the Ferus--Karcher--M\"unzner construction, and Chi's classification \cite{Chi} of isoparametric hypersurfaces with four principal curvatures and multiplicities $(7,8)$. The geometric consequence uses the reduction in \cite[Section 3]{Nik2012}. Theorem~\ref{thm:D} uses the polynomial eigenfield and rank-one restoration lemmas of \cite{Nik2006}; Corollary~\ref{cor:C} then follows from \cite{And2022,AL2023}. Jacobi-proportionality is an additional condition. In contrast, Raki\'c duality---exchange of unit Jacobi eigenvectors with the same eigenvalue---is equivalent to the Riemannian Osserman condition \cite[Section 2]{NR2013}. The related Jacobi-orthogonality condition is studied in \cite{AL2024}. Neither stronger condition is assumed in Theorem~\ref{thm:conjA}. \section{Normalization and the quartic} Write the reduced eigenvalues as $\mu\norm{x}^2$ of multiplicity $8$ and $(\mu+\kappa)\norm{x}^2$ of multiplicity $7$, and put $A(x)=(\Jac_x-\mu\Pi(x))/\kappa$. Then $A$ is a homogeneous quadratic symmetric matrix family satisfying \begin{equation}\label{eq:Aproperties} A(x)x=0,\qquad A(x)^2=\norm{x}^2A(x),\qquad \tr A(x)=7\norm{x}^2. \end{equation} The Jacobi family determines its curvature tensor. In fact, for any symmetric quadratic family $A$ with $A(x)x=0$, let $B(u,v)=(A(u+v)-A(u)-A(v))/2$. Polarization gives $B(x,y)x=-A(x)y/2$. Since $B(x,y)$ is symmetric, \[ \ip{A(x)y}y=-2\ip{B(x,y)x}y =-2\ip{B(y,x)y}x=\ip{A(y)x}x. \] Jacobi-family reconstruction \cite[Theorem 1]{AL2023} therefore applies; specializing its defining formula \cite[(12), p.~8468]{AL2023} to a quadratic family gives \begin{equation}\label{eq:reconstruct} R_A(u,v)w=\frac23\bigl(B(v,w)u-B(u,w)v\bigr) \end{equation} and gives a curvature tensor with Jacobi family $A$. Set \[ Q(x)=A(x)+xx^T,\qquad L(x)=\R x+\operatorname{im}A(x). \] For $x\ne0$, $Q(x)=\norm{x}^2\pi_{L(x)}$ and $\dim L(x)=8$. In particular, \begin{equation}\label{eq:Qproperties} Q(x)^2=\norm{x}^2Q(x),\qquad Q(x)x=\norm{x}^2x,\qquad \tr Q(x)=8\norm{x}^2. \end{equation} Sectional symmetry implies that the biquadratic form \begin{equation}\label{eq:b} b(x,y)=\ip{Q(x)y}y=\ip{A(x)y}y+\ip xy^2 \end{equation} is symmetric in $x,y$. It vanishes on either coordinate axis and satisfies $b(x,x)=\norm{x}^4$. On $\R^{32}=\R^{16}\oplus\R^{16}$ write $z=(x,y)$, $s=\norm{x}^2+\norm{y}^2$, and define \begin{equation}\label{eq:F} F(z)=s^2-8b(x,y). \end{equation} \begin{lemma}\label{lem:quartic} The polynomial $F$ satisfies \[ \norm{\nabla F}^2=16s^3,\qquad \Delta F=8s. \] Its restriction to $S^{31}$ takes both values $1$ and $-1$. \end{lemma} \begin{proof} Symmetry of $b$ gives $\nabla_yb=2Q(x)y$ and $\nabla_xb=2Q(y)x$. By \eqref{eq:Qproperties}, \[ \norm{\nabla b}^2=4sb,\qquad \ip z{\nabla b}=4b,\qquad \Delta b=2\tr Q(x)+2\tr Q(y)=16s. \] Thus \[ \norm{\nabla F}^2 =16s^3-64s\ip z{\nabla b}+64\norm{\nabla b}^2=16s^3. \] Since $\Delta s^2=4(32+2)s=136s$, we obtain $\Delta F=8s$. For unit $x$, $F(x,0)=1$. If $x,y$ are unit and $y\in L(x)$, then $F((x,y)/\sqrt2)=1-8/4=-1$. \end{proof} \Needspace{10\baselineskip} \section{The isoparametric classification input}\label{sec:external} We record the exact external results needed. \paragraph{M\"unzner.} For a homogeneous polynomial of degree $g$ satisfying $\norm{\nabla F}^2=g^2\norm z^{2g-2}$ and $\Delta F=c\norm z^{g-2}$ on $\R^{n+1}$, nonconstant on the sphere, \cite[Satz 3, p.~65]{Munzner} shows that $0$ is a regular value of $F|S^n$ and that the zero level is an isoparametric hypersurface; by its proof (p.~66) the same holds for every level in $(-1,1)$. Its Folgerung on p.~66 identifies $F$ itself with their Cartan polynomial when both $g(n-1)-c$ and $g(n-1)+c$ are nonzero; these conditions exclude the exceptional doubled-degree alternative. Satz 4(a), p.~67, gives connected level sets. Here $g=4$, $n=31$, and $c=8$, whereas $g(n-1)=120$. The resulting hypersurfaces have four principal curvatures, with \[ m_1+m_2=15,\qquad 8(m_2-m_1)=8, \] so their multiplicities are $(7,8)$. \paragraph{Chi and Ferus--Karcher--M\"unzner.} Chi's classification \cite[abstract, p.~225]{Chi} gives exactly three families with unordered multiplicities $\{7,8\}$. They are the FKM family with parameter $m=7$ and the two FKM families with $m=8$. The first is the Ozeki--Takeuchi family, as identified in \cite[Section 5.9, p.~19 of the English translation]{FKM}; the two $(8,7)$ families are distinguished in its Section 6.6, p.~22. For a symmetric Clifford system $P_0,\ldots,P_m$ on $\R^{32}$, \cite[Theorem 4.1, p.~8; Section 4.3, p.~11]{FKM} gives \begin{equation}\label{eq:FKM} F_P(z)=\norm z^4-2\sum_{i=0}^m\ip{P_iz}z^2,\qquad (m_1,m_2)=(m,15-m). \end{equation} The page numbers for FKM throughout this paper refer to its English translation, arXiv:1112.2780. By the same Folgerung of M\"unzner, with $c=\pm8\ne\pm120$, each $F_P$ used here is the Cartan polynomial of its zero level. We need congruence of the polynomials, not only of a level hypersurface. The zero level of $F$ has principal curvatures $\cot(\pi/8+j\pi/4)$, $0\le j\le3$. A congruent member of the FKM family must therefore be its zero member, with one of the two choices of normal. Let $U\in O(32)$ map the zero level of $F$ onto the zero level of $F_P$. Both zero levels are connected \cite[Satz~4(a), p.~67]{Munzner}. By \cite[Folgerung~(iii), p.~66]{Munzner}, $F$ and $F_P\circ U$ are both Cartan polynomials of this hypersurface. The Cartan polynomial of a connected oriented isoparametric hypersurface is determined by it, and reversing the orientation replaces it by its negative \cite[\S4, p.~65]{Munzner}; explicitly $F(rq)=r^4\cos 4t(q)$ near the hypersurface \cite[proof of Satz~3, p.~66]{Munzner}. Hence $F=\varepsilon F_P\circ U$ with $\varepsilon\in\{1,-1\}$. Absorb $U$ into the matrices $P_i$. Since $\Delta F_P=(136-16(m+1))s$, comparison with $\Delta F=8s$ gives precisely \begin{align} 4b(x,y)&=\sum_{i=0}^7\ip{P_iz}z^2, \label{eq:plus}\\ \text{or}\qquad 4b(x,y)&=s^2-\sum_{i=0}^8\ip{P_iz}z^2. \label{eq:minus} \end{align} These identities hold in the original coordinates $(x,y)$; no preservation of the two coordinate summands by $U$ has been assumed. \section{A lemma on quadratic maps between spheres} Quadratic maps between spheres have a classical literature \cite{Wood,Shapiro}. We give the dimension-threshold argument needed here. Part~(b) below is the quadratic case of \cite[p.~163]{Wood}; we include the short proof since (a) and (c) use it. \begin{lemma}\label{lem:threshold} Let $H_0,\ldots,H_m$ be symmetric $n\times n$ matrices and $h(x)=(\ip{H_ix}x)_{i=0}^m$. Suppose $\norm{h(x)}=\norm x^2$. For $a\in\R^{m+1}$ put $H(a)=\sum_i a_iH_i$. \begin{enumerate}[label=\textup{(\alph*)}] \item If $a\in h(S^{n-1})$, then $I-H(a)\ge0$ and the fibre over $a$ is the unit sphere of $U_a=\ker(I-H(a))$, where $\dim U_a\ge n-m$. \item If $n>2m$, then $h$ is constant on $S^{n-1}$. \item If $n=2m$, $m\ge1$, and $h$ is not constant on $S^{n-1}$, then $H_iH_j+H_jH_i=2\delta_{ij}I$. \end{enumerate} \end{lemma} \begin{proof} For unit $a,x$, $\ip{H(a)x}x=\ip a{h(x)}\le1$, with equality exactly when $h(x)=a$. This proves the fibre description. Fix a unit $x$ in that fibre and rotate the index space so that $a=e_0$. Then $H_0x=x$. Comparing the $t^2$ coefficients of $\norm{h(x+ty)}^2=\norm{x+ty}^4$ gives \[ 2\ip{H_0y}y+4\sum_{i=0}^m\ip{H_ix}y^2 =2\norm y^2+4\ip xy^2. \] The $i=0$ terms cancel, so \begin{equation}\label{eq:rankidentity} I-H_0=2\sum_{i=1}^m(H_ix)(H_ix)^T. \end{equation} Its rank is at most $m$, proving (a). The cones $U_a,U_b$ for distinct values meet only at zero. If $n>2m$, two such cones would have dimensions summing to more than $n$. This proves (b). Suppose $n=2m$ and there is more than one value. For every $a$, comparison with another fibre gives $\dim U_a=m$. Hence the right side of \eqref{eq:rankidentity} has rank $m$, and $H_1x,\ldots,H_mx$ are independent and perpendicular to $x$. On $T_xS^{n-1}=x^\perp$, \[ (dh)_x(y)=2(\ip{H_ix}y)_{i=0}^m \] has rank $m$. The image of $h$ is open in $S^m$ and is compact; since $m\ge1$, it is all of $S^m$. In particular $a$ and $-a$ are values. Their fibres are the $+1$ and $-1$ eigenspaces of $H(a)$ and each has dimension $m$. Thus $H(a)^2=I$ for every unit $a$. Polarization in $a$ proves (c). \end{proof} \section{Recovering the original eigenspaces} \begin{proposition}\label{prop:plus} Identity \eqref{eq:plus} implies \eqref{eq:Cl7}. \end{proposition} \begin{proof} Setting $y=0$ in \eqref{eq:plus} shows that the upper-left diagonal block of each $P_i$ is zero; setting $x=0$ gives the same for the lower-right block. Hence \[ P_i=\begin{pmatrix}0&C_i^T\\ C_i&0\end{pmatrix}, \qquad Q(x)=\sum_{i=0}^7(C_ix)(C_ix)^T . \] The Clifford relations give \[ C_i^TC_j+C_j^TC_i=2\delta_{ij}I,\qquad C_iC_j^T+C_jC_i^T=2\delta_{ij}I. \] Moreover $Q(x)x=\norm x^2x$ implies $\sum_i\ip{C_ix}x^2=\norm x^4$. Apply Lemma~\ref{lem:threshold}(b) to the symmetric parts of $C_i$, with $n=16$ and $m=7$. There is a unit $w\in\R^8$ such that $(\ip{C_ix}x)_i=\norm x^2w$. Rotating the index space gives $w=e_0$. Since $C_0$ is orthogonal and $\ip{C_0x}x=\norm x^2$, we have $C_0=I$. The remaining $C_i$ are skew, and their relations become $C_iC_j+C_jC_i=-2\delta_{ij}I$. Subtracting $xx^T$ from $Q(x)$ proves the claim. \end{proof} \begin{proposition}\label{prop:minus} Identity \eqref{eq:minus} implies \eqref{eq:Cl8} or \eqref{eq:Cayley}. \end{proposition} \begin{proof} In the original splitting write $P_i=\left(\begin{smallmatrix}H_i&B_i^T\\ B_i&D_i\end{smallmatrix}\right)$. The two axes in \eqref{eq:minus} give \[ \sum_{i=0}^8\ip{H_ix}x^2=\norm x^4,\qquad \sum_{i=0}^8\ip{D_iy}y^2=\norm y^4. \] Write $h(x)=(\ip{H_ix}x)_i$ and $d(y)=(\ip{D_iy}y)_i$. If $h$ is constant on the unit sphere, rotate the index so that $H_0=I$, $H_i=0$ for $i\ge1$. The upper block of $P_0^2=I$ gives $B_0=0$. Also $\tr P_0=0$, because $P_0=-P_1P_0P_1$. Thus $D_0^2=I$ and $\tr D_0=-16$, forcing $D_0=-I$. Anticommutation with $P_0=\operatorname{diag}(I,-I)$ makes the other eight matrices off-diagonal. Calling their lower blocks $C_1,\ldots,C_8$ and expanding \eqref{eq:minus} yields \[ Q(x)=\norm x^2I-\sum_{i=1}^8(C_ix)(C_ix)^T. \] The identity $Q(x)x=\norm x^2x$ gives $\sum_i\ip{C_ix}x^2=0$. Each $C_i$ is therefore skew, and the Clifford relations become those of a $\Cl(8)$-structure. This is \eqref{eq:Cl8}. If $h$ is not constant, Lemma~\ref{lem:threshold}(c) makes the $H_i$ a symmetric Clifford system. The equation $H_i^2+B_i^TB_i=I$ forces $B_i=0$. Expanding \eqref{eq:minus} now gives \[ b(x,y)=\frac12\left(\norm x^2\norm y^2-\ip{h(x)}{d(y)}\right). \] Since $b(x,x)=\norm x^4$ and $\norm{h(x)}=\norm{d(x)}=\norm x^2$, equality in Cauchy--Schwarz forces $d(x)=-h(x)$. Polarization gives $D_i=-H_i$. Consequently \[ Q(x)=\frac12\left(\norm x^2I+\sum_{i=0}^8\ip{H_ix}xH_i\right), \] which proves \eqref{eq:Cayley}. \end{proof} \section{Identification of the Cayley tensor}\label{sec:Cayley} We first establish the norm identity for an arbitrary symmetric Clifford system $S_0,\ldots,S_8$ on $\R^{16}$. For $q=\norm x^2$ put $h_i=\ip{S_ix}x$ and $S_h=\sum_i h_iS_i$. The products of zero, one, and four distinct generators are symmetric and trace-orthogonal, and their number is $1+9+126=136$. They therefore form a basis of the symmetric matrices. If $\mathcal L_k$ is their span for degree $k$, the operator $\mathcal T(Y)=\sum_i S_iYS_i$ acts on $\mathcal L_k$ by $(-1)^k(9-2k)$, namely by $9,-7,1$ for $k=0,1,4$. These assertions also appear in \cite[Lemma 3, pp.~6--7 of the preprint]{Nik2012}. Expanding $xx^T=qI/16+S_h/16+Y_4$ therefore gives \begin{equation}\label{eq:HopfM} M_x:=\sum_i(S_ix)(S_ix)^T =xx^T+\frac12(qI-S_h). \end{equation} Contraction with $x$ yields $\norm h^2=q^2+(q^2-\norm h^2)/2$, so $\norm h=q$. For this system define $Q(x)=(qI+S_h)/2$, so that $A(x)=Q(x)-xx^T$ is the matrix in \eqref{eq:Cayley} with $H_i=S_i$. For $x\ne0$ set $w=h/q$. Since $S_h=qS_w$, we have $Q(x)=q(I+S_w)/2$. Also $S_wx=x$, since $S_w$ is an orthogonal involution and $\ip{S_wx}x=q$. Its two eigenspaces have dimension eight. For $u\perp w$, $S_ux$ lies in its negative eigenspace. Thus \[ Q(x)=\frac q2(I+S_w),\qquad M_x=xx^T+qI-Q(x),\qquad S_h=2Q(x)-qI . \] In octonionic coordinates $x=(x_1,x_2)$ the standard system is \[ S_0(x_1,x_2)=(x_1,-x_2),\qquad S_i(x_1,x_2)=(e_i\overline{x_2},\,\overline{x_1}e_i),\quad1\le i\le8, \] where $(e_i)_{i=1}^8$ is an orthonormal basis of $\OCT$. With $(u\wedge v)z=\ip uz v-\ip vz u$, the Brown--Gray Cayley curvature formula \cite[Eq.~(6.12)]{BG}, in the form \cite[equation (7), p.~6 of the preprint]{Nik2012}, is \[ R_{\OCT}(u,v)=3u\wedge v+\sum_{i=0}^8 S_iu\wedge S_iv. \] In our Jacobi convention this gives \begin{align*} \Jac_x(R_{\OCT}) &=3(xx^T-qI)+M_x-S_h\\ &=4xx^T-qI-3Q(x)=-\Pi(x)-3A(x). \end{align*} Therefore $R_{\rm Cay}=-R_{\OCT}$ has Jacobi family $\Pi(x)+3A(x)$, with reduced eigenvalues $1$ and $4$ of multiplicities eight and seven. Let $\Cl_+(9)$ denote the real Clifford algebra on nine generators with squares $+1$ and pairwise anticommutation. Then $\Cl_+(9)\cong M_{16}(\R)\oplus M_{16}(\R)$. Its representations on $\R^{16}$ are equivalent up to orthogonal conjugacy and the sign of one generator; see \cite[Sections 3.4--3.5, pp.~6--7]{FKM} and \cite[p.~6 of the preprint]{Nik2012}. Changing a generator's sign leaves $\ip{S_ix}xS_i$ unchanged. This identifies every family \eqref{eq:Cayley} with the standard Cayley family and finishes the proof of Theorem~\ref{thm:A}. \section{The manifold consequence} \begin{proof}[Proof of Corollary~\ref{cor:B}] A pointwise Osserman metric is Einstein. Its Weyl Jacobi operator is a scalar shift of its curvature Jacobi operator on $x^\perp$, so it also has exactly two roots. In particular the open set $M'$ where the number of Weyl roots is locally constant, used in \cite[Lemma 4]{Nik2012}, is all of $M$. The $N=2$ branch of that lemma (pp.~10--11 of arXiv:0910.1645) uses the algebraic conjecture only to obtain the pointwise $\Cl(7)$, $\Cl(8)$, or Cayley alternatives; in the published version, \cite[Lemma~4, pp.~688--689]{Nik2012}, the alternatives are its hypothesis. Theorem~\ref{thm:A} supplies that trichotomy; disjointness of the alternatives and their smooth local realizations are established inside the cited Lemma 4. The Clifford alternatives are handled by Lemmas 5 and 6 of \cite{Nik2012} (pp.~12--18 of the preprint), including the case $\nu=8$; its Lemma 7 and Section 3.3 handle the Cayley alternative. These yield local conformal equivalence to rank-one symmetric metrics. The final Einstein argument of Section 3.4, using \cite[Theorem 4.4]{Nickerson}, gives local symmetry of the original metric, and the last paragraph of \cite[Section~3.4]{Nik2012} makes it flat or locally rank-one symmetric; flatness is excluded by the two distinct roots. (The corollary also follows from Corollary~\ref{cor:osserman16}(1); the proof given here uses only Theorem~\ref{thm:A}.) The only rank-one symmetric $16$-dimensional models with multiplicities $(7,8)$ are the Cayley projective and hyperbolic planes. Connectedness fixes the type and scale, proving the corollary. \end{proof} \section{Three-root tensors and proportionality} For the Cayley model let $E(x)=L(x)\cap x^\perp$ and $N(x)=L(x)^\perp$, of dimensions seven and eight. \begin{lemma}\label{lem:nofield} There is no nonzero linear map $K$ with $Kx\in E(x)$ for every $x\ne0$. There is also no nonzero linear map $K$ with $Kx\in N(x)$ for every $x\ne0$. \end{lemma} \begin{proof} After the orthogonal coordinate change $(x_1,x_2)\mapsto(x_1,\overline{x_2})$, the Cayley lines are \[ \mathcal C_\omega=\{(u,\omega u):u\in\OCT\},\quad \omega\in\OCT,\qquad \mathcal C_\infty=\{(0,v):v\in\OCT\}. \] Here $S_{(w_0,a)}=w_0S_0+\sum_{i=1}^8a_iS_i$ for $w_0^2+\norm a^2=1$ and $a=\sum_i a_i e_i$. Its positive eigenspace becomes $\mathcal C_\omega$ with $\omega=\overline a/(1+w_0)$ when $w_0\ne-1$, and becomes $\mathcal C_\infty$ otherwise. For $x$ in such a line, $L(x)$ is that line. Suppose first $Kx\in E(x)$. Since $\ip{Kx}x=0$, $K$ is skew. Preservation of $\mathcal C_0,\mathcal C_\infty,\mathcal C_1$ forces $K=\operatorname{diag}(D,D)$. Preservation of $\mathcal C_\omega$ gives $D(\omega u)=\omega D(u)$. Put $c=D(1)$; then $D(u)=uc$ and $(\omega u)c=\omega(uc)$ for all $\omega,u$. Thus $c$ lies in the right nucleus of $\OCT$, which is $\R$. Hence $K=cI$, and skewness forces $c=0$. Suppose next $Kx\in N(x)$. The two axes force $K=\left(\begin{smallmatrix}0&B\\ C&0\end{smallmatrix}\right)$. The graph condition for $\mathcal C_\omega^\perp$ is $B(\omega u)=-\overline\omega C(u)$. Taking $\omega=1$ gives $B=-C$; taking $u=1$ then gives $C(\omega)=\overline\omega\gamma$, where $\gamma=C(1)$. Substitution yields \[ (ab)\gamma=b(a\gamma)\qquad(a,b\in\OCT). \] Choose a quaternion subalgebra containing $\gamma$ and choose noncommuting $a,b$ in it. Associativity in that subalgebra gives $(ab-ba)\gamma=0$. Multiplicativity of the octonion norm forces $\gamma=0$, and therefore $K=0$. \end{proof} \begin{proof}[Proof of Theorem~\ref{thm:D}] Let $t\in\{1,2\}$ be the smallest multiplicity, with eigenvalue $\nu$. For the purpose of applying the polynomial eigenfield lemma, shift a different reduced eigenvalue to zero by subtracting a multiple of $R_1$. In the convention of \cite[p.~379]{Nik2006}, the shifted tensor has $k=3$ reduced eigenvalues, one of them zero. Nikolayevsky's \cite[Lemma 3.1(1),(2), p.~383]{Nik2006} gives an orthonormal frame for the $\nu$-eigenspace by odd homogeneous polynomial fields of degree at most $k-1=2$. They are therefore linear: $J_1x,\ldots,J_tx$. Their norm and orthogonality identities make each $J_i$ skew and orthogonal and make the $J_i$ anticommute. Return to the original tensor, whose eigenspaces are unchanged by the shift. Choose the other root $\mu$ so that the merger is \[ (7,7,1)\to(7,8),\quad (8,6,1)\to(8,7),\quad (8,5,2)\to(8,7),\quad (7,6,2)\to(7,8). \] For a skew orthogonal $J$, let $R_J$ denote the curvature tensor whose Jacobi family is $(Jx)(Jx)^T$, obtained for example from \eqref{eq:reconstruct}. Then \[ \widetilde R=R-(\nu-\mu)\sum_{i=1}^t R_{J_i} \] is Osserman with two distinct roots of multiplicities seven and eight. Apply Theorem~\ref{thm:A}. Its Cayley alternative is impossible: each $J_ix$ remains an eigenfield in one of the two merged distributions, contradicting Lemma~\ref{lem:nofield} after conjugating the field to the standard model: $K=TJ_iT^{-1}$. Thus $\widetilde R$ is Clifford. Restore the rank-one terms successively. By \cite[Lemma 3.2(1), pp.~383--384]{Nik2006}, if $R$ is Osserman, $J$ is orthogonal and skew, $c\ne0$, and $R-cR_J$ is Osserman and Clifford, then $R$ is Clifford. Each intermediate tensor here is Osserman: only constant eigenvalues on fixed mutually orthogonal eigenlines change. The lemma consequently applies at every step. (The proof of \cite[Lemma~3.2(1)]{Nik2006} shifts the root of largest multiplicity of $R-cR_J$ to zero. For an endpoint of type \eqref{eq:Cl8}, shift instead the base constant $\lambda_0$ of its Clifford structure to zero; the proof then applies verbatim, and when $\nu=8$ its remaining coefficient $c_0$ vanishes because $\R^{16}$ carries no nine anticommuting orthogonal complex structures.) Its source convention changes the sign of the coefficient, but only its nonvanishing is required. This proves the theorem. \end{proof} \begin{proof}[Proof of Corollary~\ref{cor:C}] Andreji\'c \cite[Theorem 5]{And2022}, restated as \cite[Theorem 3, p.~8470]{AL2023}, proves that a Riemannian Jacobi-proportional non-Clifford algebraic curvature tensor has multiplicities $(8,7)$ or $(7,7,1)$. There is no restriction on the number of roots in the hypothesis. Apply Theorems~\ref{thm:A} and \ref{thm:D}. \end{proof} \input{sec_small_roots} \input{sec_polynomial_gram} \input{sec_coherent_transpose_poles} \input{sec_real_pole_types} \input{sec_rank_seven_no_poles} \input{sec_simple_rank_four_pole} \input{sec_double_quadratic_exclusion} \input{sec_closed_kernel_and_ten_space} \input{sec_scalar_span} \input{sec_small_support_incidence} \input{sec_aligned_quadratic_two_poles} \input{sec_quadratic_proper_supports} \input{sec_real_rank_four_images} \input{sec_union_conjugate_rank_four} \input{sec_divisor_budget} \input{sec_balanced_alignment} \input{sec_zero_stratum_unbalanced} \input{sec_positive_model_factorial} \input{sec_singular_integral_extraction} \input{sec_conjugate_preliminary_reductions} \input{sec_conjugate_gluing} \input{sec_final_assembly} \section{Remarks} \begin{remark}[The quartic in other dimensions] For a two-root Osserman tensor on $\R^n$ with multiplicities $(\nu,n-1-\nu)$, the same construction has $\rank Q=\nu+1$ and \[ \norm{\nabla F}^2=16s^3,\qquad \Delta F=8(n-1-2\nu)s. \] Thus it yields a Cartan--M\"unzner quartic on $\R^{2n}$ with those multiplicities. A classification of the quartic alone does not recover the original splitting: a further argument such as Propositions~\ref{prop:plus}--\ref{prop:minus} is required. In particular the homogeneous non-FKM multiplicities $(2,2)$ and $(4,5)$ would require their own such argument. \end{remark} \begin{remark}[The definite FKM family is excluded] For nine generators on $\R^{32}$ the Clifford product $P_0\cdots P_8$ is central. The definite $(8,7)$ family has this product equal to $I$ or $-I$; the indefinite family has trace zero \cite[Sections 3.5 and 6.6]{FKM}. In the constant branch of Proposition~\ref{prop:minus} it is $\operatorname{diag}(C_1\cdots C_8,-C_1\cdots C_8)$. In the nonconstant branch it is $\operatorname{diag}(H_0\cdots H_8,-H_0\cdots H_8)$. Both have trace zero. Hence the definite FKM$(8,7)$ family cannot arise from the quartic of a two-root tensor. An orthogonal change of the generator index changes the product only by a sign. \end{remark} \section*{Acknowledgments} This work was done overwhelmingly by AI models and supervised by the author. The models were Anthropic's Claude and OpenAI's ChatGPT. The author takes full responsibility for the paper. \section*{Statements and Declarations} \noindent\textbf{Funding.} No funding was received for conducting this study. \smallskip\noindent\textbf{Competing interests.} The author has no competing interests to declare that are relevant to the content of this article. \smallskip\noindent\textbf{Data availability.} Data sharing is not applicable to this article, as no datasets were generated or analysed during the current study. \begin{thebibliography}{99} \bibitem{And2022} V.~Andreji\'c, \emph{The proportionality principle for Osserman manifolds}, J. Geom. Phys. \textbf{176} (2022), 104516. \href{https://doi.org/10.1016/j.geomphys.2022.104516}{doi:10.1016/j.geomphys.2022.104516}. \bibitem{AndTwoRoot} V.~Andreji\'c, \emph{Two-root Riemannian manifolds}, Mediterr. J. Math. \textbf{20} (2023), article 100. \href{https://doi.org/10.1007/s00009-023-02340-2}{doi:10.1007/s00009-023-02340-2}; \href{https://arxiv.org/abs/2009.12834}{arXiv:2009.12834v4}. \bibitem{AL2023} V.~Andreji\'c and K.~Luki\'c, \emph{On the existence of a curvature tensor for given Jacobi operators}, Filomat \textbf{37}:25 (2023), 8465--8471. \href{https://doi.org/10.2298/FIL2325465A}{doi:10.2298/FIL2325465A}. \url{https://www.pmf.ni.ac.rs/filomat-content/2023/37-25/37-25-6-19416.pdf}. \bibitem{AL2024} V.~Andreji\'c and K.~Luki\'c, \emph{The orthogonality principle for Osserman manifolds}, Acta Math. Hungar. \textbf{173} (2024), 246--252. \href{https://doi.org/10.1007/s10474-024-01434-x}{doi:10.1007/s10474-024-01434-x}; \href{https://arxiv.org/abs/2308.14851}{arXiv:2308.14851}. \bibitem{BochnakKucharz} J.~Bochnak and W.~Kucharz, \emph{On homology classes represented by real algebraic varieties}, Banach Center Publ. \textbf{44} (1998), 21--35. \url{https://eudml.org/doc/208886}. \bibitem{BG} R.~B. Brown and A.~Gray, \emph{Riemannian manifolds with holonomy group $\operatorname{Spin}(9)$}, in \emph{Differential Geometry (in Honor of Kentaro Yano)}, Kinokuniya, Tokyo, 1972, 41--59. \bibitem{Chi1988} Q.-S. Chi, \emph{A curvature characterization of certain locally rank-one symmetric spaces}, J. Differential Geom. \textbf{28} (1988), 187--202. \href{https://doi.org/10.4310/jdg/1214442277}{doi:10.4310/jdg/1214442277}. \bibitem{Chi} Q.-S. Chi, \emph{Isoparametric hypersurfaces with four principal curvatures, IV}, J. Differential Geom. \textbf{115} (2020), 225--301. \href{https://doi.org/10.4310/JDG/1589853626}{doi:10.4310/JDG/1589853626}; \href{https://arxiv.org/abs/1605.00976}{arXiv:1605.00976}. \bibitem{DNT2017} T.~Drescher, T.~Netzer and A.~Thom, \emph{On projections of free semialgebraic sets}, preprint (2017), introduction, pp.~1--2. \href{https://arxiv.org/abs/1709.08424}{arXiv:1709.08424}. \bibitem{EGHP2004} D.~Eisenbud, M.~Green, K.~Hulek and S.~Popescu, \emph{Small schemes and varieties of minimal degree}, arXiv:math/0404517. \url{https://arxiv.org/abs/math/0404517}. \bibitem{FKM} D.~Ferus, H.~Karcher and H.-F. M\"unzner, \emph{Cliffordalgebren und neue isoparametrische Hyperfl\"achen}, Math. Z. \textbf{177} (1981), 479--502. English translation by T.~E. Cecil, \emph{Clifford algebras and new isoparametric hypersurfaces}, \href{https://arxiv.org/abs/1112.2780}{arXiv:1112.2780}. \bibitem{GKV2002} E.~Garc\'ia-R\'io, D.~N. Kupeli and R.~V\'azquez-Lorenzo, \emph{Osserman Manifolds in Semi-Riemannian Geometry}, Lecture Notes in Math. \textbf{1777}, Springer, Berlin, 2002. \href{https://doi.org/10.1007/b83213}{doi:10.1007/b83213}. \bibitem{Gilkey2001} P.~B. Gilkey, \emph{Geometric Properties of Natural Operators Defined by the Riemann Curvature Tensor}, World Scientific, River Edge, NJ, 2001. \bibitem{Kryszewski} W.~Kryszewski, \emph{Remarks to the Vietoris theorem}, Topol. Methods Nonlinear Anal. \textbf{8} (1996), 383--405. \href{https://doi.org/10.12775/TMNA.1996.041}{doi:10.12775/TMNA.1996.041}; \url{https://projecteuclid.org/journals/topological-methods-in-nonlinear-analysis/volume-8/issue-2/Remarks-to-the-Vietoris-theorem/tmna/1479265248.full}. \bibitem{MartinezLegaz2005} J.~E.~Mart\'inez-Legaz, \emph{On Brickman's theorem}, J. Convex Anal. \textbf{12} (2005), 139--143. \url{https://www.heldermann-verlag.de/jca/jca12/jca0474_b.pdf}. \bibitem{Munzner} H.-F. M\"unzner, \emph{Isoparametrische Hyperfl\"achen in Sph\"aren. I}, Math. Ann. \textbf{251} (1980), 57--71. \bibitem{Nickerson} H.~K. Nickerson, \emph{On conformally symmetric spaces}, Geom. Dedicata \textbf{18} (1985), 87--99. \bibitem{Nik2004} Y.~Nikolayevsky, \emph{Osserman manifolds of dimension 8}, Manuscripta Math. \textbf{115} (2004), 31--53. \href{https://doi.org/10.1007/s00229-004-0480-y}{doi:10.1007/s00229-004-0480-y}; \href{https://arxiv.org/abs/math/0310387}{arXiv:math/0310387}. \bibitem{Nik2005} Y.~Nikolayevsky, \emph{Osserman conjecture in dimension $n\ne8,16$}, Math. Ann. \textbf{331} (2005), 505--522. \href{https://doi.org/10.1007/s00208-004-0580-8}{doi:10.1007/s00208-004-0580-8}; \href{https://arxiv.org/abs/math/0204258}{arXiv:math/0204258}. \bibitem{Nik2006} Y.~Nikolayevsky, \emph{On Osserman manifolds of dimension 16}, in \emph{Contemporary Geometry and Related Topics} (N.~Bokan et al., eds.), Proceedings of the Conference, Belgrade, June 26--July 2, 2005, Faculty of Mathematics, University of Belgrade, 2006, 379--398. \url{https://www.maths.tcd.ie/EMIS/proceedings/CGRT2005/Articles/cgrt24.pdf}. \bibitem{Nik2010} Y.~Nikolayevsky, \emph{Conformally Osserman manifolds}, Pacific J. Math. \textbf{245} (2010), 315--358. \href{https://doi.org/10.2140/pjm.2010.245.315}{doi:10.2140/pjm.2010.245.315}; \href{https://arxiv.org/abs/0810.5621}{arXiv:0810.5621}. \bibitem{Nik2012} Y.~Nikolayevsky, \emph{Conformally Osserman manifolds of dimension 16 and a Weyl--Schouten theorem for rank-one symmetric spaces}, Ann. Mat. Pura Appl. \textbf{191} (2012), 677--709 (published online 2011). \href{https://doi.org/10.1007/s10231-011-0201-6}{doi:10.1007/s10231-011-0201-6}. Preprint: \emph{Osserman manifolds and the Weyl--Schouten theorem for rank-one symmetric spaces}, \href{https://arxiv.org/abs/0910.1645v1}{arXiv:0910.1645v1}. \bibitem{NR2013} Y.~Nikolayevsky and Z.~Raki\'c, \emph{A note on Raki\'c duality principle for Osserman manifolds}, Publ. Inst. Math. (Beograd) (N.S.) \textbf{94}(108) (2013), 43--45. \href{https://doi.org/10.2298/PIM1308043N}{doi:10.2298/PIM1308043N}; \href{https://arxiv.org/abs/1204.1600}{arXiv:1204.1600}. \bibitem{Oss1990} R.~Osserman, \emph{Curvature in the eighties}, Amer. Math. Monthly \textbf{97} (1990), 731--756. \href{https://doi.org/10.1080/00029890.1990.11995659}{doi:10.1080/00029890.1990.11995659}. \bibitem{Samuel1964} P.~Samuel, \emph{Lectures on Unique Factorization Domains}, Tata Institute of Fundamental Research, Bombay, 1964. \url{https://mathweb.tifr.res.in/Documents/Publications/Lectures/tifr30.pdf}. \bibitem{Shapiro} D.~B. Shapiro, \emph{Compositions of Quadratic Forms}, de Gruyter Expositions in Mathematics \textbf{33}, Walter de Gruyter, Berlin--New York, 2000. \bibitem{Wood} R.~Wood, \emph{Polynomial maps from spheres to spheres}, Invent. Math. \textbf{5} (1968), 163--168. \href{https://doi.org/10.1007/BF01425547}{doi:10.1007/BF01425547}. \end{thebibliography} \end{document}