\documentclass[11pt]{article} \usepackage[letterpaper,margin=1in,headheight=14pt]{geometry} \usepackage[T1]{fontenc} \usepackage{lmodern,amsmath,amssymb,amsthm,mathtools} \usepackage{booktabs,longtable,array,tabularx} \usepackage{enumitem,microtype,fancyhdr} \usepackage{tikz,graphicx,caption} \captionsetup[figure]{name=Fig.,labelfont=bf,labelsep=space} \usepackage[dvipsnames]{xcolor} \usepackage{listings} \usepackage[hidelinks]{hyperref} \setlength{\parindent}{0pt} \setlength{\parskip}{0.48em} \setlength{\emergencystretch}{3em} \setlength{\tabcolsep}{6pt} \renewcommand{\arraystretch}{1.15} \allowdisplaybreaks \numberwithin{equation}{section} \pagestyle{fancy} \fancyhf{} \fancyhead[L]{Certified odd--even Markoff collision tiers} \fancyhead[R]{E. F. W. Heffern} \fancyfoot[C]{\thepage} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{problem}[theorem]{Problem} \newtheorem{example}[theorem]{Example} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \newcommand{\Z}{\mathbb Z} \newcommand{\unitary}{\mathbin{\|}} \newcommand{\defect}{\mathscr D} \DeclareMathOperator{\ord}{ord} \DeclareMathOperator{\Nm}{N} % Compact vector rendering; literal text: 天下为公. \newcommand{\closingmotto}{% \begin{tikzpicture}[x=.64cm,y=.64cm,line width=.9pt, line cap=round,line join=round] \draw (0.08,.86)--(.92,.86) (0,.58)--(1,.58) (.50,.86)--(.48,.37) (.48,.46)--(.10,.05) (.48,.46)--(.92,.05); \draw (1.25,.86)--(2.25,.86) (1.75,.86)--(1.75,.05) (1.78,.51)--(2.18,.29); \draw (2.80,.94)--(2.91,.78) (2.63,.70)--(2.49,.53) (2.55,.57)--(3.43,.57)--(3.30,.12)--(3.17,.04) (3.00,.88)--(2.65,.08) (2.91,.48)--(3.08,.30); \draw (3.98,.88)--(3.70,.54) (4.16,.88)--(4.48,.55) (4.07,.57)--(3.75,.10)--(4.40,.16) (4.40,.16)--(4.19,.45) (4.29,.32)--(4.50,.06); \end{tikzpicture}} \lstset{ basicstyle=\ttfamily\footnotesize, columns=fullflexible, frame=single, framesep=6pt, xleftmargin=6pt, xrightmargin=6pt, breaklines=true, backgroundcolor=\color{black!3}, showstringspaces=false } \hypersetup{ pdftitle={Hyperbolic Collision Invariants and Certified Odd--Even Tiers toward Markoff Unicity}, pdfauthor={Elleard Felix Webster Heffern}, pdfsubject={Markoff numbers and exact computational certificates}, pdfkeywords={Markoff numbers, unicity conjecture, Diophantine equations, computer-assisted proof, continued fractions} } \title{\textbf{Hyperbolic Collision Invariants and Certified Odd--Even Tiers toward Markoff Unicity}} \author{Elleard Felix Webster Heffern\\[0.25em] \small Independent Researcher, St. Louis, Missouri, USA\\[0.15em] \small Corresponding author: \href{mailto:soldier@alum.mit.edu}{soldier@alum.mit.edu}} \date{30 September 2026} \begin{document} \maketitle \begin{abstract} A collision would consist of two different ordered Markoff triples with the same largest entry $M$. We derive arithmetic conditions that every such collision must satisfy. A parametrization by hyperbolic functions factors a classical determinant identity and leads to two coupled quadratic equations. If $M$ is odd, these equations require a factorization $M=FE$ with $F,E>1$ and $\gcd(F,E)=1$. Hand proofs and exact finite computations give increasingly strong bounds, ending with $F^4,E^4\ge300000M+200000$. If $M$ is even, division of the collision quantities by four gives the same inequalities for a coprime factorization of $M/2$. These bounds yield explicit sufficient conditions for a Markoff number to be unique. The finite computations cover a proved search domain and apply necessary congruence and continued-fraction conditions. Two different arguments reject the final odd candidate: one uses continued fractions, and the other bounds the integer solutions of two quadratic equations of Pell type. Separately written programs reproduce the computations. The constants measure the strength of collision inequalities; they are not ranges through which Markoff numbers were checked. We identify the classical ingredients and the quantitative additions. Factorizations with increasingly comparable factors remain unresolved, so the general unicity conjecture is not proved. \end{abstract} \medskip \noindent\textbf{2020 Mathematics Subject Classification.} Primary 11D25; Secondary 11D09, 11A55, 11Y50. \noindent\textbf{Keywords.} Markoff numbers; Frobenius unicity conjecture; unitary divisors; continued fractions; exact finite certificate. \section{Introduction} A \emph{Markoff triple} is a positive integral solution of \begin{equation} x^2+y^2+z^2=3xyz. \label{eq:markoff} \end{equation} The equation was introduced by Markoff \cite{Markoff1879}. An integer occurring in such a triple is a Markoff number. The unicity conjecture posed by Frobenius \cite{Frobenius1913} asserts that a Markoff triple is determined, up to permutation, by its largest entry. The conjecture remains open. It is known for prime powers, twice prime powers, and several other arithmetic families; see \cite{Baragar1996,Button1998,Button2001,LangTan2007,Zhang2007, ChenChen2013,Srinivasan2009,Aigner2013}. Throughout, an \emph{ordered occurrence with maximum \(M\)} means a triple \((a,b,M)\) with \(a\le b\le M\). An \emph{ordered collision} is a pair of distinct such occurrences with the same maximum. We call \(M\) \emph{unique} when it has only one ordered occurrence, equivalently when its Markoff triple is determined up to permutation. \paragraph{Reading guide and notation.} The argument uses algebra, calculus, prime factorization, and congruences. A congruence \(u\equiv v\pmod n\) means that \(n\mid u-v\). Integers are \emph{coprime} when their greatest common divisor is one. A complete prime-power block of an integer is the factor \(p^e\) containing its entire power of a prime \(p\). The notation \(\ord_p(z)\) records the exponent of \(p\) in \(z\); this is also called its \(p\)-adic valuation. A \emph{quadratic residue} is a residue that is a square modulo the stated prime. The words \emph{packet}, \emph{floor}, and \emph{tier} will have specific meanings here. A packet is one of the displayed quadratic equations together with its integer variables and divisibility conditions. A floor is a proved lower bound for a product of its coordinates. A tier is the resulting level of the collision bound. A \emph{relaxation} keeps only some necessary conditions, so it may include tuples that do not come from Markoff triples. Excluding every tuple in this larger set also excludes actual collisions. We identify explicitly which steps use a finite computer calculation. The hyperbolic functions need no background beyond exponentials: \(\cosh t=(e^t+e^{-t})/2\) and \(\sinh t=(e^t-e^{-t})/2\). The notation \(\log\) means the natural logarithm. Later, a continued fraction \([c_1;c_2,\ldots,c_r]\) means \(c_1+1/(c_2+1/(\cdots+1/c_r))\). Its digits \(c_i\) are called partial quotients. Requiring the last digit to exceed one makes the finite expansion unique. This paper studies the arithmetic forced by an ordered collision. Let \((a,b,M)\) and \((a',b',M)\) form an ordered collision. For odd \(M\), the central chain is \[ \begin{gathered} \boxed{\text{hyperbolic four-point factorization}} \Longrightarrow \boxed{AB=M^2C} \Longrightarrow \boxed{M=FE\text{ unitary}}\\ \Longrightarrow \boxed{\text{two defect equations}} \Longrightarrow \boxed{\text{residual crossed-Pell coupling}}\\ \Longrightarrow \boxed{\text{four certified odd floors}}. \end{gathered} \] The first step supplies a geometric factorization and interpretation of Srinivasan's identity. The direct verification of that identity and all later algebraic reductions use integers. The following table separates each product floor from the coefficient it gives in the collision inequality. \begin{table}[ht] \centering \small \begin{tabular}{@{}r r l l@{}} \toprule tier & defect floor & provenance & quartic consequence \\ \midrule \(99\) & \(33\) & hand proof & \(F^4,E^4\ge 99M+66\) \\ \(360\) & \(120\) & exact local enumeration & \(F^4,E^4\ge 360M+240\) \\ \(201411\) & \(67137\) & companion-prime certificate & \(F^4,E^4\ge 201411M+134274\) \\ \(300000\) & \(100000\) & crossed-Pell/continuant certificates & \(F^4,E^4\ge 300000M+200000\) \\ \bottomrule \end{tabular} \caption{The four certified odd tiers. The coefficient is three times the defect-product floor.} \label{tab:tiers-intro} \end{table} For even maxima, dividing the collision quantities by four gives the following parallel bounds. \begin{table}[ht] \centering \small \begin{tabular}{@{}r r l l@{}} \toprule tier & defect floor & provenance & quartic consequence \\ \midrule \(216\) & \(72\) & hand proof & \(F^4,E^4\ge216M+144\) \\ \(3042\) & \(1014\) & companion-prime certificate & \(F^4,E^4\ge3042M+2028\) \\ \(300000\) & \(100000\) & continuant-root certificate & \(F^4,E^4\ge300000M+200000\) \\ \bottomrule \end{tabular} \caption{The three certified tiers for even \(M\), after normalization by \(M_1=M/2\).} \label{tab:even-tiers-intro} \end{table} In the odd table, \(120\) is sharp for the local relaxation and \(67137\) is sharp after the companion-prime condition. In the even table, \(72\) and \(1014\) are sharp for their corresponding relaxations. In both parities, \(100000\) is a certified cutoff, with no claim of sharpness. None of these numbers is a range through which Markoff unicity has merely been checked. The factorization needed for the argument keeps every complete prime-power block on one side. This is expressed by a unitary divisor. A divisor \(d\mid N\) is \emph{unitary}, written \(d\unitary N\), when \(\gcd(d,N/d)=1\). Define \begin{equation} u_*(N)=\max\{d:d\unitary N,\ d\le\sqrt N\}. \label{eq:ustar} \end{equation} The set is nonempty because it contains \(1\). \begin{theorem}[Main collision and unicity theorem] \label{thm:main} If an odd Markoff number \(M>1\) admits an ordered collision, then there are coprime unitary factors \(F,E>1\), \(M=FE\), for which \begin{equation} F^4,E^4\ge 300000M+200000. \label{eq:mainquartic} \end{equation} If an even Markoff number \(M>2\) admits an ordered collision, then \(M_1=M/2\equiv1\pmod{16}\) has coprime proper unitary factors \(F,E>1\), \(M_1=FE\), satisfying the same bound \eqref{eq:mainquartic}. Consequently, \begin{align} u_*(M)^4&<300000M+200000 &&\Longrightarrow M\text{ is unique} &&(M\text{ odd}), \label{eq:mainunitary-odd}\\ u_*(M/2)^4&<300000M+200000 &&\Longrightarrow M\text{ is unique} &&(M\text{ even}). \label{eq:mainunitary-even} \end{align} The corresponding weaker statements hold at every tier in Tables~\ref{tab:tiers-intro} and \ref{tab:even-tiers-intro}, with their stated proof provenance. \end{theorem} \begin{corollary}[Prime-power tail] \label{cor:tail-intro} Let \(M=kq^e\) be an odd Markoff number, where \(q\) is prime, \(\gcd(k,q)=1\), and \(e\ge1\). For any \[ \mathcal C\in\{99,360,201411,300000\}, \] the condition \begin{equation} k^3\le \mathcal C q^e \label{eq:tail-intro} \end{equation} guarantees that \(M\) is unique, provided the tier with coefficient \(\mathcal C\) is invoked with its stated hand or computer-assisted provenance. \end{corollary} The odd collision assertion of Theorem~\ref{thm:main} is proved in Theorem~\ref{thm:300000}, and its even counterpart in Theorem~\ref{thm:even-tiers}. Their unicity implications are Theorems~\ref{thm:unitarygap} and \ref{thm:even-unitarygap}. Corollary~\ref{cor:tail-intro} follows from Theorem~\ref{thm:compositetail} by taking a single prime-power tail. \subsection*{Historical position and quantitative comparison} Classical partial results are usually formulated as arithmetic-shape criteria. Baragar proved unicity when one of \(M,3M-2,3M+2\) is a prime, twice a prime, or four times a prime \cite{Baragar1996}. Button proved the prime case and later developed a principal-ideal and continued-fraction method \cite{Button1998,Button2001}; Lang and Tan gave an independent proof for prime powers \cite{LangTan2007}. On the companion side, Zhang proved unicity when one of \(3M\pm2\) is a prime power, four times a prime power, or eight times a prime power \cite{Zhang2007}. Chen and Chen subsequently treated \(3M\pm2=kp^\ell\) for \(1\le k\le10^{10}\) \cite{ChenChen2013}. Srinivasan gave an elementary treatment of the prime-power and twice-prime-power cases and further criteria in terms of ambiguous classes and the prime divisors of \(9M^2-4\) \cite{Srinivasan2009}. More specifically, in the present sign convention, Srinivasan's identity \cite[Lemma~2.1]{Srinivasan2009} is exactly \[ (bb'-aa')(ba'-ab')=M^2(ab-a'b'), \] that is, \(AB=M^2C\). Her analysis also contains the underlying coprime allocation of complete prime-power blocks produced by a hypothetical nonunique maximum \cite[Theorem~5.1]{Srinivasan2009}. Thus neither the determinant identity itself nor, in substance, the existence of a proper unitary split is asserted here as new. The companion-discriminant allocation formulated below likewise has a classical genus-theoretic antecedent. Srinivasan's ambiguous-class analysis and Legendre-symbol criteria detect the same mutual quadratic-residue condition between complementary factors of \(9M^2-4\) \cite[Theorems~5.4 and 6.1]{Srinivasan2009}. The elementary residual equation below is used as a bridge from the present collision variables to that condition; priority is not claimed for the unitary or residue obstruction. The additional bounded crossed-Pell reconstruction couples this residual allocation back to the primitive collision slope and supplies a second exact rejection of the final odd certificate row. For even Markoff numbers, Zhang proved the congruence \(M\equiv2\pmod{32}\) and companion-number criteria \cite[Theorems~1--2]{Zhang2007}. Srinivasan's treatment of twice prime powers already uses the determinant identity with its forced divisibility by \(4\), and her companion lemma states that every odd prime divisor of \(9M^2-4\) is \(1\pmod4\) \cite[Theorem~2.2 and Lemma~4.2]{Srinivasan2009}. The even section below combines these antecedents with the complementary conics and an exact coefficient-\(6\) enumeration; it does not claim priority for the congruence, the coprime splitting in substance, or the companion-prime condition. Geometric reformulations also predate the present work. Tornero's fixed-\(M\) hyperbola \[ H_M:\quad x^2-3Mxy+y^2+M^2=0 \] is exactly the conic used below \cite{Tornero2008}; a recent reformulation uses integral necklaces \cite{Fisac2025}. The narrower role of the hyperbolic viewpoint here is the closed-form \(Q^x\) parametrization and the explicit four-point factorization of the three terms in Srinivasan's identity. \paragraph{Recent related work.} O'Sullivan gives geometric growth and approximation estimates for real Markov trees \cite[Section~3 and Proposition~8.1]{OSullivan2026}. Yasutomi proves injectivity for substitution parameters $a\ge2$, $b=a+1$, and a partial ordering result in the classical case $a=1$, $b=2$ \cite[Theorems~1.3 and 1.5(2)]{Yasutomi2026}. The full classical case is outside his injectivity theorem. Banaian extends Aigner's ordering results to generalized $k$-Markov numbers \cite[Theorem~4]{Banaian2026}; Gyoda surveys the associated arithmetic, combinatorial, and geometric descriptions \cite{Gyoda2026}. Our question is different: given a hypothetical classical collision, how large must each factor in its forced coprime split be? The quartic bounds in Theorem~\ref{thm:main} address that question. They are not a replacement for these ordering or approximation results, and the proof does not depend on them. \subsection*{Relation to the author's prior preprint} The author's preprint \emph{A slope--scale factorization for prime-power and twice-prime-power Markoff unicity} (ResearchGate, 18 June 2026; \href{https://doi.org/10.13140/RG.2.2.23525.56800}{https://doi.org/10.13140/RG.2.2.23525.56800}) developed the \(Q^x\) parametrization, the \(A,B,C,D\) identities, the collision line, the slope--scale formulas, the unitary allocation, and the coupled defect equations. The present article substantially extends and supersedes that preprint by isolating the complementary four-point conics and, principally, by establishing the four odd defect floors \(33,120,67137,100000\), the normalized even floors \(72,1014,100000\), their unitary-gap consequences, the residual crossed-Pell reconstruction, and self-contained exact-arithmetic certificates. Claims of contribution below concern these quantitative extensions and their certificates, not the previously published or publicly circulated structural machinery. The direct coefficient-\(1\) antecedent of the present unitary-gap inequalities is Button's unitary-factor criterion \cite[Corollary~6]{Button2001}: an odd Markoff number \(M\) is unique if every coprime factorization \(M=rs\), \(rr^3\). In the notation of \eqref{eq:ustar}, this is equivalent to \[ u_*(M)^42\), and put \begin{equation} q=\frac{3M}{2},\qquad R=\sqrt{q^2-1},\qquad Q=q+R. \label{eq:qRQ} \end{equation} Thus \(Q^{-1}=q-R\), and if \(\theta=\log Q\), then \(q=\cosh\theta\), \(R=\sinh\theta\). Define \begin{equation} f(x)=\frac{M}{2R}(Q^x+Q^{-x}),\qquad g(x)=\frac{M}{2R}(Q^x-Q^{-x}). \label{eq:fg} \end{equation} Equivalently, \[ f(x)=\frac{M}{R}\cosh(x\theta),\qquad g(x)=\frac{M}{R}\sinh(x\theta). \] The curve parametrized below is precisely Tornero's fixed-\(M\) hyperbola \(H_M:X^2-3MXY+Y^2+M^2=0\) \cite[p.~596]{Tornero2008}. Tornero studies its integral points using the recurrence \(B_{j+2}=3MB_{j+1}-B_j\), whose characteristic roots are \(Q\) and \(Q^{-1}\). Here we use the corresponding continuous \(Q^x\) parametrization to place the two occurrences of an ordered collision on one real component and factor their four-point determinants. \begin{figure}[htbp] \centering \includegraphics[width=\textwidth]{Fig1_fg_log.pdf} \caption{The functions \(f\) and \(g\) for the Markoff maximum \(M=194\), which occurs in \((5,13,194)\). The horizontal axis is linear and the vertical axis is logarithmic. The shaded interval \(02\), then \(00\). Completing the square gives \[ (qX-Y)^2=R^2X^2-M^2. \] Define \(u=\operatorname{arsinh}((qX-Y)/M)\), where \(\operatorname{arsinh}\) is the inverse of \(\sinh\). Then \(\cosh u=RX/M\), since both sides are positive. Setting \(x=u/\theta\) gives \(X=f(x)\) and \(Y=qX-M\sinh u=f(1-x)\). This also proves that the parameter is unique. The function \(\cosh t\) is even and increases with \(|t|\). Thus \(f(x)\le f(1-x)\) is equivalent to \(|x|\le|1-x|\), or \(x\le1/2\). If \(x\le0\), then \[ f(1-x)\ge f(1)=\frac{3M^2}{2R}>M, \] because \(R<3M/2\). Such a point cannot have maximum \(M\). If \(x=1/2\), the two smaller integer entries agree. Pairwise coprimality, proved independently in Lemma~\ref{lem:prime-support}, makes both equal to \(1\). The equation then gives \(M=1\) or \(M=2\). Hence \(M>2\) requires \(00\) on this interval, while \(\frac{d}{dx}f(1-x)=-M\theta\sinh((1-x)\theta)/R<0\). These signs prove the monotonicity assertions. \end{proof} Let \((a,b,M)\) and \((a',b',M)\) form an ordered collision. After interchanging the two ordered occurrences if necessary, there are \begin{equation} 00\). In particular, \(B=br+as>0\) after the displacement notation introduced below, and the other signs follow from the identities. The strict inequality \(b2\) would give a common divisor. \section{Collision line, slope-scale factorization, and unitary allocation} \label{sec:unitary} The slope--scale formulas and unitary allocation in this section were previously circulated in the author's earlier slope--scale preprint (\href{https://doi.org/10.13140/RG.2.2.23525.56800}{https://doi.org/10.13140/RG.2.2.23525.56800}); they are reproduced here to make the quantitative argument self-contained. Put \begin{equation} r=a'-a>0,\qquad s=b-b'>0. \label{eq:rs} \end{equation} Then \begin{equation} C=as-br+rs,\qquad D=2bs-s^2-2ar-r^2. \label{eq:CD-rs} \end{equation} Combining these with \(D=3MC\) gives the collision line \begin{equation} (3Ms+2r)a-(3Mr+2s)b+(3Mrs+r^2+s^2)=0. \label{eq:collisionline} \end{equation} Set \begin{equation} P=3Mr+2s,\qquad S=3Ms+2r,\qquad d_0=\gcd(P,S),\qquad K=r^2+s^2+3Mrs. \label{eq:PSK} \end{equation} Equation \eqref{eq:collisionline} is equivalent to \[ S(2a+r)=P(2b-s). \] Because \(P/d_0\) and \(S/d_0\) are coprime, a positive integer \(j\) satisfies \begin{equation} 2a+r=\frac{jP}{d_0},\qquad 2b-s=\frac{jS}{d_0}. \label{eq:jparam} \end{equation} \begin{proposition}[Slope and scale identities] \label{prop:slopescale} Every ordered collision satisfies \begin{align} C&=\frac{j(s^2-r^2)}{d_0},& B&=\frac{jK}{d_0}, \label{eq:BCparam}\\ A&=\frac{M^2(s^2-r^2)}{K}, \label{eq:slopeA}\\ B^2&=\frac{K(K+4M^2)}{9M^2-4}. \label{eq:scaleB} \end{align} \end{proposition} \begin{proof} Solve \eqref{eq:jparam} explicitly: \[ a=\frac12\left(\frac{jP}{d_0}-r\right),\qquad b=\frac12\left(\frac{jS}{d_0}+s\right). \] Use \(a'=a+r\), \(b'=b-s\), and the definitions of \(B,C\). Substitution gives \[ C=as-br+rs=\frac{j(sP-rS)}{2d_0} =\frac{j(s^2-r^2)}{d_0}, \] \[ B=br+as=\frac{j(rS+sP)}{2d_0}=\frac{jK}{d_0}. \] Since \(C,j,d_0>0\), we have \(s^2-r^2>0\). Dividing \(M^2C=AB\) by the positive \(B\) proves \eqref{eq:slopeA}. The same substitution into \(A=b(b-s)-a(a+r)\) yields \[ A=\frac{(s^2-r^2)((9M^2-4)j^2-d_0^2)}{4d_0^2}. \] Equate this expression to \eqref{eq:slopeA} and cancel \(s^2-r^2>0\). The result is \[ ((9M^2-4)j^2-d_0^2)K=4M^2d_0^2. \] Rearranging and multiplying by \(K/d_0^2\), then using \(B=jK/d_0\), gives \eqref{eq:scaleB}. \end{proof} We now reduce the slope. Write \[ r=Gr_0,\qquad s=Gs_0,\qquad \gcd(r_0,s_0)=1, \] and define \begin{equation} T_0=s_0^2-r_0^2,\qquad K_0=r_0^2+s_0^2+3Mr_0s_0. \end{equation} Equation \eqref{eq:BCparam} shows that \(T_0>0\), and all quotients below are positive. Let \begin{equation} \Gamma=\gcd(K_0,T_0),\qquad K_0=L\Gamma,\qquad T_0=W\Gamma, \qquad \gcd(L,W)=1. \label{eq:LGammaW} \end{equation} Formula \eqref{eq:slopeA} becomes \begin{equation} A=\frac{M^2W}{L}. \label{eq:AL} \end{equation} Thus \(L\mid M^2\). Write \begin{equation} M^2=LN,\qquad A=NW. \label{eq:LN} \end{equation} \begin{lemma}[Scale divisor] \label{lem:scale-divisor} If \(M\) is odd, then \(L\mid B\). \end{lemma} \begin{proof} It is enough to compare the exponent of each prime in \(L\) and \(B\). Let \(p\mid L\). Because \(L\mid M^2\) and \(M\) is odd, \(p\) is an odd divisor of \(M\). Thus \(9M^2-4\equiv-4\not\equiv0\pmod p\). We have \[ K=G^2L\Gamma,\qquad K+4M^2=L(G^2\Gamma+4N). \] Both factors on the right of \((9M^2-4)B^2=K(K+4M^2)\) are divisible by \(L\). Write \(\ord_p(z)\) for the exponent of \(p\) in a positive integer \(z\). Comparing exponents gives \[ 2\ord_p(B)=\ord_p(K)+\ord_p(K+4M^2)\ge2\ord_p(L). \] Thus every prime occurs in \(B\) to at least its exponent in \(L\), which proves \(L\mid B\). \end{proof} We use two elementary facts about Markoff entries. \begin{lemma}[Coprimality and prime support] \label{lem:prime-support} The entries of a Markoff triple are pairwise coprime. Every odd prime dividing a Markoff number is \(1\pmod4\). In particular, \(3\) divides no Markoff number. \end{lemma} \begin{proof} We first prove coprimality by descent. If a prime divides two entries of a Markoff triple, the equation forces it to divide the third. Order the entries as \(x\le y\le z\). Regarding the equation as a quadratic in its last entry gives the other integer root \[ z'=3xy-z=\frac{x^2+y^2}{z}>0. \] For \(P(u)=u^2-3xyu+x^2+y^2\), \[ P(y)=x^2-y^2(3x-2)<0 \] unless \(x=y=1\). Since the leading coefficient is positive, \(P(y)<0\) places \(y\) strictly between its two roots. The root \(z\) is at least \(y\), so the other root satisfies \(01\) and positive integers \(H,W\) such that \begin{equation} M=FE,\qquad A=F^2W,\qquad B=E^2H,\qquad C=HW. \label{eq:unitarypacket} \end{equation} Each of \(F,E\) is a unitary divisor of \(M\). \end{theorem} \begin{proof} Consider a complete prime-power factor \(p^e\) of \(M\). Its square \(p^{2e}\) occurs in \(M^2=LN\). If a positive power of \(p\) occurred in both \(L\) and \(N\), then \(A=NW\) and \(L\mid B\) would give \(p\mid A,B\). Lemma~\ref{lem:no-split} excludes this. Thus the entire power \(p^{2e}\) belongs to one of \(L,N\). Doing this for every prime gives \[ L=E^2,\qquad N=F^2,\qquad M=FE,\qquad\gcd(F,E)=1. \] Both factors exceed one. If \(L=1\), then \(A=M^2W\), contradicting \(00\). Substitute these expressions in \(M^2C=AB\) and cancel \(M^2=F^2E^2\) to obtain \(C=HW\). The coprimality of \(F,E\) is exactly the assertion that they are complementary unitary divisors. \end{proof} \begin{remark} Proper coprime factorizations of a hypothetical nonunique maximum underlie Srinivasan's Theorem~5.1 and Button's Corollary~6 \cite{Srinivasan2009,Button2001}. Theorem~\ref{thm:allocation} records the exact allocation to \(A,B,C\) in the notation required for the defect packets. \end{remark} \section{Complementary hyperbolic factors and the defect packets} \label{sec:defects} The coupled defect equations below were also circulated in the author's earlier slope--scale preprint (\href{https://doi.org/10.13140/RG.2.2.23525.56800}{https://doi.org/10.13140/RG.2.2.23525.56800}); they are reproved here from the complementary four-point conics in the form required by the certified tiers. The same four hyperbolic points used to factor \(AB=M^2C\) produce the two equations on which all four odd tiers rest. Define \begin{equation} \widehat A=3Maa'-a'b-ab',\qquad \widehat B=3Mab'-aa'-bb'. \label{eq:AhatBhat} \end{equation} Along with \(\xi,\eta\) from \eqref{eq:xieta}, put \begin{equation} \tau=c+d,\qquad \upsilon=1-d+c. \end{equation} Then \(0<\tau,\upsilon<1\), \(\eta+\tau=1\), and \(\xi+\upsilon=1\). \begin{lemma}[Complementary four-point conics] \label{lem:fourpoint} One has \begin{equation} \widehat A=Mg(\tau),\qquad \widehat B=Mg(\upsilon), \label{eq:hatg} \end{equation} and, for \(00\). For the conic, the addition formula for \(\sinh(u+v)\) gives \[ \sinh^2u+\sinh^2v+2\cosh(u+v)\sinh u\sinh v =\sinh^2(u+v). \] Take \(u=x\theta\), \(v=(1-x)\theta\), so \(u+v=\theta\). Multiply by \(M^2/R^2\) and use \(\sinh\theta=R\), \(2\cosh\theta=3M\). This proves \eqref{eq:gconic}. The pairs \((\eta,\tau)\) and \((\xi,\upsilon)\) each sum to one. Substituting them and multiplying by \(M^2\) proves \eqref{eq:Aconic} and \eqref{eq:Bconic}. \end{proof} \begin{theorem}[Coupled defect packets] \label{thm:defect-packets} For every hypothetical odd collision there are positive integers \(n,m\) such that \begin{align} n^2+3MHn+H^2&=F^4, \label{eq:defectF}\\ m^2+3MWm+W^2&=E^4. \label{eq:defectE} \end{align} Moreover, \begin{equation} \gcd(F,H)=\gcd(H,n)=1,\qquad \gcd(E,W)=\gcd(W,m)=1. \label{eq:defectgcds} \end{equation} \end{theorem} \begin{proof} First we show that the complementary factors have the required square divisors. Let \(p^e\unitary F\), and suppose \(v=\ord_p(\widehat A)<2e\). In \eqref{eq:Aconic}, the term \(\widehat A^2\) has exponent \(2v\). The exponents of \(A^2\) and \(M^4\) are at least \(4e>2v\), and that of \(3MA\widehat A\) is at least \(3e+v>2v\). This is impossible: dividing the equation by \(p^{2v}\) and reducing modulo \(p\) would leave exactly one nonzero term. Hence \(\ord_p(\widehat A)\ge2e\). For every prime of \(F\), this proves \(F^2\mid\widehat A\). The identical argument with \(E,B,\widehat B\) proves \(E^2\mid\widehat B\). Write \[ \widehat A=F^2m,\qquad\widehat B=E^2n. \] Substitute these and \eqref{eq:unitarypacket} into the two conics. Dividing the first by \(F^4\) gives \eqref{eq:defectE}; dividing the second by \(E^4\) gives \eqref{eq:defectF}. The integers \(m,n\) are positive because the complementary factors are positive. If a prime divided both \(F\) and \(H\), it would divide both \(A=F^2W\) and \(B=E^2H\), contrary to Lemma~\ref{lem:no-split}. Thus \(\gcd(F,H)=1\); the same reasoning gives \(\gcd(E,W)=1\). A prime dividing \(H,n\) would, by \eqref{eq:defectF}, also divide \(F^4\), hence \(F\), which is impossible. This proves \(\gcd(H,n)=1\). Applying the same argument to the other equation proves \(\gcd(W,m)=1\). \end{proof} The quadratic equations also have a useful interpretation through conjugate numbers. In the ring of numbers of the form \(u+vQ_M\), with \(u,v\in\Z\), namely the real quadratic order \[ \mathcal O_M=\Z\left[\frac{3M+\sqrt{9M^2-4}}2\right], \] one has \[ \Nm(x+yQ_M)=x^2+3Mxy+y^2,\qquad Q_M=\frac{3M+\sqrt{9M^2-4}}2. \] Thus \eqref{eq:defectF}--\eqref{eq:defectE} are norm equations of fourth powers. Here the norm is the product obtained by replacing the square root by each of its two signs. This interpretation provides context; the bounds below use only the displayed integer equations. \section[The hand tier: 33 and 99]{The hand tier: \(33\) and \(99\)} \label{sec:hand} Abstract one defect equation as \begin{equation} x^2+y^2+3TOxy=T^4. \label{eq:genericdefect} \end{equation} For a collision packet, either \[ (T,O,x,y)=(F,E,H,n) \quad\text{or}\quad (T,O,x,y)=(E,F,W,m). \] By Theorem~\ref{thm:allocation}, Lemma~\ref{lem:prime-support}, and \eqref{eq:defectgcds}, the local block conditions are \begin{equation} T,O>1,\quad \gcd(T,O)=1,\quad 2\nmid TO,\quad 3\nmid TO, \quad p\mid TO\Longrightarrow p\equiv1\pmod4, \quad \gcd(x,y)=1. \label{eq:blockconditions} \end{equation} \begin{lemma}[The \(33\)-lemma] \label{lem:33} Every solution of \eqref{eq:genericdefect} satisfying \eqref{eq:blockconditions} has \begin{equation} xy\ge33. \end{equation} \end{lemma} \begin{proof} Reduce \eqref{eq:genericdefect} modulo \(3\). Since \(3\nmid T\), it gives \(x^2+y^2\equiv1\pmod3\). The only square residues are zero and one, so exactly one of \(x,y\) is divisible by \(3\). In particular, \(3\mid xy\). Reduction modulo \(T\) also gives \(T\mid x^2+y^2\). Suppose \(xy<33\), and use symmetry to take \(x\le y\). Listing the coprime factor pairs of the positive multiples of three below \(33\) gives \[ \begin{gathered} (1,3),(1,6),(1,9),(1,12),(1,15),(1,18),(1,21),(1,24),(1,27),(1,30),\\ (2,3),(2,9),(2,15),(3,4),(3,5),(3,7),(3,8),(3,10),(5,6). \end{gathered} \] Because \(T\) is odd, \(T^4\equiv1\pmod{16}\). Every integer whose prime divisors are \(1\pmod4\) is itself \(1\pmod4\), so \(TO\) must have residue \(1,5,9\), or \(13\) modulo \(16\). For each listed pair, test these four residues in the equation, and then list the divisors \(T>1\) of \(x^2+y^2\) having the required prime divisors. The surviving six choices are \[ \begin{array}{c@{\qquad}c@{\qquad}c} \{x,y\}&x^2+y^2&T\\\hline \{1,9\}&82&41\\ \{1,21\}&442&13,17,221\\ \{3,7\}&58&29\\ \{3,8\}&73&73. \end{array} \] For each choice, the equation determines \(O\): \begin{equation} O=\frac{T^3-(x^2+y^2)/T}{3xy}. \label{eq:solveO} \end{equation} In the same order, these values are \[ \frac{68919}{27},\quad\frac{2163}{63},\quad \frac{4887}{63},\quad\frac{10793859}{63},\quad \frac{24387}{63},\quad5403. \] The first five are not integers. The last is divisible by three, which the block conditions exclude. Thus every possible pair below \(33\) is rejected. \end{proof} \begin{theorem}[Hand quartic tier] \label{thm:99} Every hypothetical odd collision satisfies \begin{equation} Hn,Wm\ge33,\qquad F^4,E^4\ge99M+66. \label{eq:99tier} \end{equation} It also satisfies the strict cubic imbalance \begin{equation} E<\frac{F^3}{99},\qquad F<\frac{E^3}{99}. \label{eq:99imbalance} \end{equation} \end{theorem} \begin{proof} Each of the two collision packets satisfies the hypotheses of Lemma~\ref{lem:33}. Therefore \(Hn\ge33\) and \(Wm\ge33\). The inequality \((x-y)^2\ge0\) gives \(x^2+y^2\ge2xy\), so either packet yields \[ T^4=x^2+y^2+3Mxy\ge(3M+2)xy\ge33(3M+2). \] This proves both quartic inequalities. For the first cubic inequality, use the strictly positive square terms to get \[ F^4>3MHn\ge99M=99FE. \] Divide by \(99F>0\) to obtain \(E1 with gcd(T,O)=1 retain only admissible O \end{lstlisting} The submitted source uses explicit failure checks rather than Python \texttt{assert}, so optimization flags cannot disable verification. It deliberately searches the larger domain in which \(\gcd(x,y)=1\) is not imposed; eliminating every point in this superset proves the required claim. A second program, written separately and runnable on its own, orders the search by the product \(N=xy\), enumerates the factor pairs of \(N\), and reimplements every arithmetic routine without importing the primary program. The primary program's \texttt{--mode both} option is an internal product-order cross-check that reuses its arithmetic helpers; it is not the standalone verifier. \subsection[The local floor 120 and coefficient 360] {The local floor \(120\) and coefficient \(360\)} The exact search over \(33\le xy<120\) leaves no admissible row. Before the final prime-support test on \(O\), the sole row is \[ (x,y,T,O)=(1,33,109,13081),\qquad 13081=103\cdot127, \] and both prime factors are \(3\pmod4\). At the boundary, \begin{equation} (x,y,T,O)=(5,24,601,603005),\qquad xy=120, \label{eq:localboundary} \end{equation} satisfies the equation and all local block conditions. Hence \(120\) is sharp for this relaxation. \begin{proposition}[Local exact tier] \label{prop:360} Every hypothetical odd collision satisfies \begin{equation} Hn,Wm\ge120,\qquad F^4,E^4\ge360M+240. \label{eq:360tier} \end{equation} \end{proposition} \begin{proof} Lemma~\ref{lem:33} already excludes \(xy<33\). The finite search described above covers every pair with \(33\le xy<120\), every possible divisor \(T\), and the resulting forced value of \(O\). It leaves no admissible row, even though its search domain is larger than the collision domain. Thus each collision packet must have \(xy\ge120\). Apply this conclusion to \((x,y)=(H,n)\) and \((W,m)\). In each equation, \(x^2+y^2\ge2xy\) then gives \[ T^4\ge(3M+2)xy\ge120(3M+2)=360M+240. \] This is a computer-assisted step: the proved finite reduction and its exact exhaustive certificate together establish the product floor. \end{proof} \subsection[The companion-prime filter and coefficient 201411] {The companion-prime filter and coefficient \(201411\)} The next filter is an automatic property of a genuine Markoff entry. \begin{lemma}[Companion-prime necessity] \label{lem:companion} If \(M\) is a Markoff number, every odd prime divisor of \((3M-2)(3M+2)=9M^2-4\) is \(1\pmod4\). \end{lemma} \begin{proof} Rearranging the Markoff equation gives two sum-of-squares identities: \begin{equation} (b-a)^2+M^2=ab(3M-2),\qquad (a+b)^2+M^2=ab(3M+2). \label{eq:companionidentities} \end{equation} Let an odd prime \(p\) divide either companion \(3M-2\) or \(3M+2\). It cannot divide \(M\), since that would also force \(p\mid2\). Reduce the corresponding identity modulo \(p\) and divide by \(M^2\). It shows that \(-1\) is a square modulo \(p\). As in Lemma~\ref{lem:prime-support}, this forces \(p\equiv1\pmod4\). \end{proof} The intentionally enlarged enumeration for \(120\le xy<100000\) has nineteen rows, listed in Table~\ref{tab:survivors}. Eighteen satisfy \(\gcd(x,y)=1\); the row with \(xy=75000\) has \(\gcd(x,y)=25\) and is retained to make the superset search explicit. Of all nineteen rows, eighteen contain a displayed prime \(\ell\equiv3\pmod4\) dividing a companion \(3TO-2\) or \(3TO+2\). The sole companion-admissible row is \begin{equation} (x,y,T,O)=(23,2919,1145,7453),\qquad xy=67137. \label{eq:exceptionalrow} \end{equation} It proves sharpness at \(67137\) for the local-plus-companion relaxation. \begin{proposition}[Companion exact tier] \label{prop:201411} Every hypothetical odd collision satisfies \begin{equation} Hn,Wm\ge67137,\qquad F^4,E^4\ge201411M+134274. \label{eq:201411tier} \end{equation} \end{proposition} \begin{proof} For each packet, the product \(TO\) is the actual common maximum \(M\). Lemma~\ref{lem:companion} therefore supplies a necessary condition on both of its companions. The local tier already excludes products below \(120\). In the complete table for \(120\le xy<100000\), every row with \(xy<67137\) fails that necessary companion condition. Hence both packets satisfy \(xy\ge67137\). Using \(x^2+y^2\ge2xy\) once more gives \[ T^4\ge(3M+2)xy\ge67137(3M+2)=201411M+134274. \] The conclusion uses the exact finite enumeration; no assumption that an arbitrary surviving row is a Markoff collision is made. \end{proof} \subsection[Residual companion allocation and a crossed Pell-type equation] {Residual companion allocation and a crossed Pell-type equation} \label{sec:residual-companion} The slope and scale formulas also divide the companion discriminant \(\Delta_M\) into coprime factors. The word discriminant here refers to the square-root expression arising from the quadratic equation; we will use its explicit value. Put \[ \Delta_M=9M^2-4,\qquad P_0=3Mr_0+2s_0,\qquad S_0=3Ms_0+2r_0. \] \begin{proposition}[Residual companion allocation] \label{prop:residual-companion} For every hypothetical odd collision, \begin{equation} d_0=G\Gamma,\qquad H=jG, \label{eq:d0-H-residual} \end{equation} and \(\Gamma\) is a unitary divisor of \(\Delta_M\). More precisely, put \begin{equation} \Gamma^\ast=\frac{\Delta_M}{\Gamma}. \label{eq:Gamma-star-residual} \end{equation} Then \begin{equation} \gcd(\Gamma,\Gamma^\ast)=1,\qquad \Gamma^\ast j^2-\Gamma G^2=4F^2. \label{eq:residual-Pell} \end{equation} Furthermore, \begin{equation} \gcd(\Gamma,M)=\gcd(\Gamma,j)=\gcd(\Gamma^\ast,G)=1. \label{eq:residual-coprime} \end{equation} If \begin{equation} Z_F=2n+3MH-2F^2, \label{eq:ZF} \end{equation} then \begin{equation} Z_F=G^2\Gamma,\qquad G^2=\gcd(Z_F,H^2). \label{eq:ZF-extraction} \end{equation} Thus the residual data \(G,j,\Gamma,\Gamma^\ast\) are determined by the ordered packet \((F,E,H,n)\), with \(M=FE\). \end{proposition} \begin{proof} We first identify the common divisor in the slope formulas. Since \(P=GP_0\) and \(S=GS_0\), write \(d_0=G\delta\), where \(\delta=\gcd(P_0,S_0)\). At least one of \(P_0,S_0\) is odd, because \(M\) is odd and the coprime integers \(r_0,s_0\) cannot both be even. Thus \(\delta\) is odd. The identities \[ s_0P_0-r_0S_0=2T_0,\qquad s_0P_0=K_0+T_0,\qquad r_0S_0=K_0-T_0 \] show that \(\delta\mid T_0\), and then \(\delta\mid K_0\). Hence \(\delta\mid\Gamma\). For the reverse divisibility, \(K_0\) is odd for either possible parity pattern of \(r_0,s_0\), so \(\Gamma\) is odd. Also \(\gcd(\Gamma,r_0s_0)=1\): a prime dividing \(\Gamma\) and one slope coordinate would divide the other by \(T_0=s_0^2-r_0^2\). The last two displayed identities therefore allow cancellation of \(r_0,s_0\) modulo \(\Gamma\), giving \(\Gamma\mid P_0,S_0\). Thus \(\delta=\Gamma\) and \(d_0=G\Gamma\). Next, direct linear combinations give \begin{equation} 3MP_0-2S_0=\Delta_Mr_0,\qquad 3MS_0-2P_0=\Delta_Ms_0. \label{eq:companion-adjugate} \end{equation} Both left sides are divisible by \(\Gamma\). Since \(r_0,s_0\) are coprime, an integer combination of them is one, so \(\Gamma\mid\Delta_M\). As \(M\) is odd, \(\gcd(M,\Delta_M)=\gcd(M,4)=1\). In particular, \(\gcd(\Gamma,M)=1\). Use \(K=G^2E^2\Gamma\) and \(d_0=G\Gamma\) in the formula for \(B\): \[ B=\frac{jK}{d_0}=jGE^2=HE^2. \] Cancel \(E^2\) to obtain \(H=jG\). Substitute these same quantities into \eqref{eq:scaleB}. Cancellation gives \[ \Delta_Mj^2=\Gamma(G^2\Gamma+4F^2). \] Writing \(\Delta_M=\Gamma\Gamma^\ast\) and dividing by \(\Gamma\) proves \eqref{eq:residual-Pell}. If a prime divided both \(\Gamma,\Gamma^\ast\), that equation would force it to divide \(4F^2\). It is odd and coprime to \(F\mid M\), a contradiction. Hence the allocation is unitary. The same reduction modulo a common prime of \(\Gamma,j\), or of \(\Gamma^\ast,G\), proves the remaining coprimalities. Finally, complete the square in the first defect equation: \[ (2n+3MH)^2=4F^4+H^2\Delta_M. \] The right side exceeds \(4F^4\), so \(Z_F=2n+3MH-2F^2>0\), and \[ H^2\Delta_M=Z_F(Z_F+4F^2). \] The residual equation and \(H=jG\) give the same product as \[ (G^2\Gamma)(G^2\Gamma+4F^2). \] The function \(t(t+4F^2)\) increases strictly for \(t>0\). Its two positive inputs must therefore agree: \(Z_F=G^2\Gamma\). Since \(H^2=G^2j^2\) and \(\gcd(\Gamma,j)=1\), taking a greatest common divisor gives \(\gcd(Z_F,H^2)=G^2\). Thus the packet determines \(G\), then \(j=H/G\), then \(\Gamma=Z_F/G^2\), and finally \(\Gamma^\ast=\Delta_M/\Gamma\). \end{proof} \begin{remark}[Relation with the principal-genus obstruction] \label{rem:principal-genus} For every prime \(p\mid\Gamma\), equation \eqref{eq:residual-Pell} shows that \(\Gamma^\ast\) is a quadratic residue modulo \(p\); for every \(p\mid\Gamma^\ast\), it shows that \(-\Gamma\) is a quadratic residue modulo \(p\). By Lemma~\ref{lem:companion}, all these primes are \(1\pmod4\), so the second condition may equivalently be stated with \(\Gamma\) in place of \(-\Gamma\). These are the quadratic-residue conditions, written using Legendre symbols, occurring in Srinivasan's principal-genus method \cite[Theorem~5.1, Definition~3.1, and Theorems~5.4 and 6.1]{Srinivasan2009}; no priority is claimed for this residue obstruction. The additional ingredient below is its simultaneous coupling to the primitive collision slope. \end{remark} Factor the two coprime companions according to the residual allocation: \begin{equation} 3M-2=\alpha\bar\alpha,\qquad 3M+2=\beta\bar\beta,\qquad \Gamma=\alpha\beta,\qquad \Gamma^\ast=\bar\alpha\bar\beta, \label{eq:companion-four-blocks} \end{equation} The bars in this notation denote complementary integer factors, not complex conjugates. Here \(\alpha=\gcd(\Gamma,3M-2)\) and \(\beta=\gcd(\Gamma,3M+2)\). The allocation is unitary within each companion. \begin{theorem}[Crossed slope equation] \label{thm:crossed-slope} For every hypothetical odd collision there are positive integers \(\lambda,\mu\) such that \begin{align} r_0+s_0&=\alpha\lambda, &s_0-r_0&=\beta\mu, \label{eq:slope-companion-divisibility}\\ W&=\lambda\mu\beta\mu,\qquad \bar\beta\lambda>\bar\alpha\mu. \label{eq:crossed-chamber} \end{equation} Moreover, \begin{equation} 1<\Gamma<\Delta_M,\qquad \alpha<3M-2,\qquad \beta<3M+2. \label{eq:proper-companion-split} \end{equation} \end{theorem} \begin{proof} Put \(u=r_0+s_0\) and \(v=s_0-r_0\); both are positive. From \(\Gamma\mid P_0,S_0\), we get \[ \Gamma\mid S_0+P_0=(3M+2)u,\qquad \Gamma\mid S_0-P_0=(3M-2)v. \] The odd numbers \(3M-2\) and \(3M+2\) are coprime: a common divisor would divide four. Thus \(\alpha\mid3M-2\) is coprime to \(3M+2\), and \(\beta\mid3M+2\) is coprime to \(3M-2\). Cancellation gives \(\alpha\mid u\) and \(\beta\mid v\). Write \(u=\alpha\lambda\), \(v=\beta\mu\), with \(\lambda,\mu>0\). Since \(T_0=uv\) and \(\Gamma=\alpha\beta\), \[ W=T_0/\Gamma=\lambda\mu. \] Also \(A=F^2Wv\). The second follows from \[ 2P_0=(3M+2)u-(3M-2)v =\Gamma(\bar\beta\lambda-\bar\alpha\mu)>0. \] To prove that the companion split is proper, use the original ordering: \[ G(r_0+s_0)=r+s=(b-a)-(b'-a')5M^2\) for \(M>1\), a contradiction. Thus \(\Gamma>1\). Also \[ \alpha\le r_0+s_00\), and \(XY<7453^2\). If the row is the second packet, then \((E,F)=(1145,7453)\) and \(\{W,m\}=\{23,2919\}\). The slope reconstruction of Lemma~\ref{lem:second-packet-reconstruction} gives \[ \begin{array}{c@{\qquad}c@{\qquad}c@{\qquad}c} (W,m)&k&(r_0,s_0)&\Gamma\\\hline (23,2919)&417&(7,3144)&429769\\ (2919,23)&23&(1,57128)&1118057. \end{array} \] The corresponding complementary divisors \(\Gamma^\ast\) are \(1525037909\) and \(586208053\). Therefore the residual equation \eqref{eq:residual-Pell} is again \eqref{eq:exceptional-pell-1} or \eqref{eq:exceptional-pell-2}, now with \((X,Y)=(j,G)\). Since \(B=E^2H8N^2\). Since \(X\ge1\), \[ SY^2=RX^2-4N^2>\frac R2X^2. \] Multiply by \(X^2\) and use \(XY2\cdot429769\cdot7453^4, \] so \(X\le1148\). Reducing that equation modulo \(S=429769=137\cdot3137\), and inverting \(R\) modulo \(S\), gives \[ X^2\equiv275216\pmod{429769}. \] Its four roots between zero and \(429768\) are \[ 85941,\quad104763,\quad325006,\quad343828. \] None lies in \(1\le X\le1148\), so this case is impossible. For the second equation, \[ 586208053\cdot1853^4>2\cdot1118057\cdot7453^4, \] so \(X\le1852\). Reduction modulo \(1118057=137\cdot8161\) gives \[ X^2\equiv391856\pmod{1118057}. \] Its four roots in the standard residue interval are \[ 129197,\quad471959,\quad646098,\quad988860. \] Again none is small enough. Each root list is complete because each of its two distinct odd prime moduli has exactly two roots, and the Chinese remainder theorem combines their four choices into exactly four residues. The supplementary programs verify the prime factors, all four residues, and the finite bounds independently. \end{proof} \subsection[An independent continuant filter and coefficient 300000] {An independent canonical-continuant filter and coefficient \(300000\)} Proposition~\ref{prop:exceptional-residual-rejection} already eliminates the sole odd companion-admissible row below \(100000\). We retain the following different necessary test as an independent audit of that rejection and because the even \(100000\) tier uses it. We use only a necessary part of the classical Cohn--Christoffel description of a Markoff occurrence \cite{Cohn1955,Cohn1971,Reutenauer2026}; the precise finite continued-fraction form needed here is supplied by \cite[Theorem~4.1]{CanakciSchiffler2020}. \begin{lemma}[Canonical continuant necessity] \label{lem:continuant} For every occurrence of a Markoff number \(M>2\), there is an integer \(\rho\) satisfying \begin{equation} \rho^2\equiv-1\pmod M,\qquad \frac{M}{3}<\rho<\frac{M}{2}, \label{eq:rootwindow} \end{equation} such that every partial quotient of the canonical finite regular continued fraction of \(M/\rho\) (whose last partial quotient is greater than \(1\)) belongs to \(\{1,2\}\). \end{lemma} \begin{proof} The Markoff-tree construction assigns a reduced slope to each ordered occurrence with \(M>2\). Theorem~4.1 of \c{C}anak\c{c}{\i} and Schiffler \cite{CanakciSchiffler2020} gives that occurrence an even palindrome of continued-fraction digits \[ [a_r,\ldots,a_1,a_1,\ldots,a_r], \qquad a_i\in\{1,2\},\quad a_r=2. \] Call the digits \(c_1,\ldots,c_{2r}\). Each matrix associated with a digit is symmetric. Transposing their product reverses the digits, and the palindrome leaves the product unchanged. The continued-fraction matrix is therefore \begin{equation} \prod_{i=1}^{2r}\begin{pmatrix}c_i&1\\1&0\end{pmatrix} =\begin{pmatrix}M&\rho\\\rho&\sigma\end{pmatrix}. \label{eq:symmetriccontinuant} \end{equation} Each factor has determinant \(-1\). The number of factors is even, so \(M\sigma-\rho^2=1\), proving \(\rho^2\equiv-1\pmod M\). The usual matrix rule for a continued fraction gives \[ M/\rho=[c_1;c_2,\ldots,c_{2r}]. \] Its last digit is \(2>1\), so it is the canonical finite expansion. Its first digit is also two, and the tail \([c_2;\ldots,c_{2r}]\) is strictly greater than one. Hence \(22\) be an even Markoff number and write \(M=2M_1\). Then \begin{equation} M\equiv2\pmod{32},\qquad M_1\equiv1\pmod{16}. \label{eq:even-congruence} \end{equation} In every occurrence \((a,b,M)\), the entries \(a,b\) are \(1\pmod4\). \end{lemma} \begin{proof} Pairwise coprimality makes \(a,b\) odd. If \(4\mid M\), reducing the Markoff equation modulo four would give \(1+1\equiv0\), which is false. Thus \(M=2M_1\) with \(M_1\) odd. Every odd prime dividing an entry is \(1\pmod4\), by Lemma~\ref{lem:prime-support}. It follows that \(a,b,M_1\equiv1\pmod4\), including when one of them is one. Write \[ a=1+4u,\qquad b=1+4v,\qquad M_1=1+4w. \] Substitute these expressions in \(a^2+b^2+4M_1^2=6M_1ab\) and reduce modulo \(32\). The result is \[ 16\bigl((u^2-u)+(v^2-v)\bigr)-24w\equiv0\pmod{32}. \] Both \(u(u-1)\) and \(v(v-1)\) are even, so the first term vanishes modulo \(32\). Hence \(32\mid24w\), or \(4\mid w\). Therefore \(M_1=1+4w\equiv1\pmod{16}\), and doubling gives \(M\equiv2\pmod{32}\). \end{proof} For an even ordered collision, put \begin{equation} A_0=\frac A4,\quad B_0=\frac B4,\quad C_0=\frac C4,\quad \widehat A_0=\frac{\widehat A}{4},\quad \widehat B_0=\frac{\widehat B}{4}. \label{eq:even-normalized-variables} \end{equation} These are integers. Indeed, all four smaller entries are \(1\pmod4\), so the definitions of \(A,B,C\) give \(4\mid A,B,C\); and, since \(M\equiv2\pmod4\), \[ \widehat A=3Maa'-a'b-ab'\equiv2-1-1\equiv0\pmod4, \] with the same calculation for \(\widehat B\). Dividing the determinant identity and the complementary conics by \(16\) gives \begin{align} A_0B_0&=M_1^2C_0, \label{eq:even-ABC}\\ A_0^2+\widehat A_0^2+3MA_0\widehat A_0&=M_1^4, \label{eq:even-Aconic}\\ B_0^2+\widehat B_0^2+3MB_0\widehat B_0&=M_1^4. \label{eq:even-Bconic} \end{align} \begin{theorem}[Normalized even collision allocation] \label{thm:even-allocation} If an even Markoff number \(M>2\) admits an ordered collision, then there are coprime proper unitary factors \(F,E>1\) of \(M_1=M/2\) and positive integers \(H,W,n,m\) such that \begin{equation} M_1=FE,\qquad A=4F^2W,\quad B=4E^2H,\quad C=4HW, \label{eq:even-unitary-packet} \end{equation} and \begin{equation} \widehat A=4F^2m,\qquad \widehat B=4E^2n. \label{eq:even-hat-packet} \end{equation} Moreover, \begin{align} n^2+3MHn+H^2&=F^4, \label{eq:even-defectF}\\ m^2+3MWm+W^2&=E^4, \label{eq:even-defectE} \end{align} and \begin{equation} \gcd(F,H)=\gcd(H,n)=1,\qquad \gcd(E,W)=\gcd(W,m)=1. \label{eq:even-defect-gcds} \end{equation} \end{theorem} \begin{proof} We allocate each complete prime-power factor of \(M_1\) to one side of the normalized identity. Let \(p^e\unitary M_1\). The prime \(p\) is odd. Lemma~\ref{lem:no-split} says it cannot divide both \(A_0\) and \(B_0\). Comparing prime exponents in \eqref{eq:even-ABC} gives \[ \ord_p(A_0)+\ord_p(B_0)=2e+\ord_p(C_0). \] One exponent on the left is zero, so the other is at least \(2e\). Allocate \(p^e\) to \(F\) if its square divides \(A_0\), and to \(E\) if its square divides \(B_0\). This gives \[ M_1=FE,\quad\gcd(F,E)=1,\quad A_0=F^2W,\quad B_0=E^2H, \] and the normalized identity then gives \(C_0=HW\). The factors are proper. If \(F=1\), then \(M_1^2=E^2\mid B_0\), whereas \(01\) are coprime, neither is divisible by \(3\), every prime dividing \(TO\) is \(1\pmod4\), and \(\gcd(x,y)=1\). \begin{lemma}[The sharp local floor \(72\)] \label{lem:even72} Every solution of \eqref{eq:even-generic-defect} satisfying these block conditions has \(xy\ge72\). The value \(72\) is sharp for this local relaxation. \end{lemma} \begin{proof} Reduction modulo three shows that exactly one of \(x,y\) is divisible by three. Since \(T^4\equiv1\pmod{16}\) and \(TO\equiv1\pmod{16}\), reduction modulo \(16\) gives \begin{equation} x^2+6xy+y^2\equiv1\pmod{16}. \label{eq:even-residue-packet} \end{equation} Take \(x\le y\). Under \(xy<72\), the coprime pairs satisfying these two congruence conditions are exactly \[ \begin{gathered} (1,18),(1,24),(1,42),(1,48),(1,66),\\ (2,3),(2,9),(2,27),(2,33),\\ (3,4),(3,10),(3,20),(5,6),(5,12),(6,7). \end{gathered} \] For each pair, the equation forces \(T\mid x^2+y^2\) and \begin{equation} O=\frac{T^3-(x^2+y^2)/T}{6xy}. \label{eq:even-solveO} \end{equation} List the divisors \(T>1\) having only prime divisors \(1\pmod4\), and substitute each in this formula. The only integer values \(O>1\) are the following; every one fails a block condition. \[ \begin{array}{@{}r r r r l@{}} \toprule x&y&T&O&\text{obstruction}\\\midrule 1&18&325&317853&3\mid O\\ 1&24&577&1334028&2\mid O\\ 1&42&1765&21818937&3\mid O\\ 1&48&2305&42522648&2\mid O\\ 1&66&4357&208865877&3\mid O\\ 2&3&13&61&TO\equiv9\pmod{16}\\ 3&4&25&217&7\mid O\\ 5&6&61&1261&TO\equiv9\pmod{16}\\ 6&7&85&2437&TO\equiv9\pmod{16}\\\bottomrule \end{array} \] Thus \(xy<72\) is impossible. To check sharpness for the stated local conditions, take \[ (x,y,T,O)=(8,9,145,7057),\qquad145=5\cdot29. \] The number \(7057\) is prime and \(1\pmod4\); the two blocks are coprime and have product \(1\pmod{16}\). Moreover, \[ 8^2+9^2=145,\qquad1+432\cdot7057=145^3. \] Multiplying the second identity by \(145\) verifies the defect equation. This admissible local row has product \(72\), proving sharpness for the relaxation. \end{proof} The separate even certificate enumerates the intentionally enlarged domain \(72\le xy<100000\) twice, first by pairs and then by products. It retains \(65\) local rows, of which \(63\) satisfy \(\gcd(x,y)=1\). Applying Lemma~\ref{lem:companion} to the actual candidate maximum \(M=2TO\) leaves exactly four primitive rows: \begin{equation} \begin{array}{@{}r r r r r r@{}} \toprule xy&x&y&T&O&M=2TO\\ \midrule 1014&3&338&61&37&4514\\ 7140&84&85&14281&67987081&1941847007522\\ 74256&272&273&148513&7352086561&2183760862867586\\ 96624&9&10736&613&397&486722\\ \bottomrule \end{array} \label{eq:even-companion-rows} \end{equation} Since \(M=2M_1\), one has \(3M\mathbin{\pm}2=2(3M_1\mathbin{\pm}1)\); hence the odd-prime companion test is equivalently applied to \(3M_1\mathbin{\pm}1\). The first row proves sharpness for that relaxation: with \(M_1=2257\), \[ 3M_1-1=2\cdot5\cdot677,\qquad 3M_1+1=2^2\cdot1693, \] so all odd companion primes are \(1\pmod4\). The canonical test of Lemma~\ref{lem:continuant} uses the even maximum \(M=2TO\), not the half-maximum. For \(M=4514\), the interval roots are \(1597,1819\), with continued fractions \[ [2,1,4,1,3,3,1,4,1,2],\qquad [2,2,13,13,2,2]. \] The middle two rows in \eqref{eq:even-companion-rows} have no root of \(-1\) in \((M/3,M/2)\). For \(M=486722\), the interval roots are \(191291,193673\), with continued fractions \[ [2,1,1,5,7,1,2,2,1,7,5,1,1,2], \quad [2,1,1,18,1,1,3,3,1,1,18,1,1,2]. \] Every displayed expansion has a digit greater than \(2\). Thus all four rows fail the canonical-continuant necessity. \begin{theorem}[Certified even tiers] \label{thm:even-tiers} Every hypothetical even collision satisfies the hand tier \begin{equation} Hn,Wm\ge72,\qquad F^4,E^4\ge216M+144. \label{eq:even216} \end{equation} The exact companion-prime certificate strengthens this to \begin{equation} Hn,Wm\ge1014,\qquad F^4,E^4\ge3042M+2028, \label{eq:even3042} \end{equation} and the canonical-continuant certificate gives \begin{equation} Hn,Wm\ge100000,\qquad F^4,E^4\ge300000M+200000. \label{eq:even300000} \end{equation} The floors \(72\) and \(1014\) are sharp for their stated relaxations; \(100000\) is a certified cutoff, not a sharpness claim. \end{theorem} \begin{proof} Each packet in Theorem~\ref{thm:even-allocation} has \((T,O)=(F,E)\) or \((E,F)\), with actual maximum \(M=2TO\). Lemma~\ref{lem:even72} gives the hand floor \(xy\ge72\). In the complete search above this floor and below \(100000\), the companion filter leaves exactly the four displayed rows, whose smallest product is \(1014\). Thus the necessary companion condition improves the floor to \(1014\). The continued-fraction condition must be applied modulo the full maximum \(M\). Two of those four rows have no root in its required interval, and the other two have only the displayed continued fractions, each containing a digit above two. Thus none is a Markoff occurrence, and the final floor is \(100000\). For each established floor \(B_0\), either packet satisfies \[ T^4=x^2+y^2+3Mxy\ge(3M+2)xy\ge B_0(3M+2). \] Substituting \(B_0=72,1014,100000\) gives the three claimed quartic bounds. The rows at products \(72\) and \(1014\) prove sharpness only for their specified relaxations; no row establishes sharpness at \(100000\). \end{proof} \begin{remark}[Provenance of the even tiers] The \(72\) tier is proved entirely in the article. The \(1014\) and \(100000\) tiers are computer-assisted theorems certified by two separately implemented exact-integer traversals supplied in the supplementary archive. The companion and canonical filters are necessary for every actual Markoff occurrence by Lemmas~\ref{lem:companion} and \ref{lem:continuant}. \end{remark} \section{Intrinsic unitary-gap and tail consequences} \label{sec:consequences} For the odd case, it is useful to state all four tiers at once. Let \(B_0\) be one of \[ 33,\quad120,\quad67137,\quad100000, \] and put \(\mathcal C=3B_0\). The corresponding tier says \begin{equation} F^4,E^4\ge \mathcal C M+\frac{2\mathcal C}{3}. \label{eq:Ctier} \end{equation} \begin{theorem}[Intrinsic unitary-gap criterion] \label{thm:unitarygap} Let \(M\) be an odd Markoff number. At any certified coefficient \(\mathcal C\in\{99,360,201411,300000\}\), \begin{equation} u_*(M)^4<\mathcal C M+\frac{2\mathcal C}{3} \quad\Longrightarrow\quad M\text{ is unique}. \label{eq:generalunitarygap} \end{equation} \end{theorem} \begin{proof} The maximum \(M=1\) has the single ordered occurrence \((1,1,1)\). For \(M>1\), suppose a collision existed and set \(d=\min(F,E)\) in Theorem~\ref{thm:allocation}. This is a unitary divisor of \(M\) and satisfies \(d\le\sqrt M\), so it belongs to the set defining \(u_*(M)\). Hence \(d\le u_*(M)\). The chosen collision tier gives \(d^4\ge\mathcal C M+2\mathcal C/3\). Together with the assumed inequality, this would give \[ \mathcal C M+2\mathcal C/3\le d^4\le u_*(M)^4 <\mathcal C M+2\mathcal C/3, \] a contradiction. Thus no collision exists. \end{proof} \begin{theorem}[Even intrinsic unitary-gap criterion] \label{thm:even-unitarygap} Let \(M>2\) be an even Markoff number and put \(M_1=M/2\). At any certified even coefficient \[ \mathcal C\in\{216,3042,300000\}, \] with respectively hand, companion-certificate, and continuant-certificate provenance, \begin{equation} u_*(M_1)^4<\mathcal C M+\frac{2\mathcal C}{3} \quad\Longrightarrow\quad M\text{ is unique}. \label{eq:even-generalunitarygap} \end{equation} \end{theorem} \begin{proof} Suppose a collision existed. Theorem~\ref{thm:even-allocation} gives a unitary split \(M_1=FE\). Its smaller factor \(d=\min(F,E)\) is a unitary divisor at most \(\sqrt{M_1}\), so \(d\le u_*(M_1)\). The chosen tier in Theorem~\ref{thm:even-tiers} bounds each factor's fourth power below by \(\mathcal C M+2\mathcal C/3\). Therefore \[ \mathcal C M+2\mathcal C/3\le d^4\le u_*(M_1)^4 <\mathcal C M+2\mathcal C/3, \] which is impossible. Notice that the divisor function uses \(M_1\), while the quartic bound uses the full maximum \(M\). \end{proof} For a composite tail, define \begin{equation} \mathcal G(N)=\frac{N}{u_*(N)^4} =\min_{\substack{UV=N\\\gcd(U,V)=1}} \frac{\max(U,V)}{\min(U,V)^3}. \label{eq:Gfunctional} \end{equation} The equality follows by putting \(d=\min(U,V)\): the displayed ratio is \(N/d^4\), minimized by the largest unitary \(d\le\sqrt N\). \begin{theorem}[Composite unitary tail] \label{thm:compositetail} Let \(M=kN\) be an odd Markoff number with \(\gcd(k,N)=1\). At any certified coefficient \(\mathcal C\), the condition \begin{equation} k^3u_*(N)^4\le\mathcal C N, \quad\text{equivalently}\quad k^3\le\mathcal C\mathcal G(N), \label{eq:compositetail} \end{equation} guarantees that \(M\) is unique. \end{theorem} \begin{proof} Assume there is a collision, with unitary split \(M=FE\). Because \(\gcd(k,N)=1\), every prime-power block of \(N\) lies wholly in one of these two factors. Write its resulting split as \(N=XY\), with \(\gcd(X,Y)=1\), and choose the names so that \(X\le Y\). Then \(X\le u_*(N)\). The collision factor containing \(X\) has the form \(dX\), where \(d\) is a unitary divisor of \(k\). It is therefore at most \(k u_*(N)\). Apply the collision tier to this factor and the assumed bound to its upper estimate: \[ \mathcal C kN+\frac{2\mathcal C}{3} \le(dX)^4\le k^4u_*(N)^4\le\mathcal C kN. \] The positive term \(2\mathcal C/3\) makes this chain impossible. It also explains why a non-strict inequality is allowed in the hypothesis. \end{proof} Taking \(N=q^e\) makes \(u_*(N)=1\), proving Corollary~\ref{cor:tail-intro}. More generally, if \(N=XY\) has exactly two coprime primary blocks with \(X