\documentclass[11pt]{amsart} \usepackage[T1]{fontenc} \usepackage{amsmath,amssymb,amsthm,mathtools} \usepackage{microtype} \usepackage{xcolor} \usepackage{booktabs,longtable,array} \usepackage{enumitem} \usepackage[a4paper,top=24mm,bottom=25mm,left=26mm,right=26mm]{geometry} \usepackage[colorlinks=true,linkcolor=blue!45!black,citecolor=blue!45!black,urlcolor=blue!55!black,bookmarks=true,bookmarksnumbered=true,unicode=true]{hyperref} \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{remark}[theorem]{Remark} \newtheorem{conjecture}[theorem]{Conjecture} \DeclareMathOperator{\rad}{rad} \DeclareMathOperator{\Norm}{N} \newcommand{\N}{\mathbb{N}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\Q}{\mathbb{Q}} \newcommand{\eps}{\varepsilon} \newcommand{\pos}[1]{(#1)_+} \newcommand{\R}{\{1,\dots,6\}} \newcommand{\FullOC}{\mathrm{FullOC}} \newcommand{\old}{\mathrm{old}} \newcommand{\sset}{\mathcal O} \newcommand{\wto}{\lesssim} \title[Complete modulus locking for the abc exceptional set]{Complete modulus locking in the exceptional set\\of the abc conjecture} \author{Chunping Zeng} \date{19 November 2026} \subjclass[2020]{Primary 11D45; Secondary 11N36, 11D75, 11Y16} \keywords{abc conjecture, exceptional set, radical, collision inequality, exact rational certificate, branch and bound} \hypersetup{ pdftitle={Complete modulus locking in the exceptional set of the abc conjecture}, pdfauthor={Chunping Zeng}, pdfsubject={A 0.584 exceptional-set exponent with exact rational certification}, pdfkeywords={abc conjecture, radical, exceptional set, modulus locking, rational certificate} } \begin{document} \begin{abstract} Let $N_\lambda(X)$ count primitive positive triples $a+b=c\le X$ with $\rad(abc)0$, define \[ N_\lambda(X)=\#\left\{(a,b,c)\in\N^3:\ a+b=c\le X,\ \gcd(a,b,c)=1,\ \rad(abc)0$, \begin{equation}\label{eq:main-count} N_\lambda(X)\ll_{\lambda,\eps}X^{U+\eps}. \end{equation} In particular, \[ N_\lambda(X)\ll_{\lambda,\eps}X^{73/125+\eps} =X^{0.584+\eps}, \qquad \frac{73}{125}-U=\frac1{2500001500}. \] \end{theorem} The finite optimization step is not inferred from a floating-point optimum. It is proved by a complete disjunctive tree and positive rational multipliers, independently reconstructed and checked in exact arithmetic. The passage from the exponent program to \eqref{eq:main-count}, including all logarithmic, divisor and decomposition losses, is given in Section~\ref{sec:counting}. For orientation, Table~\ref{tab:comparison} lists the stated exponents in the principal sources used here. Decimal values are rounded only for display. No claim of a comprehensive literature optimum is made. \begin{table}[htbp] \centering \caption{Selected general bounds for $N_\lambda(X)$ at the budget endpoint $\lambda=1$; the BBLT $3/5$ bound is stated for $\lambda<1$.}\label{tab:comparison} \begin{tabular}{@{}lll@{}} \toprule Source & Certified exponent & Approximation\\ \midrule BBLT \cite{BBLT} & $3/5$ & $0.600000000$\\ Li, collision program \cite{Li17} & $54/91$ & $0.593406593$\\ Li, squares and twisted energy \cite{Li28} & $589903/1000008$ & $0.589898281$\\ Theorem~\ref{thm:main} & $\mathbf{11680007/20000012}$ & $\mathbf{0.5839999996}$\\ \bottomrule \end{tabular} \end{table} Relative to the exact exponent of \cite{Li28}, the gain is \[ \frac{589903}{1000008}-U =\frac{29491659695}{5000043000024} =0.0058982812137\ldots . \] \subsection{What is new} The argument has five connected outputs. \begin{enumerate}[label=(\arabic*),leftmargin=2.2em] \item \textbf{All fixed degrees through $6$.} The grouped difference estimate is used with $e=2,3,4,5,6$. The underlying uniform count for $aU^e-bV^e=m$ in arbitrary fixed degree is already proved in \cite[Appendix~B]{Li28}; we reproduce its proof and do not claim a new general-degree counting theorem. \item \textbf{Complete whole-value modulus locking.} For fixed remaining collision data, all third-letter sets $O_C\subseteq\R$ are considered. A one-Lipschitz deletion argument proves index $1$ redundant, leaving the $32$ subsets of $\{2,3,4,5,6\}$. This completeness statement is specific to the whole-value lock and the savings $\delta_0,\delta_2,\dots,\delta_6$; it is not asserted for an arbitrary coordinate-lock program. \item \textbf{A new exact upper certificate.} The resulting finite implication has $65{,}576$ leaves, $4{,}591$ geometric nodes and $7{,}623$ collision nodes. Its maximum leaf bound is exactly $U$. \item \textbf{Strict separation from the old family.} The old modulus family has an exact feasible point with $D=403/690$. At that point a new FullOC instance gives $53/920$, there is a fixed $d=d(\eps_0)\ge6$ such that the restricted count is at most $X^{\eps_0}$ times a constant times the largest relevant box count. In a box, small coefficients are fixed and the three letters have the form \[ \mathsf a_0\prod_{i=1}^d x_{a,i}^i,\qquad \mathsf b_0\prod_{i=1}^d x_{b,i}^i,\qquad \mathsf c_0\prod_{i=1}^d x_{c,i}^i, \qquad x_{s,i}\sim P_{s,i}. \] The equation is \begin{equation}\label{eq:box-equation} \mathsf a_0\prod_i x_{a,i}^i+ \mathsf b_0\prod_i x_{b,i}^i= \mathsf c_0\prod_i x_{c,i}^i. \end{equation} The small coefficients are pairwise coprime and at most $X^{\eps_0}$. We may restrict to solutions whose three displayed letter values are pairwise coprime: in a primitive triple $a+b=c$, a common divisor of any two letters divides the third. Discarding further within-letter restrictions can only enlarge an upper bound. The dyadic parameters obey the radical and height budgets \begin{equation}\label{eq:dyadic-budgets} 2^{-d-1}X^\mu\le\prod_{s,i}P_{s,i}\le X^{\mu+3\eps_0/4}, \qquad \prod_iP_{s,i}^i\le X, \end{equation} and the original third letter satisfies $\prod_iP_{c,i}^i\ge X^{1-\eps_0/2}$. Here $0<\mu\le\lambda$ is the radical budget of the restricted count. All later estimates are uniform over boxes and frozen coefficients once $d$ is fixed. Set \begin{equation}\label{eq:exponent-definitions} P_{s,i}=X^{\alpha_{s,i}},\qquad \alpha_s=\sum_{i=1}^d\alpha_{s,i},\qquad L=\alpha_a+\alpha_b+\alpha_c, \end{equation} and write the nonzero box count as $X^D$. The quantities $\alpha_{s,i}$, $\alpha_s$ and $D$ depend on the box, but we suppress this in the notation. Errors below are $O(\eps_0)+o(1)$ as $X\to\infty$, with $d$ fixed. \subsection{The four inherited estimates} The limiting exponent system uses only coordinates $1\le i\le6$ explicitly; the remaining tail is retained through $\alpha_s-\sum_{i\le6}\alpha_{s,i}$. All variables are nonnegative. \begin{proposition}[Inherited height and radical rows]\label{prop:base-rows} Every anatomic box satisfies, up to the stated limiting errors, \begin{equation}\label{eq:height-radical} L\le\lambda,\qquad \sum_{i=1}^6\alpha_{s,i}\le\alpha_s,\qquad 7\alpha_s-\sum_{i=1}^6(7-i)\alpha_{s,i}\le1 \quad(s\in\{a,b,c\}). \end{equation} The last row follows from $\sum_{i\le d}i\alpha_{s,i}\le1$ by using $i\ge7$ in the tail. \end{proposition} \begin{proposition}[Inherited pair count]\label{prop:pair-count} One has \begin{equation}\label{eq:pairs} D\le\alpha_a+\alpha_b,\qquad D\le\alpha_a+\alpha_c,\qquad D\le\alpha_b+\alpha_c. \end{equation} \end{proposition} \begin{proposition}[Inherited Fourier estimate]\label{prop:fourier} Define \begin{equation}\label{eq:fourier-max} F_s=\max\{\alpha_{s,2}+\alpha_{s,4}+\alpha_{s,6},\ \alpha_{s,3}+\alpha_{s,6},\ \alpha_{s,5}\}. \end{equation} Then \begin{equation}\label{eq:fourier} F_a+F_b+F_c\le4L-6D. \end{equation} \end{proposition} \begin{proposition}[Inherited geometric estimate]\label{prop:geometry} For arbitrary $I_a,I_b,I_c\subseteq\R$, put \[ \Sigma_0=\sum_{s\in\{a,b,c\}}\sum_{i\in I_s}\alpha_{s,i}, \qquad \Sigma_1=\sum_{s\in\{a,b,c\}}\sum_{i\in I_s}(i-1)\alpha_{s,i}. \] Then \begin{equation}\label{eq:geometry} D\le\max\{L-\Sigma_0,\ L-1+\Sigma_1\}. \end{equation} \end{proposition} For proofs of \eqref{eq:pairs}--\eqref{eq:geometry}, see \cite[Propositions~2.3, 3.1 and 4.1]{BBLT}. In the last two estimates the numerator initially contains the actual product $\prod_{s,i}P_{s,i}$, before any replacement by the radical budget. This gives $L$, rather than merely $\lambda$, in \eqref{eq:fourier} and \eqref{eq:geometry}. The factor $X^{-1+O(\eps_0)}$ in the second geometric alternative comes from the height of the original third letter and is independent of the radical budget. Thus $\lambda$ enters these limiting rows only through $L\le\lambda$. For computation, each $F_s$ is a nonnegative auxiliary variable bounded below by the three expressions in \eqref{eq:fourier-max}. The complete inequality system is invariant under permutations of the letters. We may therefore impose the symmetry-breaking order \begin{equation}\label{eq:symmetry} \alpha_{a,4}+2\alpha_{a,5}\ge \alpha_{b,4}+2\alpha_{b,5}\ge \alpha_{c,4}+2\alpha_{c,5}. \end{equation} After relabelling, every feasible point has a representative satisfying \eqref{eq:symmetry}. The geometric estimate remains valid after such a permutation: its final displayed expression is symmetric, although its proof uses the height of the original third letter. \section{Uniform power differences and twisted energy}\label{sec:arithmetic} The two arithmetic lemmas in this section are inherited from \cite[Sections~3.1--3.2 and Appendix~B]{Li28}. Their proofs are included because the new certificate uses them in degrees and modulus configurations not exhausted there. \subsection{Weighted differences of arbitrary fixed degree} \begin{lemma}[Inherited uniform count; proof reproduced]\label{lem:general-degree} Fix an integer $e\ge2$ and positive real numbers $C,\eta$. For $T\ge2$, positive integers $\mathsf a,\mathsf b$ and a nonzero integer $m$, the number of positive pairs $(U,V)$ satisfying \begin{equation}\label{eq:weighted-thue} \mathsf aU^e-\mathsf bV^e=m, \qquad \mathsf a,\mathsf b,U,V,|m|\le T^C \end{equation} is $O_{e,C,\eta}(T^\eta)$, uniformly in $\mathsf a,\mathsf b,m$. No coprimality hypothesis is required. \end{lemma} \begin{proof} Multiply by $\mathsf a^{e-1}$ and put \[ t=\mathsf a^{e-1}\mathsf b,\qquad H=\mathsf a^{e-1}m, \qquad \theta=t^{1/e}>0,\qquad K=\Q(\theta),\qquad n=[K:\Q]\le e. \] Then $\theta$ is an algebraic integer and \[ (\mathsf aU)^e-tV^e=H\ne0. \] Let $f$ be the monic minimal polynomial of $\theta$. Since $f\in\Z[Z]$, monic division in $\Z[Z]$ gives \[ Z^e-t=f(Z)g(Z),\qquad g\in\Z[Z],\qquad \deg g=e-n. \] For the algebraic integer $\alpha=\mathsf aU-\theta V$, define \[ z:=\Norm_{K/\Q}(\alpha)=V^nf(\mathsf aU/V). \] Evaluating the factorization of $Z^e-t$ yields \begin{equation}\label{eq:norm-factor} H=z\,V^{e-n}g(\mathsf aU/V). \end{equation} Both factors on the right are integers. Since $H\ne0$, the norm $z$ is a nonzero divisor of $H$. All quantities in \eqref{eq:weighted-thue} have polynomial size in $T$. Hence there is $C_1=C_1(e,C)$ such that every embedding $\sigma:K\hookrightarrow\mathbb C$ satisfies $|\sigma(\alpha)|\le T^{C_1}$. The product of these absolute values is the nonzero integer $|z|$, so \begin{equation}\label{eq:embedding-box} T^{-C_1(n-1)}\le|\sigma(\alpha)|\le T^{C_1}. \end{equation} Moreover, the integral ideal $(\alpha)$ has norm $|z|$ dividing $|H|$. We count possible ideals uniformly in $K$. Above a rational prime $p$ there are at most $e$ prime ideals. If $p^h\Vert H$, an ideal whose norm divides $H$ has exponent at most $h$ at each of these primes. Unique factorization of ideals therefore gives at most \begin{equation}\label{eq:ideal-count} \prod_{p^h\Vert H}(h+1)^e=\tau(|H|)^e \end{equation} integral ideals whose norms divide $|H|$. This bound has no discriminant-dependent constant. Since $|H|\le T^{Ce}$, the divisor bound makes \eqref{eq:ideal-count} at most $O_{e,C,\eta}(T^{\eta/2})$. Fix one generator $\alpha_0$ arising from a counted solution for each principal ideal that occurs. Every other counted generator of that ideal is $\alpha=\epsilon\alpha_0$ for a unit $\epsilon$. By \eqref{eq:embedding-box}, the vector $(\log|\sigma(\epsilon)|)_\sigma$ lies in a cube of side $O_{e,C}(\log T)$. We need a separation constant uniform over fields of degree at most $e$. Algebraic integers of degree at most $e$ whose conjugates all have modulus at most $\exp(1)$ form a finite set: their monic minimal polynomials have integer coefficients bounded in terms of $e$. If all conjugates of an algebraic integer have modulus $1$, all its positive powers belong to this same finite set, so it is a root of unity. Consequently there is $c_e>0$ such that every non-torsion unit of degree at most $e$ has \[ \max_\sigma\left|\log|\sigma(\epsilon)|\right|\ge c_e. \] Outside the finite set a conjugate already has modulus greater than $\exp(1)$; inside it, take the least positive distance. The number of roots of unity of degree at most $e$ is bounded by the same finiteness argument. Apply this separation to the quotient of two units occurring in $K$. Distinct unit logarithmic vectors are at distance at least $c_e$ in the supremum norm. Thus cubes of side $c_e/2$ contain at most one vector, and each vector has a bounded number of unit preimages. One ideal therefore contributes $O_{e,C}((1+\log T)^e)$ generators. If $n\ge2$, then $\theta$ is irrational and $(U,V)\mapsto\mathsf aU-\theta V$ is injective. The ideal and generator counts now prove the lemma after absorbing the logarithm into $T^{\eta/2}$. Finally, if $n=1$, then $\theta$ is a positive integer. The nonzero integer $s=\mathsf aU-\theta V$ divides $H$. For each signed divisor $s$, substitution gives \[ (\theta V+s)^e-(\theta V)^e=H. \] This is a nonzero polynomial equation of degree $e-1$, with leading coefficient $e\theta^{e-1}s$. It has at most $e-1$ roots, each determining $U$. The divisor bound for $H$ completes the proof. \end{proof} \begin{corollary}[Grouped values]\label{cor:grouped} Let $\varnothing\ne I\subseteq\R$ and $x_i,x_i'\sim X^{\alpha_i}$ for $i\in I$. For fixed nonzero $m$ of polynomial size in $X$, the number of pairs of values \[ v=\prod_{i\in I}x_i^i,\qquad v'=\prod_{i\in I}(x_i')^i, \qquad v-v'=m \] is at most $X^{\delta+o(1)}$, where $\delta$ may be the minimum of \begin{equation}\label{eq:delta-grouped} \delta_0=\sum_{i\in I}\alpha_i, \qquad \delta_e=2\sum_{\substack{i\in I\\e\nmid i}}\alpha_i \quad(e\ge2\ \text{fixed}). \end{equation} \end{corollary} \begin{proof} The bound $\delta_0$ counts $v$, after which $v'$ is determined. For $\delta_e$, freeze both copies of every coordinate with $e\nmid i$. The remaining products have the form $U^e$ and $V^e$, so the equation becomes $\mathsf aU^e-\mathsf bV^e=m$. Lemma~\ref{lem:general-degree} applies uniformly, and the frozen choices cost $\delta_e$. A value of $U$ has at most $\tau(U)^6=X^{o(1)}$ coordinate representations, since every coordinate divides $U$; the same holds for $V$. If no index in $I$ is divisible by $e$, then $\delta_e=2\delta_0$ and this alternative is redundant. \end{proof} The numerical certificate below uses $\delta_e$ precisely for $e=2,3,4,5,6$. The new feature is this systematic use, not Lemma~\ref{lem:general-degree} itself. \subsection{Energy with an arbitrary unit twist} \begin{lemma}[Inherited energy estimate; proof reproduced]\label{lem:energy} Fix positive integers $j,k$. Let $M\ge1$, let $\gamma$ be a unit modulo $M$, and let $P,Q\ge1$. The number $\mathcal K_\gamma$ of unit-coordinate solutions to \begin{equation}\label{eq:twisted-congruence} x^jy^k\equiv\gamma(x')^j(y')^k\pmod M, \qquad x,x'\sim P,\quad y,y'\sim Q, \end{equation} satisfies, for every $\eta>0$, \begin{equation}\label{eq:energy-bound} \mathcal K_\gamma\ll_{j,k,\eta}(PQM)^\eta PQ \sqrt{(1+P^2/M)(1+Q^2/M)}. \end{equation} In particular, no coprimality condition on $j,k$ is needed. \end{lemma} \begin{proof} Let $G=(\Z/M\Z)^\times$, interpreted as the one-element group when $M=1$. For $g\in G$, let $R_P(g)$ count pairs $(x,x')$ with $x/x'=g$, and define \[ S_j(t)=\sum_{g^j=t}R_P(g),\qquad S_k(t)=\sum_{h^k=t}R_Q(h). \] Writing $g=x/x'$ and $h=y/y'$, the congruence becomes $g^jh^k=\gamma$. Therefore \[ \mathcal K_\gamma=\sum_{t\in G}S_j(t)S_k(\gamma/t). \] Since $t\mapsto\gamma/t$ permutes $G$, Cauchy--Schwarz gives \begin{equation}\label{eq:energy-cs} \mathcal K_\gamma^2\le\|S_j\|_2^2\|S_k\|_2^2 \le\rho_j(M)\rho_k(M)\mathcal M_P\mathcal M_Q, \end{equation} where $\mathcal M_P=\sum_gR_P(g)^2$ and $\rho_j(M)$ is the kernel size of $g\mapsto g^j$. The second inequality applies Cauchy--Schwarz on each fibre of the power map. For an odd prime power the unit group is cyclic, and for a power of $2$ it is a product of at most two cyclic groups. Hence \[ \rho_j(M)\le j^{2\omega(M)}\ll_{j,\eta}M^\eta, \] including $M=1$. Next, $\mathcal M_P$ counts $xt'\equiv x't\pmod M$ with all four coordinates in the interval of size $P$. Fixing $(x,t')$ gives $O(P^2)$ choices. The product $x't$ then lies in one residue class in $[1,4P^2]$, so it takes $O(1+P^2/M)$ integer values. Each value has $P^{o(1)}$ factorizations. Thus \[ \mathcal M_P\ll_\eta P^{2+\eta}(1+P^2/M). \] The analogous estimate for $\mathcal M_Q$, inserted into \eqref{eq:energy-cs}, proves \eqref{eq:energy-bound} after reducing the intermediate error exponents. The argument is uniform in $\gamma$. \end{proof} If $P=X^p$, $Q=X^q$, $M\ge X^w$ and $M\le X^{O(1)}$, then \eqref{eq:energy-bound} has exponent \begin{equation}\label{eq:kappa} \kappa(p,q;w)=p+q+\tfrac12\bigl(\pos{2p-w}+\pos{2q-w}\bigr). \end{equation} Only the lower bound for $M$ is used in the two main factors; the polynomial upper bound absorbs $M^\eta$. \section{The collision inequality with a third-letter modulus}\label{sec:collision} We now reproduce the collision formula of \cite[Section~4]{Li28} in the generality needed later. The formula is an inherited arithmetic input; its exhaustive degree-and-modulus deployment and the resulting certificates are new. Fix distinct letters $A,B,C$. Choose nonempty $I_A,I_B\subseteq\R$, a pair $j