\documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage{lmodern,amsmath,amssymb,amsthm,microtype} \usepackage[letterpaper,margin=0.93in]{geometry} \usepackage[hidelinks]{hyperref} \hypersetup{pdftitle={The sum of the reciprocal Mersenne cubes is irrational}, pdfauthor={twinheterodox}} \newtheorem{theorem}{Theorem} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \newcommand{\Q}{\mathbb Q} \newcommand{\Z}{\mathbb Z} \newcommand{\C}{\mathbb C} \newcommand{\Ord}{\operatorname{ord}} \newcommand{\cl}{\mathcal L} \newcommand{\ph}{\varphi} \newcommand{\ep}{\varepsilon} \DeclareMathOperator{\rank}{rank} \allowdisplaybreaks[1] \raggedbottom \title{\Large\bfseries The sum of the reciprocal Mersenne cubes is irrational} \author{twinheterodox\\[2pt]\normalsize Prepared with Codex} \date{18 November 2026} \makeatletter \renewcommand{\@maketitle}{% \begin{center} {\@title\par}\medskip \begin{tabular}{c}\@author\end{tabular}\par\smallskip {\small\@date\par} \end{center}} \makeatother \begin{document} \maketitle \begin{abstract} The sum of the reciprocal cubes of the Mersenne numbers \(2^n-1\) is irrational. More generally, for \(|b|>1\) let \(M_j(b)=\sum_{n\ge1}(b^n-1)^{-j}\). For every integer \(b\ge2\), we prove that \(1,M_1(b),M_2(b),M_3(b)\) are linearly independent over \(\Q\). These four numbers are also linearly independent over \(\Q(b)\) whenever \(b\) is an algebraic integer with \(|b|>1\) and every conjugate other than \(b\) and \(\bar b\) lies in the closed unit disk. This includes all Pisot and Salem numbers. \end{abstract} The reciprocals of the Mersenne numbers \(1,3,7,15,\ldots\) sum to the Erd\H{o}s--Borwein constant. The number \(1\), this constant, and the sum of the squares of those reciprocals are linearly independent over \(\Q\) \cite[Theorem~1.3]{PVA}. We prove that adjoining the sum of their cubes raises the dimension of the rational span from three to four. For a complex base \(b\) with \(|b|>1\) and a positive integer \(j\), write \[ M_j(b)=\sum_{m\ge1}\frac{1}{(b^m-1)^j}. \] \begin{theorem}\label{thm:main} For every integer \(b\ge2\), the four numbers \(1,M_1(b),M_2(b),M_3(b)\) are linearly independent over \(\Q\). In particular \(1+3^{-3}+7^{-3}+15^{-3}+\cdots\) is irrational. \end{theorem} At the golden ratio \(b=(1+\sqrt5)/2\), even coefficients in \(\Q(\sqrt5)\) give no nontrivial relation among these four numbers. Corollary~\ref{cor:field} gives the general condition on the conjugates of an algebraic base. To compare the theorem with earlier results, consider the Lambert sums \[ Z_s(b)=\sum_{n\ge1}\frac{n^{s-1}}{b^n-1}. \] Expanding \(x^j/(1-x)^j\) at \(x=b^{-m}\) and summing in \(m\) gives \begin{equation}\label{eq:triangle} M_1=Z_1,\qquad M_2=Z_2-Z_1,\qquad 2M_3=Z_3-3Z_2+2Z_1 . \end{equation} This triangular change of variables is invertible over \(\Q\). The first two identities translate the independence of \(1,Z_1,Z_2\), proved by Postelmans and Van Assche and also by Zudilin \cite[Theorem~2]{ZudPair}, into the three-number result used above. The public inventory for the weight-three determinant reports a companion result asserting independence of \(1,Z_2,Z_3\) \cite[entry FUT-LI]{UpliftCompanion}. Both three-number statements are compatible with \(Z_3-3Z_2+2Z_1\) being rational, which by \eqref{eq:triangle} would make the reciprocal cube sum rational. We argue by contradiction. For a fixed nonzero triple \(A,B,C\), let \[ \xi=A Z_1(b)+B Z_2(b)+C Z_3(b). \] Under a rational relation we can take \(A,B,C,\xi\) all integral. We adapt the weight-three determinant of \cite[\S\S1--2]{Uplift} to this combination of Lambert sums. After clearing its denominators and dividing out a common power of \(b\), we obtain an integer. We show that its absolute value tends to zero with the size of the determinant and that infinitely many of these integers are nonzero. The weight-one term \(AZ_1\) requires a subtraction at zero in each summand. That subtraction affects both the common power of \(b\) and the estimate at the value \(\xi\). \section{One determinant for a quadratic weight} The series converge absolutely for complex \(|b|>1\). Put \(q=1/b\). For \(j\ge1\), summing \(x^j\) over the points \(q^m\), \(m\ge1\), gives \(1/(b^j-1)\). The Euler derivative \(\theta=x\,d/dx\) multiplies \(x^j\) by \(j\), so \(A+B\theta+C\theta^2\) supplies the numerator \(A+Bj+Cj^2\). For a general holomorphic function \(f\), the term \(Af(q^m)\) tends to \(Af(0)\). To obtain a convergent series we subtract this limit in every summand and put \begin{equation}\label{eq:L} L(f)=\sum_{m\ge1} \bigl[A(f(q^m)-f(0))+B(\theta f)(q^m)+C(\theta^2f)(q^m)\bigr]. \end{equation} For \(f\) holomorphic on a neighborhood of \(|x|\le |q|\), each summand is \(O_f(|q|^m)\) for fixed \(A,B,C,q\). Thus \(L(1)=0\), and \(L(x^j)=(A+Bj+Cj^2)/(b^j-1)\) for \(j\ge1\). To see why poles at \(b,b^2,\ldots\) are useful, write the first \(a\) summands at a variable base \(t\) as \(S_a(t)\), where \[ h(v)=\frac{A}{v-1}+\frac{Bv}{(v-1)^2} +\frac{Cv(v+1)}{(v-1)^3},\qquad S_a(t)=\sum_{m=1}^a h(t^m). \] For example, \((A,B,C)=(2,-3,1)\) gives \(h(v)=2/(v-1)^3\) and \(\xi=2M_3(b)\). A general triple lets us test every putative relation. Expanding the geometric series gives \(\xi=\sum_{m\ge1}h(b^m)\). In \eqref{eq:L}, replacing \(f(x)=(1-x)^{-1}\) by \((1-q^a x)^{-1}\) shifts the summation index from \(m\) to \(m+a\). Only the first \(a\) terms of the series for \(\xi\) are lost. Thus \[ L((1-q^a x)^{-1})=\xi-S_a(b). \] The points \(q^m\) where \(L\) samples a function lie inside the unit disk. The poles \(b^a\) of these test fractions lie outside it, and each fraction introduces the same unknown \(\xi\). To clear denominators before evaluating at the actual base and series value, replace \(b,\xi\) by indeterminates \(t,X\) and treat \(A,B,C\) as independent indeterminates as well. For \(K\ge1\) set \[ F(x)=\prod_{a=1}^K(x-t^a),\qquad D(x)=\prod_{a=1}^K(1-x/t^a). \] Define \(\cl_{t,X}\) on fractions with denominator dividing \(D\) by partial fractions, including the polynomial quotient. Assign \[ \mu_0=0,\qquad \cl_{t,X}(x^j)=\mu_j=\frac{A+Bj+Cj^2}{t^j-1}\quad(j\ge1),\qquad \cl_{t,X}((1-x/t^a)^{-1})=X-S_a(t). \] The determinant of these moments is \begin{equation}\label{eq:delta} \Delta_K(t,X)= \det\left[\cl_{t,X}\left(\frac{x^{i+j}}{D(x)}\right)\right]_{0\le i,jK/2\), at most two nodes coincide at any point. This accounts for the two ranges in the multiplier below. Let \(\Phi_d\) be the \(d\)-th cyclotomic polynomial and put \begin{equation}\label{eq:clearer} H_K(t)=\prod_{d=1}^K\Phi_d(t)^{c_d},\qquad c_d=\begin{cases} \lceil K^2/(4d)+2K+17d/16\rceil,&2d\le K,\\ 5(K-d)+3,&2d>K. \end{cases} \end{equation} The degree \(\sum_{d\le K}c_d\varphi(d)\), where \(\varphi\) is Euler's totient, controls the size of this product at \(b\). On the scale \(K^3\log|b|\), division by the base power saves \(5/12\), and the triple zeros in Section~\ref{sec:decay} will save another \(1/9\). The multiplier must cost less than their sum. Its exponents come from the sharpened pure weight-three estimate \cite[\S4 and Appendix A]{Uplift}. We show why the same exponents suffice for arbitrary \(A,B,C\). \begin{lemma}\label{lem:clear} The polynomial \(P_K=t^{-e_K}H_K\Delta_K\) belongs to \(\Z[t,A,B,C,X]\). Moreover \[ \deg H_K\le\gamma K^3+O(K^2\log(2K)),\qquad \gamma=\frac{321}{64\pi^2},\qquad \sum_{d\le K}c_d=O(K^2\log(2K)). \] \end{lemma} \begin{proof} Fix a primitive \(d\)-th root \(\omega\), put \(\tau=t-\omega\), and work locally with \(A,B,C,X\) independent. A cluster \(I_r\) consists of the \(a\) in \(1,\ldots,K\) congruent to \(r\bmod d\). Write \(m=|I_r|,\ F_r=\prod_{a\in I_r}(x-t^a)\), and use the basis \((F/F_r)(x-\omega^r)^j,\ 0\le jK\), put \(u=K-d\). The \(u\) double clusters are \(\{a,a+d\}\) with \(1\le a\le u1\) and complex \(A,B,C\). At the true value \(\xi\), \[ |P_K(b,A,B,C,\xi)|\le \exp\{-\ep K^3\log |b|+O_{b,A,B,C}(K^2\log(2K))\}, \qquad \ep=\frac{19}{36}-\frac{321}{64\pi^2}>0 . \] \end{lemma} \begin{proof} Put \(r=|q|\) and, for \(j\ge0\), take \[ B_j(x)=\prod_{\ell=1}^j(x-q^{\lceil\ell/3\rceil}),\quad s_j=\sum_{\ell=1}^j\lceil\ell/3\rceil,\quad R_j=\lceil j/3\rceil . \] These polynomials are monic, so changing both rows and columns preserves \(\Delta_K\). For one entry write \(R=\max(R_i,R_j)\), \(W=1/D\), and \(h=B_iB_jW\). On \(|x|\le1\), \(W\) is bounded by \(\prod_{a\ge1}(1-r^a)^{-1}\). Cauchy's inequalities bound \(W,W',W''\) on \(|x|\le r\) uniformly in \(K\). If \(1\le m1\), the cyclotomic product formula gives \(\log|\Phi_d(b)|=\ph(d)\log|b|+O_b(1)\) uniformly in \(d\). Indeed the error is bounded by \(\sum_{h\ge1}-\log(1-|b|^{-h})<\infty\). In units of \(K^3\log|b|\), the triple zeros supply \(1/9\) of decay and division by the common power of \(b\) supplies another \(5/12\) by Lemma~\ref{lem:base}. The cyclotomic multiplier costs at most \(\gamma\) by Lemma~\ref{lem:clear}. Their difference \(1/9+5/12-\gamma=\ep\) is positive already from \(\pi>31/10\). \end{proof} \section{Nonvanishing for arbitrarily large sizes} The coefficients \(A,B,C\) may have either sign in a rational relation, so we need a nonvanishing argument that also works without positivity. We encode the moments in a meromorphic function. A singular determinant of size \(k\) imposes interpolation conditions at the exterior points \(b,\ldots,b^k\). If all large sizes were singular, successive approximations would coincide. The function's poles at the interior points \(q,q^2,\ldots\) will rule this out. We use the interpolation argument of \cite[\S5]{Uplift}, in the complex-node form of \cite[\S5]{Golden}. \begin{lemma}\label{lem:nonzero} If \(|b|>1\) and \((A,B,C)\ne(0,0,0)\), then \(\Delta_K(b,\xi)\ne0\) for arbitrarily large \(K\). \end{lemma} \begin{proof} For \(|z|>|q|\), define \(\mathcal F(z)=L((z-x)^{-1})\). Its summand at \(u=q^m\) is \[ A\left(\frac1{z-u}-\frac1z\right) +\frac{(B+C)u}{(z-u)^2}+\frac{2Cu^2}{(z-u)^3}. \] The series is analytic at infinity. It converges locally normally away from zero and the nodes, giving a meromorphic continuation to \(\C\setminus\{0\}\). At each node the highest nonzero weight supplies a genuine pole, of order three if \(C\ne0\), two if \(C=0,B\ne0\), and one otherwise. Hence \(\mathcal F\) is nonrational. If the size \(k\) determinant is zero, there is a nonzero polynomial \(Q_k\) of degree less than \(k\) with \(L(x^jQ_k/D_k)=0\) for \(0\le j1\). Suppose all of its conjugates other than \(b\) and \(\bar b\) have modulus at most one. Then \(1,M_1(b),M_2(b),M_3(b)\) are linearly independent over \(\Q(b)\). This includes all real Pisot and Salem bases. \end{corollary} \begin{proof} We first bound \(P_K(t,A,B,C,Y)\) on \(|t|\le1\) for any fixed complex \(A,B,C,Y\). We give the details of the circle argument from \cite[\S4]{Golden} because the conjugate of \(\xi\) at such a place is simply a fixed number \(Y\). On \(|t|=1+1/K\), the nodes \(r_a=t^a\) have modulus at most \(e\), and \(|1-r_a/r_c|\ge1/(K+1)\) for \(a\ne c\). Thus \[ 1/D(x)=\sum_{a=1}^K\frac{w_a}{1-x/r_a},\qquad |w_a|=\prod_{c\ne a}|1-r_a/r_c|^{-1}\le(K+1)^{K-1}. \] Each tail \(S_a\) is \(O_{A,B,C}(K^4)\) and \(\mu_j=O_{A,B,C}(K^3)\) for \(0\le j\le2K\), since \(|t^m-1|\ge1/K\) for \(m\ge1\). For \(U_{a,j}=\cl_{t,Y}(x^j/(1-x/r_a))\) the exact recurrence \[ U_{a,0}=Y-S_a,\qquad U_{a,j+1}=r_a(U_{a,j}-\mu_j) \] bounds \(|U_{a,j}|\) by \(\exp(O_{A,B,C,Y}(K+\log(2K)))\) for \(j\le2K\). Summing \(w_a U_{a,j}\) and expanding the determinant gives \(\log^+|\Delta_K|=O_{A,B,C,Y}(K^2\log(2K))\). On that circle \(1/(K+1)\le |1-t^{-h}|\le2\) for \(h\ge1\). M\"obius inversion bounds \(\log|\Phi_d(t)|\le\ph(d)\log(1+1/K)+\tau(d)\log(2K+2)\), where \(\tau(d)\) counts the divisors. In \eqref{eq:clearer}, \(c_d=O(K^2/d)\), and \(\sum_{d\le K}\tau(d)/d\le(1+\log K)^2\). Lemma~\ref{lem:clear} therefore gives \(\log^+|H_K|=O(K^2\log^3(2K))\). Also \(|t^{-e_K}|\le1\). The maximum modulus principle for the integral polynomial now yields \[ \max_{|t|\le1}\log^+|P_K(t,A,B,C,Y)| =O_{A,B,C,Y}(K^2\log^3(2K))=o(K^3). \] Scale any putative field relation by one integer so that its coefficients \(A,B,C\) and value \(\xi\) are algebraic integers. For the unbounded sequence of nonzero sizes, \(P_K(b,A,B,C,\xi)\) then has a nonzero integer norm. At the specified embedding, Lemma~\ref{lem:decay} gives exponential decay on the \(K^3\) scale. If \(b\) is nonreal, the complex conjugate of this embedding gives the same absolute value. Each remaining embedding costs only \(o(K^3)\) in the logarithm by the disk bound at its fixed conjugate parameters. The absolute norm tends to zero on this sequence, a contradiction. The change of variables \eqref{eq:triangle} also holds by absolute convergence for complex \(|b|>1\). \end{proof} \section*{Verification} Another AI model independently audited the mathematics and exposition of the 18 November 2026 draft. It verified Theorem~\ref{thm:main} and Corollary~\ref{cor:field}, rederived the decay margin, and reproduced the finite checks with independently written code. \small \begin{thebibliography}{9} \setlength{\itemsep}{1pt} \bibitem{PVA} K. Postelmans and W. Van Assche, Irrationality of \(\zeta_q(1)\) and \(\zeta_q(2)\), \emph{J. Number Theory} \textbf{126} (2007), 119--154. \href{https://arxiv.org/abs/math/0604312}{arXiv:math/0604312}. \bibitem{ZudPair} W. Zudilin, Approximations to \(q\)-logarithms and \(q\)-dilogarithms, with applications to \(q\)-zeta values, \emph{J. Math. Sci. (N.Y.)} \textbf{137} (2006), 4673--4683. \href{https://www.math.ru.nl/~zudilin/PS/qhyp.pdf}{Theorem 2}. \bibitem{Uplift} uplift-universal-inc, Irrationality of the Lambert-series value \(\zeta_{1/p}(3)\), research draft, 12 October 2026. \href{https://github.com/uplift-universal-inc/qzeta_3-irrationality/blob/main/paper/main.pdf}{Manuscript}. \bibitem{UpliftCompanion} uplift-universal-inc, Results reserved for companion articles, entry FUT-LI. \href{https://github.com/uplift-universal-inc/qzeta_3-irrationality/blob/main/docs/FUTURE_RESULTS.md}{Research inventory}. \bibitem{Kratt} C. Krattenthaler, A determinant identity for moments of orthogonal polynomials that implies Uvarov's formula for the orthogonal polynomials of rationally related densities. \href{https://arxiv.org/abs/2103.03969}{arXiv:2103.03969}. \bibitem{ZudHankel} W. Zudilin, On the irrationality of generalized \(q\)-logarithm, \emph{Res. Number Theory} \textbf{2} (2016), article 15. \href{https://doi.org/10.1007/s40993-016-0042-x}{doi:10.1007/s40993-016-0042-x}. \bibitem{BZ} P. Bundschuh and W. Zudilin, Irrationality measures for certain \(q\)-mathematical constants, \emph{Math. Scand.} \textbf{101} (2007), 104--122. \href{https://doi.org/10.7146/math.scand.a-15034}{doi:10.7146/math.scand.a-15034}. \bibitem{Golden} twinheterodox, The weight-three Lambert sum at the golden ratio is irrational, Hexagon submission 2611.00623v1 (2026), \S\S3--5. \href{https://github.com/twinheterodox/algebraic-base-lambert-values/blob/main/output/pdf/algebraic-base-lambert-values.pdf}{Manuscript PDF}. \bibitem{MersenneCode} twinheterodox, Verification programs for the reciprocal Mersenne cubes (2026). \href{https://github.com/twinheterodox/algebraic-base-lambert-values/tree/main/checks/mersenne}{Code and reference results}. \end{thebibliography} \end{document}