\documentclass[11pt]{amsart} \usepackage{amsmath} \usepackage{amssymb} \usepackage{amsthm} \usepackage{booktabs} \usepackage{hyperref} \newtheorem{theorem}{Theorem}[section] \newtheorem{proposition}[theorem]{Proposition} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \numberwithin{equation}{section} \setlength{\emergencystretch}{2em} \raggedbottom \newcommand{\ZZ}{\mathbb{Z}} \newcommand{\QQ}{\mathbb{Q}} \newcommand{\CC}{\mathbb{C}} \newcommand{\cO}{\mathcal{O}} \newcommand{\cP}{\mathcal{P}} \newcommand{\cR}{\mathcal{R}} \newcommand{\Nm}{\operatorname{N}} \newcommand{\Tr}{\operatorname{Tr}} \newcommand{\eu}{\phi_{\mathrm E}} \newcommand{\Tot}{\mathcal T} \newcommand{\spn}{\operatorname{span}} \newcommand{\file}[1]{{\def\_{\textunderscore\allowbreak}\texttt{#1}}} % Source: handoff/fibzeta/manyU/PROOF_MANYU.md v2, whole-chain refereed (REFEREE_MANYU.md, minor items m1--m6 applied). % Label: [Proved, mod Nesterenko's criterion (Fischler--Zudilin Thm A / KRZ Prop. 2) and Sum phi(l) = 3n^2/pi^2 + % O(n log n); not Lean-checked]. Lemma/equation numbers of [Kri26c] are those of Hexagon 2611.00693v1 % (= lean_qzeta3/paper/main.tex as built 2026-11-19, build/main.aux). \begin{document} \hypersetup{pdftitle={Many odd-power reciprocal sums of Lucas sequences are irrational},pdfauthor={Shawn Krisman}} \title[Odd-power reciprocal sums of Lucas sequences]{Many odd-power reciprocal sums\\ of Lucas sequences are irrational} \author{Shawn Krisman} \subjclass[2020]{Primary 11J72; Secondary 11B39, 11R11, 33D15} \keywords{Lucas sequences, Fibonacci numbers, reciprocal sums, Lambert series, irrationality, linear independence, Nesterenko's criterion} \begin{abstract} Let $a\ge3$, let $U_0=0$, $U_1=1$, $U_{n+1}=aU_n-U_{n-1}$, and let $\delta=\pm1$. We prove that $1$ and the numbers $\sum_{n\ge1}\delta^n/U_n^l$, $l=1,3,\dots,A-1$, span a $\QQ$-vector space of dimension at least $(\pi/\sqrt{12+\pi^2}+o(1))\sqrt A$. So infinitely many of these sums with $l$ odd are irrational, for instance infinitely many of $\sum_n1/F_{2n}^l$. With the known linear independence of $1$, $\sum_n\delta^n/U_n$ and $\sum_n\delta^n/U_n^3$, one of $\sum_n\delta^n/U_n^l$, $l=5,7,\dots,23$, lies outside the span of these three numbers. The proof adapts the author's construction for odd $q$-zeta values at real quadratic units, with a weighted bilateral sum. No single sum such as $\sum_n1/U_n^5$ is decided, and sequences of norm $-1$, such as the Fibonacci numbers, are not covered. \end{abstract} \maketitle % =========================================================================== \section{Introduction} % =========================================================================== Fix an integer $a\ge3$ and let $U_0=0$, $U_1=1$, $U_{n+1}=aU_n-U_{n-1}$. Then $U_n=(\eta^n-\eta^{-n})/(\eta-\eta^{-1})$ with $\eta=(a+\sqrt{a^2-4})/2>1$, a unit of norm $+1$ of $K=\QQ(\sqrt{a^2-4})$, and every unit $\eta>1$ of norm $+1$ of a real quadratic field arises in this way, with $a=\eta+\eta^{-1}$. Examples are $U_n=F_{2n}$ ($a=3$), $U_n=F_{kn}/F_k$ for even $k$ ($a=L_k$) and $U_n=\cP_{2n}/2$, where $\cP_n$ are the Pell numbers ($a=6$). Let $\delta\in\{1,-1\}$ and, for even $A\ge4$ and $1\le rk-1$, so that Theorem~\ref{thm:lucas} gives $\dim\ge k$ for the sums with odd $l\le A_k-1$. For $2\le k\le11$ the values are those of Table~\ref{tab:thresholds}, attained at $r=k-1$; at every even $A1$ be a unit of norm $+1$ of a real quadratic field, with trace $a$ and Lucas sequence $(U_n)$, let $\delta\in\{1,-1\}$, $\nu=\delta/\eta$, and let $d$ be the squarefree part of $a^2-4$. For odd $l\ge1$, \begin{equation}\label{eq:basis} \sum_{n\ge1}\frac{\delta^n}{U_n^l}=(\eta-\eta^{-1})^l\,V_l,\qquad V_l:=\sum_{n\ge1}\frac{\nu^{nl}}{(1-\nu^{2n})^l}=\sum_{\substack{m\ge1\\ m\text{ odd}}}\cP_l(m)\frac{\nu^m}{1-\nu^m}, \end{equation} where $\cP_l(m)=\prod_{i}(m^2-i^2)/(2^{l-1}(l-1)!)$, the product over odd $i$ with $1\le i\le l-2$. Hence $V_l=\sum_{j}a_{l,j}W_j(\nu)$, over odd $j\le l$, with $a_{l,j}=[m^{j-1}]\cP_l\in\QQ$ and $a_{l,l}\ne0$; and for every odd $L\ge1$, \[ \spn_\QQ\Bigl(1,\sum_n\frac{\delta^n}{U_n},\dots,\sum_n\frac{\delta^n}{U_n^L}\Bigr) =\spn_\QQ\bigl(1,\sqrt d\,W_1(\nu),\dots,\sqrt d\,W_L(\nu)\bigr). \] \end{lemma} \begin{proof} Since $1/U_n^l=(\eta-\eta^{-1})^l\eta^{-nl}(1-\eta^{-2n})^{-l}$ and $l$ is odd, $\delta^n\eta^{-nl}=\nu^{nl}$ and $\eta^{-2n}=\nu^{2n}$. Expanding $(1-\nu^{2n})^{-l}$ and summing over $n$ (absolutely convergent), $V_l=\sum_{i\ge0}\binom{i+l-1}{l-1}\nu^{l+2i}/(1-\nu^{l+2i})$. With $m=l+2i$, $\binom{(m-l)/2+l-1}{l-1}=\prod_{t=1}^{l-1}(m-l+2t)/(2^{l-1}(l-1)!)=\cP_l(m)$, since the factors $m-l+2t$ are the numbers $m\pm i$, $i$ odd, $i\le l-2$. As $\cP_l$ vanishes at $m=1,3,\dots,l-2$, the sum may run over all odd $m$. $\cP_l$ is even of degree $l-1$ with leading coefficient $1/(2^{l-1}(l-1)!)$, which gives the triangular expansion. Finally $(\eta-\eta^{-1})^2=a^2-4$, so $(\eta-\eta^{-1})^l=(a^2-4)^{(l-1)/2}f\sqrt d$ with $f\in\QQ^\times$. \end{proof} For example $V_3=(W_3-W_1)/8$ and $V_5=(9W_1-10W_3+W_5)/384$. \begin{remark}\label{rem:equiv} By Lemma~\ref{lem:basis}, Theorem~\ref{thm:lucas} for $(a,\delta)$ is Theorem~\ref{thm:lambert} at $\nu=\delta/\eta$. Conversely, for $\nu$ as in Theorem~\ref{thm:lambert}, $\eta=1/|\nu|>1$ is a unit of norm $+1$, its trace $a=\eta+\eta^{-1}$ is an integer $>2$, and $\nu=\delta/\eta$ with $\delta$ the sign of $\nu$. Since $\eta\in K\setminus\QQ$, $\QQ(\sqrt{a^2-4})=K$, so the squarefree $d$ of Lemma~\ref{lem:basis} is that of Theorem~\ref{thm:lambert}. \end{remark} % =========================================================================== \section{What is used from \texorpdfstring{\cite{Kri26c}}{[Kri26c]}}\label{sec:reused} % =========================================================================== From now on $A\ge4$ is even, $1\le r0$ depending only on $\kappa$, \[ |c_{s,j}(q)|\le\rho^{s-A(j+1)}c_+^{A-2r+1}\gamma_+\gamma_-^{-A}(1+\rho)^{rn+e}\kappa^{-\mathcal E(j)}, \] where $\mathcal E(j)\le(A+4r^2)n^2/8+rn/2$. \item[(f)] \cite[Lemma~7.1, Remark~7.2, and \S9 before (9.2)]{Kri26c}: $\sigma(d_n(q))=d_n(1/q)=-q^{-\Tot(n)}d_n(q)$, $\log|d_n(q)|=O(n)$, and $\Tot(n)=3n^2/\pi^2+O(n\log n)$ \cite[Theorem~330]{HW08}. \end{enumerate} We also use the trace argument of \cite[Lemma~9.3(a)]{Kri26c}: $\Tr_{K/\QQ}(x/\sqrt d)\in\ZZ$ for $x\in\cO_K$, $d$ squarefree; and Nesterenko's criterion as stated in \cite[\S9, proof of Theorem~1.1]{Kri26c}, from \cite{Nes85}, \cite[Theorem~A]{FZ10}, \cite[Proposition~2]{KRZ06}: \emph{if $\vartheta_1,\dots,\vartheta_N$ are real and there are integers $p_{j,n}$ with $\frac1{n^2}\log|\sum_jp_{j,n}\vartheta_j|\to\alpha<0$ and $\limsup_n\frac1{n^2}\log|p_{j,n}|\le\beta$ for every $j$, then $\dim_\QQ\sum_j\QQ\vartheta_j\ge1-\alpha/\beta$.} We apply all of this at $q=\nu^2$ with $\nu$ a unit of norm $+1$, $0<|\nu|<1$; then $q$ is a unit of norm $+1$ with $00. \] If $\alpha\ge0$ then $\tau(A,r)\le0$ and there is nothing to prove. Otherwise the dimension is at least \[ 1-\frac\alpha\beta=1+\frac{r(A-2r)/2-3A/(2\pi^2)}{3A/(2\pi^2)+(A+4r^2)/8}=1+\tau(A,r).\qedhere \] \end{proof} \begin{proof}[Proof of Theorem~\ref{thm:lucas}] By Lemma~\ref{lem:basis} at $\nu=\delta/\eta$, the span in Theorem~\ref{thm:lucas} is that of Theorem~\ref{thm:lambert}, with $d$ the squarefree part of $a^2-4$. The asymptotic is \cite[proof of Corollary~1.2]{Kri26c}: with $t=\sqrt{12+\pi^2}/(2\pi)$ and $r=\lceil t\sqrt A\rceil$ (admissible for $A\ge6$), $\tau(A,r)/\sqrt A\to4\pi^2t/(12+\pi^2+4\pi^2t^2)=\pi/\sqrt{12+\pi^2}$. If at most $u$ of the sums with odd $l\le A-1$ were irrational, the span would have dimension at most $u+1$; this gives the count, and $\max_r\tau(A,r)\to\infty$. \end{proof} \begin{proof}[Proof of Corollaries~\ref{cor:24} and~\ref{cor:examples}] Let $V$ be the span of Theorem~\ref{thm:lucas} at $A=A_k$, so $\dim V\ge k$ by Corollary~\ref{cor:table}. As $V$ is $V_3$ plus the span of the sums with $5\le l\le A_k-1$, and $\dim V_3=3$ by \cite[Theorem~1.1]{Kri26d} (its case $(P,Q)=(a,1)$, for both $\delta$; in its Lean statement $(a,E)=(a,-1)$), $\dim V/V_3\ge k-3$; (1) and (2) are $k=4$ and $k=5$. For Corollary~\ref{cor:examples}: $F_{2n}$ is the sequence with $a=3$; for even $k$, $F_{kn}=F_kU_n$ with $\eta=\varphi^k$ and $a=L_k$, so $\sum_n\delta^n/F_{kn}^l=F_k^{-l}\sum_n\delta^n/U_n^l$; and $\cP_{2n}=2U_n$ with $a=6$. \end{proof} % =========================================================================== \section{Thresholds and computations}\label{sec:thresholds} % =========================================================================== \begin{proof}[Proof of Corollary~\ref{cor:table}] Since $12A+\pi^2(A+4r^2)>0$, the inequality $1+\tau(A,r)>k-1$ is equivalent to \[ \pi^2\bigl[4r(A-2r)-(k-2)(A+4r^2)\bigr]>12A(k-1). \] For $2\le k\le11$, every even $A\le A_k$ and every $1\le r0$. (c) \emph{Sharper thresholds.} At $q=1/p$ the bounds of \cite{KRZ06} were improved by saving one power of $d_n$ \cite{JM10} and by a refinement of Nesterenko's criterion \cite{FZ10}; we have not adapted either. A full power of $d_n(\nu^2)$ cannot be saved in Lemma~\ref{lem:den}: in every case we computed exactly (in $\QQ(\nu)$, $A\le24$) the denominator of $P_0$ contains $\Phi_{2n}(\nu)^A$, so $d_n(\nu^2)^{A-1}P_0\notin\ZZ[\nu^{\pm1}]$ there. In these cases the exponent $A$ occurs only at $\Phi_{2\ell}(\nu)$ with $\ell$ in roughly $(2n/3,n]$ and at a few small indices, so a partial saving is not excluded. (d) \emph{Odd $A$.} We have not proved a version of the construction for odd $A$ (then $e$ is a half-integer for odd $n$). Such a version, with the same function $\tau$, would give nothing new. For fixed $r\ge1$, \[ \tau(A,r)=\frac{(4\pi^2r-12)A-8\pi^2r^2}{(12+\pi^2)A+4\pi^2r^2} \] is a M\"obius function of $A$ whose denominator is positive for $A>0$ and whose determinant $(4\pi^2r-12)\cdot4\pi^2r^2+8\pi^2r^2(12+\pi^2)$ is positive, so it is strictly increasing in $A$. Moreover $r