The Riemann Hypothesis in Oaxaca

Carlos Segovia · 2026

Read The Riemann Hypothesis in Oaxaca by Carlos Segovia (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-22413.txt.

Introduction

Carlos Segovia links the Riemann Hypothesis, the most famed open claim in math, to a grid of number partitions. The hypothesis concerns where a key function hits zero, and it governs how primes spread. His 2026 preprint uses a known equivalent form of the claim.

That form rests on a threshold for how large a divisor-sum can grow. Segovia builds a direct bridge from that form to the Young lattice, a grid that sorts box diagrams. A new bridge gives attackers of the claim a fresh path. It may also reveal why the threshold takes its odd value. The work is a link, not a proof.

Catalog record

CREATORCarlos Segovia
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE16 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-22413

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The Riemann Hypothesis in Oaxaca Carlos Segovia Abstract An equivalence of the Riemann Hypothesis (RH) enables a direct bridge to the Young lattice. In specific, the classical threshold lim n → ∞ σ ⁡ ( n ) / ( n ​ log ⁡ log ⁡ n ) = e γ ≈ 1.78107 \lim_{n\to\infty}\sigma(n)/(n\log\log n)=e^{\gamma}\approx 1.78107 , derived from the asymptotic behavior of the sum-of-divisors function, admits a combinatorial interpretation via limiting proportions associated to specific families of integer partitions. Riemann [ 7 ] introduced the zeta function \zeta(s) with the assertion that all of the nontrivial zeros have real part equal to 1/2 . Later, Robin [ 8 ] established an equivalence between the RH and the inequality: \frac{\sigma(n)}{n}<e^{\gamma}\log\log n,\quad\textrm{ for }n>5040\,, (1) with \sigma(n) the sum of divisors and \gamma the Euler–Mascheroni constant. Lagarias [ 5 ] provided an elementary equivalence of the RH with the inequality: \sigma(n)\leq H_{n}+e^{H_{n}}\log H_{n}\,, (2) where H_{n} is the n -Harmonic number. Recently, Espinosa published an equivalence of the RH in [ 3 ] . In what follows I present an sketch of his proof: A(n):=\sum_{k=0}^{\infty}a_{k}\frac{(H_{n})^{k}}{k!}\geq\sigma(n)\,, (3) for H_{k}-\gamma\leq a_{k}\leq\log(k)+\frac{1}{k} , k\geq 1 . Proof. Take the sum \displaystyle\sum_{k=0}^{\infty}\left(\frac{H_{k}-\gamma}{k!}\right)x^{k} \displaystyle=\sum_{k=1}^{\infty}H_{k}\frac{x^{k}}{k!}-\gamma e^{x}\,, (4) \displaystyle=e^{x}(\log x+\gamma+E_{1}(x))-\gamma e^{x}\,, (5) \displaystyle=e^{x}(\log x+E_{1}(x))\,, (6) where we use \displaystyle\sum_{k=1}^{\infty}H_{k}\frac{x^{k}}{k!} \displaystyle=\int_{0}^{1}\frac{e^{x}-e^{xt}}{1-t}dt\,, (7) \displaystyle=e^{x}\int_{0}^{x}\frac{1-e^{-t}}{t}dt\,, (8) which is equal to e^{x}(\log x+\gamma+E_{1}(x)) by [ 1 , 5.1.39] . By Lagarias [ 5 ] , for n\geq 3 , e^{\gamma}n\log\log n\leq e^{H_{n}}\log H_{n}\,. (9) Suppose the RH, by Robin [ 8 ] , for n>5040 , we have \sigma(n)\leq e^{\gamma}n\log\log n\,, (10) For 1\leq n\leq 5040 we can verify by computer that \sigma(n)\leq A(n) . Suppose the inequality \sigma(n)\leq A(n) . If the RH is false, by Robin [ 8 ] , there exists constants 0<\beta<\frac{1}{2} and C>0 such that the inequality \sigma(n)\geq e^{\gamma}n\log\log n+\frac{Cn\log\log n}{(\log n)^{\beta}}\,, (11) holds for infinitely many n . In addition, by Lagarias [ 5 ] , for n\geq 3 , we have the inequality e^{H_{n}}\log H_{n}\leq e^{\gamma}n\log\log n+\frac{3e^{\gamma}n}{2\log n}\,. (12) Because of ( 6 ), we obtain \displaystyle\frac{Cn\log\log n}{(\log n)^{\beta}} \displaystyle\leq\frac{3e^{\gamma}n}{2\log(n)}+e^{H_{n}}E_{1}(H_{n})\,, (13) \displaystyle\leq\frac{3e^{\gamma}n}{2\log n}+\frac{e^{\gamma}(n+1)}{\log n}\,, (14) \displaystyle=\frac{e^{\gamma}}{\log n}\left(\frac{5}{2}n+1\right)\,, (15) where e^{H_{n}}E_{1}(H_{n})\leq e^{H_{n}}/H_{n}\leq e^{\gamma}(n+1)/\log n since \log n<H_{n}<\log(n+1)+\gamma for n\geq 1 . Therefore, we have the inequality: \frac{C}{e^{\gamma}}\log\log n\leq(\log n)^{\beta-1}\left(\frac{5}{2}+\frac{1}{n}\right)\,. (16) Taking limits with n\rightarrow\infty , we obtain the left-side tend to infinity, while the right-side tend to zero, which is a contradiction. We can increase the coefficients H_{k}-\gamma\leq a_{k}\leq\log k+1/k providing the following inequality (see de Appendix): A(n)\leq e^{H_{n}}\log H_{n}+\frac{2e^{\gamma}(n+1)}{\log n}\,. (17) This inequality implies we do not exceed the threshold \lim_{n\rightarrow\infty}A(n)/n\log\log n=e^{\gamma} . ∎ The Young lattice is one of the most important and beautiful structures in mathematics. The connection with the RH arises through the interpretation of A(n) in terms of integer partitions: \exp(x_{1}+\cdots+x_{n})=\sum_{j=1}^{\infty}\sum_{\mu\vdash j}\frac{m_{\mu}\left(x_{1},x_{2}\ldots,x_{n}\right)}{\mu_{1}!\,\mu_{2}!\,\ldots\,\mu_{\ell(\mu)}!}\,, (18) for m_{\mu} , the monomial symmetric polynomial associated with the partition \mu=[\mu_{1},\cdots,\mu_{l(\mu)}] . Consider partitions of j with order i and define the sums: E_{i,j}(n):=\sum_{\mu\vdash j,\,\ell(\mu)=i}\frac{m_{\mu}(x_{1},\dots,x_{n})}{\mu_{1}!\mu_{2}!\dots\mu_{i}!}\,. (19) Recently, Espinosa in [ 4 ] , provides an interesting identity summing appropriately the monomials in E_{i,j}(n) called the assembly theorem. The following is an improvement adding the coefficients in order to obtain complete sums. We denote the r -sum as follows: A_{r}(n)=\sum_{i=1}^{n}\log(i+r)E_{i,i+r-1}(n)\,, (20) and we obtain A(n)=\sum_{r=1}^{\infty}A_{r}(n) with x_{i}=1/i , for i=1,\ldots,n . The case r=1 is a well understood logarithmic structure A_{1}(n)=\sum_{m=1}^{n}\log(m+1)e_{m}(n)\,, (21) where the e_{m} are the elementary symmetric polynomials with e_{m}(n):=e_{m}\left(1,\frac{1}{2},\ldots,\frac{1}{n}\right)=\genfrac{[}{]}{0.0pt}{}{n+1}{m+1}/n!\,, (22) the numerator is the sign-less Stirling number of first kind, see [ 2 ] . Denote by C_{n} the number of cycles of a permutation of n elements. The elementary symmetric polynomials have the following probabilistic interpretation: e_{m}(n)=(n+1)\mathbb{P}(C_{m+1}=m+1)\,, (23) where \mathbb{P}(C_{n+1}=m+1) denotes the probability that a permutation of n+1 elements, chosen uniformly at random, has exactly m+1 cycles; see [ 2 ] . Consequently, A_{1}(n)=(n+1)\sum_{k=1}^{n+1}\log k\cdot\mathbb{P}(C_{n+1}=k)=(n+1)\cdot\mathbb{E}[\log C_{n+1}]\,. (24) For r\geq 2 , we found the following determinantal formulas: p_{k}=\begin{vmatrix}e_{1}&1&0&0&\cdots&0\\ 2e_{2}&e_{1}&1&0&\cdots&0\\ 3e_{3}&e_{2}&e_{1}&1&\cdots&0\\ \vdots&\vdots&\vdots&\vdots&\ddots&\vdots\\ ke_{k}&e_{k-1}&e_{k-2}&\cdots&e_{2}&e_{1}\end{vmatrix}\,,\hskip 28.45274ptm_{[r,1^{l}]}=\begin{vmatrix}e_{1}&1&0&0&\cdots&0\\ 2e_{2}&e_{1}&1&0&\cdots&0\\ 3e_{3}&e_{2}&e_{1}&1&\cdots&0\\ \vdots&\vdots&\vdots&\vdots&\ddots&\vdots\\ (r-1)e_{r-1}&e_{r-2}&e_{r-3}&\cdots&e_{1}&1\\ (l+r)e_{l+r}&e_{l+r-1}&e_{l+r-2}&\cdots&e_{l+2}&e_{l+1}\end{vmatrix}\,, (25) where the first one can be found in [ 6 ] . Recall the p_{k} and e_{i} denote the power and elementary symmetric polynomials, respectively, and m_{[r,1^{l}]} are the monomial symmetric polynomials for the partitions [r] , [r,1] , [r,1,1] , \ldots (denoted by [r,1^{l}] for shortly). We denote the sub-summand of A_{r}(n) : \tilde{A}_{r}(n)=\frac{1}{r!}\sum_{l=0}^{n-1}\log(l+r+1)\,m_{[r,1^{l}]}(x_{1},x_{2},\ldots,x_{n})\,. (26) The use of the previous determinantal formulas give the expresion: \tilde{A}_{r}(n)=\frac{1}{r!}\det\begin{pmatrix}e_{1}(n)&1&0&\cdots&0\\ 2e_{2}(n)&e_{1}(n)&1&\cdots&0\\ \vdots&\vdots&\vdots&\ddots&\vdots\\ (r-1)e_{r-1}(n)&e_{r-2}(n)&\cdots&e_{1}&1\\ S_{1}^{(r)}&S_{2}^{(r)}&\cdots&S_{r-1}^{(r)}&S_{r}^{(r)}\end{pmatrix}\,, (27) where S^{(r)}_{1}=\sum_{m=1}^{n}\log(m+r)\cdot(m+r-1)e_{m+r-1}(n) (28) and S^{(r)}_{j}=\sum_{m=1}^{n}\log(m+r)\cdot e_{m+r-j}(n),\quad j=2,\ldots,r\,. (29) Remember e_{m}(n)=(n+1)\mathbb{P}(C=m+1) , m\geq 0 . For C:=C_{n+1} , denote the following: E_{1}=\mathbb{E}[(C-1)\log C]\textrm{ and }E_{j}=\mathbb{E}[\log(C+j-1)],\textrm{ for }j\geq 2\,. (30) Consequently, we write S_{1}^{(r)}=(n+1)E_{1}-(n+1)\sum_{t=1}^{r}\log t\cdot(t-1)\,\mathbb{P}(C=t)=(n+1)E_{1}-\sum_{t=1}^{r}\log t\cdot(t-1)\,e_{t-1}(n) (31) and for j\geq 2 , S_{j}^{(r)}=(n+1)E_{j}-(n+1)\sum_{t=1}^{r-j+1}\log(t+j-1)\,\mathbb{P}(C=t)=(n+1)E_{j}-\sum_{t=1}^{r-j+1}\log(t+j-1)\,e_{t-1}(n),\quad j\geq 2\,. (32) where e_{t-1}(n)=(n+1)\mathbb{P}(C=t) states our previous notation. Set \Delta_{1}=-\sum_{t=1}^{r}\log t\cdot(t-1)e_{t-1} and \Delta_{j}=-\sum_{t=1}^{r-j+1}\log(t+j-1)e_{t-1} , for j\geq 2 . Because the matrix below has zero determinant: \begin{pmatrix}e_{1}&1&0&0&\cdots&0\\ 2e_{2}&e_{1}&1&0&\cdots&0\\ 3e_{3}&e_{2}&e_{1}&1&\cdots&0\\ \vdots&\vdots&\vdots&\vdots&\ddots&\vdots\\ (r-1)e_{r-1}&e_{r-2}&e_{r-3}&\cdots&e_{1}&1\\ \Delta_{1}&\Delta_{2}&\Delta_{3}&\cdots&\Delta_{r-1}&\Delta_{r}\end{pmatrix}\,, (33) therefore, we obtain the formula, for r\geq 2 : \tilde{A}_{r}(n)=\frac{n+1}{r!}\sum_{j=1}^{r}(-1)^{r+j}p_{j-1}(n)E_{j}\,. (34) In order to find expressions for E_{j} , 1\leq j\leq r , we have the following: \mathbb{E}\bigl[\log(C_{n}+a)\bigr]=\log\bigl(H_{n}+a\bigr)+\sum_{m=2}^{\infty}\frac{(-1)^{m-1}}{m}\frac{\mathbb{E}[(C_{n}-H_{n})^{m}]}{(H_{n}+a)^{m}}, (35) and \mathbb{E}[C_{n}\log C_{n}]=H_{n}\log H_{n}-\sum_{m=2}^{\infty}\frac{(-1)^{m-1}}{m(m-1)}\frac{\mathbb{E}[(C_{n}-H_{n})^{m}]}{H_{n}^{m-1}}\,. (36) The central moments \mu_{m}:=\mathbb{E}[(C-H_{n})^{m}] are given in terms of the cumulants of the number of cycles of permutations C_{n} : \begin{array}[]{c}\kappa_{1}=H_{n}=p_{1}(n)\,,\\ \kappa_{2}=p_{1}(n)-p_{2}(n)\,,\\ \kappa_{3}=p_{1}(n)-3p_{2}(n)+2p_{3}(n)\,,\\ \vdots\\ \kappa_{m}=\sum_{\ell=1}^{m}(-1)^{\ell-1}(\ell-1)!\;S(m,\ell)\;p_{\ell}\,,\end{array} (37) where S(m,\ell) is the Stirling of second kind. The central moments \mu_{m}=B_{m}(0,\kappa_{2},\ldots,\kappa_{m}) are the complete Bell polynomials with positive integer coefficients evaluated in the cumulants: \mu_{1}=0,\quad\mu_{2}=\kappa_{2},\quad\mu_{3}=\kappa_{3},\quad\mu_{4}=\kappa_{4}+3\kappa_{2}^{2},\quad\mu_{5}=\kappa_{5}+10\kappa_{3}\kappa_{2},\quad\mu_{6}=\kappa_{6}+15\kappa_{4}\kappa_{2}+10\kappa_{3}^{2}+15\kappa_{2}^{3},\ldots (38) It is not difficult to convince ourself that B_{m}(0,\kappa_{2},\ldots,\kappa_{m})\sim c(\log n)^{\lfloor m/2\rfloor} for c a constant. In addition, we use H_{n+1}=\log n+\gamma+O(1/n) and \log H_{n+1}=\log\log n+\gamma/\log n+o(1/\log n) and we obtain: \displaystyle E_{1} \displaystyle=\mathbb{E}[C\log C]-\mathbb{E}[\log C]\,, (39) \displaystyle=\left[H_{n+1}\log H_{n+1}+\frac{\kappa_{2}}{2H_{n+1}}+O\left(\frac{1}{\log n}\right)\right]-\left[\log H_{n+1}+O\left(\frac{1}{\log n}\right)\right]\,, (40) \displaystyle=\log H_{n+1}(H_{n+1}-1)+\frac{\kappa_{2}}{2H_{n+1}}+O\left(\frac{1}{\log n}\right)\,, (41) \displaystyle=\left(\log\log n+\frac{\gamma}{\log n}+o\left(\frac{1}{\log n}\right)\right)\left(\log n+\gamma-1+O\left(\frac{1}{n}\right)\right)+O(1)\,, (42) \displaystyle=\log n\log\log n+(\gamma-1)\log\log n+\gamma+O(1)\,, (43) and \displaystyle E_{j} \displaystyle=\mathbb{E}(\log(C+j-1))\,, (44) \displaystyle=\left[\log H_{n+1}+\frac{j-1}{H_{n+1}}+o\left(\frac{1}{H_{n+1}^{2}}\right)\right]+O\left(\frac{1}{\log n}\right)\,, (45) \displaystyle=\log\log n+O(1)\,, (46) for j\geq 2 . As a consequence, we have the following: H_{n}E_{2}-E_{1}=\log\log n-\gamma+O(1)\,. (47) We deduce the following limits: \rho_{1}=\lim_{n\rightarrow\infty}\frac{A_{1}(n)}{n\log\log n}=1,\quad\rho_{2}=\lim_{n\rightarrow\infty}\frac{A_{2}(n)}{n\log\log n}=\lim_{n\rightarrow\infty}\frac{(n+1)}{2n}\frac{H_{n}E-E_{1}}{\log\log n}=\frac{1}{2}\,, (48) and for r\geq 2 , we have: \rho_{r}:=\lim_{n\rightarrow\infty}\frac{\tilde{A}_{r}(n)}{n\log\log n}=\lim_{n\rightarrow\infty}\frac{(n+1)(-1^{r})}{r!n}\left(\frac{-E_{1}+p_{1}(n)E_{2}-p_{2}(n)E_{3}+\ldots+(-1)^{r}p_{r-1}(n)E_{r}}{\log\log n}\right)\,. (49) Therefore, we obtain \rho_{r}=\frac{1}{r!}\left[(-1)^{r}+\sum_{j=3}^{r}(-1)^{r+j}\zeta(j-1)\right]\,. (50) From the proof of the RH equivalence we know that \lim_{n\rightarrow\infty}A(n)/n\log\log n=e^{\gamma}\approx 1.7810\ldots The previous formula provides the proportions: \begin{array}[]{l|l|l}r&\rho_{r}&S_{r}=\sum_{k=1}^{r}\rho_{k}\\ \hline\cr 1&1&1\\ 2&0.5&1.5\\ 3&0.107489011141371&1.607489011141371\\ 4&0.023213451512974&1.630702462654345\\ 5&0.004376669978331&1.635079132632676\\ 6&0.000710732441866&1.635789865074542\\ 7&0.000100320576238&1.635890185650781\\ 8&0.000012465811043&1.635902651461824\\ 9&0.000001381381681&1.635904032843505\\ 10&0.000000138005094&1.635904170848599\\ \end{array} Nevertheless, we get a strict inequality 1.6359<1.7810 where we need to add the sub-summands for partitions which are not of the form [r,1^{l}] . For instance, for r=3 , we have those partitions associated to [2,2] and for r=4 , we have to add the partitions associated to [3,2] and [2,2,2] . The correction term for the partitions associated to [2,2] , has the following form: R_{3}(n)=\frac{1}{4}\sum_{l=0}^{n-2}\log(l+5)m_{[2,2,1^{l}]}=\frac{1}{4}\sum_{l=0}^{n-2}\log(l+5)\left[e_{l+2}e_{2}-(l+2)e_{l+3}e_{1}+\frac{(l+1)(l+4)}{2}e_{l+4}\right]\,. (51) Computational calculations provides the proportion between A_{r}(n) and R_{r}(n)=A_{r}(n)-\tilde{A}_{r}(n) increases at the beginning but at some moment it should decrease, leaving the question of when this occurs and how. Our interest comes from speculative arrangements of the divisors: positive integers n with enough divisors, not a perfect square, we enumerate the divisors below the square root, 1=d_{1}<d_{2}<\cdots<d_{k}<\sqrt{n} (52) and the sum of divisors satisfies: \sigma(n)=\sum_{i=1}^{k}d_{i}+\frac{n}{d_{i}} (53) For the first summand A_{1}(n)\approx n\log\log n , we consider all the sum of divisors in pairs with divisors d_{i}\leq d^{(1)}(n) , where d^{(1)}(n) denotes the largest divisor satisfying the inequality: \sum_{d_{i}|n}^{d^{(1)}(n)}d_{i}+\frac{n}{d_{i}}\leq n\log\log n (54) For colossally

Read the complete text