Ramanujan’s partition generating functions modulo ℓ \ell
Kathrin Bringmann, William Craig and Ken Ono · 2025
Read Ramanujan’s partition generating functions modulo ℓ \ell by Kathrin Bringmann, William Craig and Ken Ono (2025) 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-2506-06101.txt.
Introduction
Kathrin Bringmann, William Craig, and Ken Ono extend Ramanujan's famous partition congruences to every prime from five up. Their work appeared in twenty twenty-five. Partitions count the ways to split a number into pieces, and Ramanujan found eerie patterns modulo five and seven.
His identities turned those patterns into exact generating formulas. The authors seek the same for all larger primes at once. They prove the relevant series collapses into a simple fraction. Its numerator counts traces tied to special values of Dirichlet series for cusp forms. Since no such forms exist in the smallest weights, Ramanujan's five, seven, and eleven congruences fall out anew. The proof unifies a century-old wonder.
Catalog record
| CREATOR | Kathrin Bringmann, William Craig and Ken Ono |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-06101
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Ramanujan’s partition generating functions modulo ℓ \ell Thanks: 2020 Mathematics Subject Classification. 05A17, 11P82 Kathrin Bringmann, William Craig and Ken Ono Abstract . For the partition function p ( n ) p(n) , Ramanujan proved the striking identities 𝒫 5 ( q ) := ∑ n ≥ 0 p ( 5 n + 4 ) q n = 5 ∏ n ≥ 1 ( q 5 , q 5 ) ∞ 5 ( q , q ) ∞ 6 , 𝒫 7 ( q ) := ∑ n ≥ 0 p ( 7 n + 5 ) q n = 7 ∏ n ≥ 1 ( q 7 , q 7 ) ∞ 3 ( q , q ) ∞ 4 + 49 q ∏ n ≥ 1 ( q 7 , q 7 ) ∞ 7 ( q , q ) ∞ 8 , \begin{split}\mathcal{P}_{5}(q):=\sum_{n\geq 0}p(5n+4)q^{n}&=5\prod_{n\geq 1}\frac{\left(q^{5};q^{5}\right)_{\infty}^{5}}{(q;q)_{\infty}^{6}},\\ \mathcal{P}_{7}(q):=\sum_{n\geq 0}p(7n+5)q^{n}&=7\prod_{n\geq 1}\frac{\left(q^{7};q^{7}\right)_{\infty}^{3}}{(q;q)_{\infty}^{4}}+49q\prod_{n\geq 1}\frac{\left(q^{7};q^{7}\right)_{\infty}^{7}}{(q;q)_{\infty}^{8}},\end{split} where ( q , q ) ∞ := ∏ n ≥ 1 ( 1 − q n ) . (q;q)_{\infty}:=\prod_{n\geq 1}(1-q^{n}). As these identities imply his celebrated congruences modulo 5 and 7, it is natural to seek, for primes ℓ ≥ 5 , \ell\geq 5, closed form expressions of the power series 𝒫 ℓ ( q ) := ∑ n ≥ 0 p ( ℓ n − δ ℓ ) q n ( mod ℓ ) , \mathcal{P}_{\ell}(q):=\sum_{n\geq 0}p(\ell n-\delta_{\ell})q^{n}\ \left(\mathrm{mod}\,\ell\right), where δ ℓ := ℓ 2 − 1 24 . \delta_{\ell}:=\frac{\ell^{2}-1}{24}. In this paper, we prove that 𝒫 ℓ ( q ) ≡ c ℓ 𝒯 ℓ ( q ) ( q ℓ , q ℓ ) ∞ ( mod ℓ ) , \mathcal{P}_{\ell}(q)\equiv c_{\ell}\dfrac{{\mathcal{T}}_{\ell}(q)}{\left(q^{\ell};q^{\ell}\right)_{\infty}}\ \left(\mathrm{mod}\,\ell\right), where c ℓ ∈ ℤ c_{\ell}\in\mathbb{Z} is explicit and 𝒯 ℓ ( q ) {\mathcal{T}}_{\ell}(q) is the generating function for the Hecke traces of ℓ \ell -ramified values of special Dirichlet series for weight ℓ − 1 \ell-1 cusp forms on SL 2 ( ℤ ) \mathrm{SL}_{2}(\mathbb{Z}) . This is a new proof of Ramanujan’s congruences modulo 5, 7, and 11, as there are no nontrivial cusp forms of weight 4, 6, and 10. 1. Introduction and Statement of Results A partition of n is any nonincreasing sequence of positive integers that sum to n. The number of partitions of n is denoted p(n) (by convention, we let p(0):=1 and p(n):=0 for n<0 ). Ramanujan famously proved (see [ 2 , 7 ] ), for every non-negative integer n , that \begin{split}p(5n+4)&\equiv 0\ \left(\mathrm{mod}\,5\right),\\ p(7n+5)&\equiv 0\ \left(\mathrm{mod}\,7\right),\\ p(11n+6)&\equiv 0\ \left(\mathrm{mod}\,11\right).\end{split} For the congruences with modulus 5 and 7, he used the beautiful identities \begin{split}\mathcal{P}_{5}(q):=\sum_{n\geq 0}p(5n+4)q^{n}&=5\prod_{n\geq 1}\frac{\left(q^{5};q^{5}\right)_{\infty}^{5}}{(q;q)_{\infty}^{6}},\\ \mathcal{P}_{7}(q):=\sum_{n\geq 0}p(7n+5)q^{n}&=7\prod_{n\geq 1}\frac{\left(q^{7};q^{7}\right)_{\infty}^{3}}{(q;q)_{\infty}^{4}}+49q\prod_{n\geq 1}\frac{\left(q^{7};q^{7}\right)_{\infty}^{7}}{(q;q)_{\infty}^{8}},\end{split} where (q;q)_{\infty}:=\prod_{n\geq 1}(1-q^{n}). In 1969, with the help of binary theta functions, Winquist [ 8 ] was able to offer a similar identity that proved Ramanujan’s congruence with modulus 11. In the spirit of these identities, for every prime \ell\geq 5, we determine the q -series \mathcal{P}_{\ell}(q)\in\mathbb{F}_{\ell}[[q]] \mathcal{P}_{\ell}(q):=\sum_{n\geq 0}p(\ell n-\delta_{\ell})q^{n}\ \left(\mathrm{mod}\,\ell\right), where \delta_{\ell}:=\frac{\ell^{2}-1}{24}. These expressions involve the generating functions of “weighted Hecke traces” of special values of specific Dirichlet series associated to weight \ell-1 Hecke eigenforms on \mathrm{SL}_{2}(\mathbb{Z}) (for background see [ 3 ] or [ 6 ] ). To define these Hecke traces, first suppose that ( q:=e^{2\pi iz} throughout) f(z):=q+\sum\limits_{n\geq 2}a_{f}(n)q^{n}\in S_{2k} is an even integer weight 2k Hecke eigenform on \mathrm{SL}_{2}(\mathbb{Z}) . For s\in\mathbb{C} with \mathrm{Re}(s)>2k, the twisted quadratic Dirichlet series is defined by D(f;s):=\sum\limits_{n\geq 1}\frac{\genfrac{(}{)}{}{}{12}{n}a_{f}\left(\frac{n^{2}-1}{24}\right)}{n^{s}}, where \genfrac{(}{)}{}{}{\cdot}{\cdot} denotes the Kronecker symbol. Furthermore, if k\geq 2 , 0\leq j\leq k-2 , and m\geq 0, then we let \displaystyle\beta(k,j,m) \displaystyle\hskip 14.22636pt:=\frac{(-1)^{j+1}\Gamma\!\left(k-\frac{1}{2}\right)\Gamma\!\left(k+\frac{1}{2}\right)}{9}\left(\frac{6}{\pi}\right)^{2k}\frac{(2k+m-2)!(k-j-1)^{[k]}\left(\frac{3}{2}\right)^{\![j]}}{j!m!(2k-j-2)!\left(-\frac{1}{2}-j\right)^{\![k]}\left(\frac{5}{2}\right)^{\![j]}}, where \Gamma(\cdot) is the usual Gamma-function. Moreover the rising factorial is given by (x)^{[j]}:=\begin{cases}x(x+1)\cdots(x+j-1)\ \ \ \ \ &{\text{\rm if}}\ j\geq 1,\\ 1\ \ \ \ \ &{\text{\rm if }j=0},\end{cases} which are companions of the usual falling factorials (x)_{m}:=\begin{cases}x(x-1)\cdots(x-m+1)\ \ \ \ &{\text{\rm if $m\geq 1$}},\\ 1\ \ \ \ \ &{\text{\rm if $m=0$}},\\ \frac{1}{(x)_{-m}}\ \ \ \ \ &{\text{\rm if $m\leq-1$}},\end{cases} For such f\in S_{2k}, we define the following sums of values of Dirichlet series by D_{f}:=\sum\limits_{j=0}^{k-2}\sum_{m\geq 0}\beta(k,j,m)D(f;2k+1+2m+2j). Moreover we define the weight 2k Hecke trace by \Tr_{2k}(n):=\sum_{f}a_{f}(n)\frac{D_{f}}{||f||}, where the sum runs over the normalized Hecke eigenforms f\in S_{2k}, and the Petersson norms of f , ||f|| , is defined as ( z=x+iy throughout) ||f||:=\int_{\mathrm{SL}_{2}(\mathbb{Z})\backslash\mathbb{H}}|f(z)|^{2}y^{2k}\frac{dxdy}{y^{2}}. As a_{f}(n) is the eigenvalue of the Hecke operator T_{n} , we refer to the numbers \Tr_{2k}(n) as Hecke traces. Finally, for primes \ell\geq 5, we collect the \ell -ramified values (i.e., the arguments that are multiples of \ell ) if 2k=\ell-1 as the Fourier coefficients of the generating function \mathcal{T}_{\ell}(q):=\sum_{n\geq 1}\Tr_{\ell-1}(\ell n)q^{n}. Theorem 1.1 . If \ell\geq 5 is a prime, then \mathcal{P}_{\ell}(q)\equiv c_{\ell}\dfrac{\mathcal{T}_{\ell}(q)}{\left(q^{\ell};q^{\ell}\right)_{\infty}}\ \left(\mathrm{mod}\,\ell\right), where c_{\ell}:=2\cdot\overline{3}(\frac{-1}{\ell})(\frac{\ell+1}{2})!^{\ell-3}\ \left(\mathrm{mod}\,\ell\right) , where throughout \overline{a} denotes the inverse of a\ \left(\mathrm{mod}\,\ell\right) . For \ell\in\{5,7,11\} , we have that S_{\ell-1}=\{0\}. As there are no nontrivial cusp forms in these spaces, we immediately obtain a new proof of Ramanujan’s famous partition congruences. Corollary 1.2 . For n\in\mathbb{N} , we have \begin{split}p(5n+4)&\equiv 0\ \left(\mathrm{mod}\,5\right),\\ p(7n+5)&\equiv 0\ \left(\mathrm{mod}\,7\right),\\ p(11n+6)&\equiv 0\ \left(\mathrm{mod}\,11\right).\end{split} Moreover Theorem 1.1 immediately implies the following congruence formula for p(\ell n-\delta_{\ell})\ \left(\mathrm{mod}\,\ell\right) in terms of p(0),p(1),\dots,p(n-1) . Corollary 1.3 . If \ell\geq 5 is a prime and n\in\mathbb{N} , then we have p(\ell n-\delta_{\ell})\equiv c_{\ell}\sum_{\begin{subarray}{c}j,m\geq 0\\ \ell j+m=n\end{subarray}}p(j)\Tr_{\ell-1}(\ell(m-j))\ \left(\mathrm{mod}\,\ell\right). Example . For the prime \ell=13, Theorem 1.4 and Corollary 1.3 of [ 4 ] gives \displaystyle\mathcal{T}_{13}(q) \displaystyle=-\dfrac{33108590592}{691}\Delta|U_{13}(z)\equiv 7\Delta|U_{13}(z)\ \left(\mathrm{mod}\,13\right), where f|U_{j}(z):=\sum_{n\geq 1}a_{f}(jn)q^{n} for j\in\mathbb{N} . Using c_{13}\equiv 6\ \left(\mathrm{mod}\,13\right) , this gives \begin{split}c_{13}\dfrac{\mathcal{T}_{13}(q)}{\left(q^{13};q^{13}\right)_{\infty}}&\equiv\dfrac{3\Delta|U_{13}(z)}{\left(q^{13};q^{13}\right)_{\infty}}\\ &\equiv 11q+9q^{2}+3q^{3}+6q^{4}+12q^{5}+6q^{6}+q^{8}+\dots\ \left(\mathrm{mod}\,13\right).\end{split} To illustrate Theorem 1.1 , we note that \begin{split}\mathcal{P}_{13}(q)&=\sum_{n\geq 1}p(13n-7)q^{n}=11q+490q^{2}+8349q^{3}+89134q^{4}+715220q^{5}+\dots\\ &\equiv 11q+9q^{2}+3q^{3}+6q^{4}+12q^{5}+6q^{6}+q^{8}+\dots\ \left(\mathrm{mod}\,13\right).\end{split} Furthermore, Corollary 1.3 implies, for n\in\mathbb{N} , that \begin{split}p(13n-7)&\equiv 3\sum_{\begin{subarray}{c}j,m\geq 0\\ 13j+m=n\end{subarray}}p(j)\tau(13(m-j))\ \left(\mathrm{mod}\,13\right).\\ \end{split} To obtain Theorem 1.1 , we make use of recent work of Gomez, the third author, Saad, and Singh [ 4 ] that offers an infinite family of generalizations of Euler’s “Pentagonal Number” recurrence for p(n) . In Section 2 we recall these formulas, and in Section 3 we use them to obtain Theorem 1.1 . 2. Generalizations of Euler’s “Pentagonal number” recurrence For n\in\mathbb{N}, Euler famous recurrence relation asserts that (see p. 12 of [ 1 ] ) \displaystyle p(n) \displaystyle=p(n-1)+p(n-2)-p(n-5)-p(n-7)+\dots (2.1) \displaystyle=\sum_{m\in\mathbb{Z}\setminus\{0\}}(-1)^{m+1}p(n-\omega(m)), where \omega(m):=\frac{3m^{2}+m}{2} is the m -th pentagonal number . This recurrence is one of the most efficient methods for computing partition numbers. Gomez, the third author, Saad, and Singh [ 4 ] proved that Euler’s recurrence is the first case of an infinite family of rich recurrence relations satisfied by the partition numbers. To make this precise, we make use of Dedekind’s eta-function \eta(z):=q^{\frac{1}{24}}\prod_{n\geq 1}\left(1-q^{n}\right)=\sum_{n\in\mathbb{Z}}(-1)^{n}q^{\frac{1}{24}(6n+1)^{2}}, where z\in\mathbb{H}, the upper half of the complex plane. To define these relations, we require the differential operator D:=\frac{1}{2\pi i}\frac{d}{dz}=q\frac{d}{dq}. For k\in\mathbb{N}_{0}, we define 1 1 1 To avoid confusing notation, we note that R_{k}(z) is denoted P_{k}(z) in [ 4 ] . R_{k}(z):=\frac{(2k-1)(2k-2)_{k-1}^{2}}{2^{2k-2}}\sum_{\begin{subarray}{c}r,s\geq 0\\ r+s=k\end{subarray}}(-1)^{r}\frac{2s-1}{(2r)!(2s)!}D^{r}\left(\frac{1}{\eta(z)}\right)D^{s}(\eta(z)). By [ 4 ] , we have \displaystyle R_{k}(z)=\sum_{\begin{subarray}{c}n\geq 0\\ m\in\mathbb{Z}\end{subarray}}(-1)^{m+1}g_{k}(n,m)p(n-\omega(m))q^{n}, where g_{k}(n,m):=\frac{(2k-1)(2k-2)_{k-1}^{2}}{2^{2k-2}}\\ \times\sum_{r=0}^{k}(-1)^{k+r}\frac{2k-2r-1}{(2r)!(2k-2r)!}(6m+1)^{2r}(24n-(6m+1)^{2})^{k-r}. By Theorem 1.1 of [ 4 ] , for each k\geq 0 , R_{k} is a weight 2k holomorphic modular form on \mathrm{SL}_{2}(\mathbb{Z}) . These expressions are simple