On the correlations between character sums of division polynomials under shifts
Subham Bhakta, Igor E. Shparlinski · 2025
Read On the correlations between character sums of division polynomials under shifts by Subham Bhakta, Igor E. Shparlinski (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-2507-09271.txt.
Introduction
Subham Bhakta and Igor Shparlinski bound sums that pair values of division polynomials on an elliptic curve at shifted points, in a math paper from 2025. The sums weigh each value by a multiplicative character over a finite field, and the shift moves the second factor along the curve.
Their bound holds on average over shifts in a short range of shift sizes. That short range reach is the hard part of the result. A multi dimensional form of the bound is also proved for higher cases. Such sums feed work on curve based codes and on the spread of curve points in number theory.
Catalog record
| CREATOR | Subham Bhakta, Igor E. Shparlinski |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-09271
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the correlations between character sums of division polynomials under shifts Subham Bhakta, Igor E. Shparlinski Abstract . Let E E be an elliptic curve over the finite field 𝔽 p \mathbb{F}_{p} , and P ∈ E ( 𝔽 p ) P\in E(\mathbb{F}_{p}) be an 𝔽 p {\mathbb{F}}_{p} -rational point. We study the sums S χ , P ( N , h ) = ∑ n = 1 N χ ( ψ n ( P ) ) χ ( ψ n + h ( P ) ) , S_{\chi,P}(N,h)=\sum_{n=1}^{N}\chi(\psi_{n}(P))\chi(\psi_{n+h}(P)), where ψ n ( P ) \psi_{n}(P) denotes the n n -th division polynomial evaluated at P P , and χ \chi is a multiplicative character of 𝔽 p ∗ {\mathbb{F}}_{p}^{*} . We estimate S χ , P ( N , h ) S_{\chi,P}(N,h) on average over h h over a rather short interval h ∈ [ 1 , H ] h\in[1,H] . We also obtain a multidimensional generalisation of this result. 2020 Mathematics Subject Classification 11G07, 11L40 1. Introduction 1.1. Set-up and motivation Let E be an elliptic curve over {\mathbb{F}}_{p} of characteristic p>3 given by the Weierstrass equation as following, with coefficients in {\mathbb{F}}_{p} . y^{2}=x^{3}+ax+b. For any integer n\geq 0 , define the n th division polynomial \psi_{n}\in{\mathbb{F}}_{p}[x,y] as follows. \displaystyle\psi_{0} \displaystyle=0,\quad\psi_{1}=1,\quad\psi_{2}=2y, \displaystyle\psi_{3} \displaystyle=3x^{4}+6ax^{2}+12bx-a^{2}, \displaystyle\psi_{4} \displaystyle=4y(x^{6}+5ax^{4}+20bx^{3}-5a^{2}x^{2}-4abx-8b^{2}-a^{3}), for more details, see [ 8 , Exercise 3.7] , combined with the discussion in [ 8 , Chapter III.1] . Denote by E({\mathbb{F}}_{p}) the group of points on E defined over {\mathbb{F}}_{p} . We can interpret each \psi_{n} as rational functions on E({\mathbb{F}}_{p}) . The multiplication of P\in E by n is given as a rational map by \displaystyle[n](P) \displaystyle=\biggl(\frac{x(P)\psi_{n}^{2}(P)-\psi_{n-1}(P)\psi_{n+1}(P)}{\psi_{n}(P)^{2}}, \displaystyle\qquad\qquad\frac{\psi_{n-1}(P)^{2}\psi_{n+2}(P)-\psi_{n-2}(P)\psi_{n+1}(P)^{2}}{4y(P)\psi_{n}(P)^{3}}\biggr). It is well known that \displaystyle\psi_{m+n}(P) \displaystyle\psi_{m-n}(P)\psi_{r}^{2}(P) \displaystyle=\psi_{m+r}(P)\psi_{m-r}(P)\psi_{n}^{2}(P)-\psi_{n+r}(P)\psi_{n-r}(P)\psi_{m}(P)^{2}, for any integers m,n,r , see, for example, [ 8 , Exercise 3.7(g)] . It turns out that \psi_{n}(P) is periodic, and the period T , which could be as large as (p-1)\operatorname{ord}P , where \operatorname{ord}P is the order of P in the group E({\mathbb{F}}_{p}) , see [ 7 , Corollary 9] . Let \chi be a multiplicative character of {\mathbb{F}}_{p}^{*} . We consider the sequence \chi(\psi_{n}(P)) , where of course we set \chi(0)=0 . Then, \chi(\psi_{n}(P)) is periodic with a much smaller period (see Section 1.3 ), compared to the period of \psi_{n}(P) . For points P\in E({\mathbb{F}}_{p}) of large orders, Shparlinski and Stange [ 6 ] have studied the character sums of the form S_{\chi,P}(N)=\sum_{n=1}^{N}\chi(\psi_{n}(P)), for the quadratic character of {\mathbb{F}}_{p} and optained a nontrivial bound provided \operatorname{ord}P\geq N\geq p^{1/2+\varepsilon} for some fixed \varepsilon>0 . Here, we study a related but more difficult question on the correlation of \chi(\psi_{n}(P)) with its shifts, that is, about bounding the sums S_{\chi,P}(N,h)=\sum_{n=1}^{N}\chi(\psi_{n}(P))\chi(\psi_{n+h}(P)) with an integer h\neq 0 . More precisely, we ask the following: Question 1.1 . Let p be a prime, P\in E(\mathbb{F}_{p}) be a point and \chi be a multiplicative character of {\mathbb{F}}_{p}^{*} . When do we have S_{\chi,P}(N,h)=o(N)? Due to the limitations of our techniques, we do not have an affirmative answer to the question above. However, we prove an average version. For instance, we estimate the following sum \sum_{h=1}^{H}|S_{\chi,P}(N,h)|=\sum_{h=1}^{H}\left|\sum_{n=1}^{N}\chi(\psi_{n}(P))\chi(\psi_{n+h}(P))\right|. Question 1.1 itself, as well as our results, are motivated by the progress towards the Chowla conjecture for \lambda(n) and \mu(n) , that is, for the Liouville and Möbius functions, respectively. It is well-known that the prime number theorem is equivalent to any of the following statements \sum_{n=1}^{N}\lambda(n)=o(N)\quad\text{and}\quad\sum_{n=1}^{N}\mu(n)=o(N),\qquad\text{as}\ N\to\infty, Chowla [ 1 ] has conjectured that for any integer h>0 , the following holds (1.1) \sum_{n=1}^{N}\lambda(n)\lambda(n+h)=o(N)\quad\text{and}\quad\sum_{n=1}^{N}\mu(n)\mu(n+h)=o(N), as N\to\infty . Despite several striking results towards these conjectures, both remain widely open, see [ 4 , 5 , 9 , 10 , 11 ] and references therein. An averaged version of this conjecture has been studied in [ 4 ] , in fact for a more general class of multiplicative functions. Our function \chi\left(\psi_{n}(P)\right) is not multiplicative, although it satisfies an almost multiplicative property, as stated in Lemma 2.1 . Thus, Question 1.1 essentially studies an elliptic analogue of Chowla’s conjectures ( 1.1 ), for the function \chi\left(\psi_{n}(P)\right) , and also remains widely open. 1.2. Notation Throughout the whole article, as usual, the notations U=O(V) , U\ll V , and V\gg U are equivalent to |U|\leq cV for some positive constant c , which may depend on the the order d of the multiplicative character \chi and on the dimension m . We use \omega(n) and \tau(n) to denote the number of distinct prime and positive integer factors of an integer n\neq 0 and use \varphi(n) to denote the Euler function. We also denote {\mathbf{\,e}}_{R}(n)=\exp(2\pi in/R) . Everywhere, we use p to denote a prime, and h,k,m,n are reserved for integers. 1.3. Main result Let \chi be a multiplicative character on {\mathbb{F}}_{p}^{*} of order d , and P\in E(\mathbb{F}_{p}) be a point. Denote \operatorname{ord}P be the order of P in E(\mathbb{F}_{p}) , and set (1.2) R=d\operatorname{ord}P. Applying [ 6 , Lemma 3.1] for s=d , the sequences \chi(\psi_{n}(P)) is periodic with a period dividing R , as long as \operatorname{ord}P\geq 3 . Here, given m integers h_{1},\ldots,h_{m} , we consider the following average values of multidimensional correlations between the values of \chi(\psi_{n}(P)) : \displaystyle U_{m,\chi,P}(H,N)=\frac{1}{H^{m-1}}\sum_{h_{2},\ldots,h_{m}=1}^{H}\left|\sum_{n=1}^{N}\chi(\psi_{n}(P))\prod_{j=2}^{m}\chi(\psi_{n+h_{j}}(P))\right|, \displaystyle V_{m,\chi,P}(H,N)=\frac{1}{H^{m}}\sum_{h_{1},\ldots,h_{m}=1}^{H}\left|\sum_{n=1}^{N}\prod_{j=1}^{m}\chi(\psi_{n+h_{j}}(P))\right|. Theorem 1.2 . For any integer m\geq 2 , and integers 1\leq H,N\leq R , we have (1.3) \begin{split}U_{m,\chi,P}(H,N)&,V_{m,\chi,P}(H,N)\\ &\ll H^{-1/8}R\exp\left(O\left((\log R)^{1/2}/\log\log R\right)\right)\\ &\qquad\qquad\qquad\qquad\qquad\quad+H^{1/2}R^{3/4}p^{1/8}\log R,\end{split} and if R\geq p^{1/2}\exp\left(2.1\log p/\log\log p\right) , then we also have (1.4) \begin{split}U_{m,\chi,P}(H,N)&,V_{m,\chi,P}(H,N)\\ &\ll H^{-1/4}R^{7/6}p^{1/24}(\log R)(\log\log R)^{1/6}.\end{split} Remark 1.3 . It is easy to see from the proof of Theorem 1.2 that for complete sums N=R , one can remove \log R from the second term of the bound ( 1.3 ) and from the bound ( 1.4 ). To understand the strength of Theorem 1.2 we consider the extreme case of very long sums with R\geq N\gg R . Then the bound ( 1.3 ) is nontrivial in the range \exp\left(C(\log R)^{1/2}/\log\log R\right)\leq H\leq R^{1/2}p^{-1/4}(\log R)^{-2}f(p)^{-1} for some absolute constant C>0 and any function f(p)\to 0 as p\to\infty . In fact, for any fixed \varepsilon>0 , Remark 3.2 allows us to have a non-trivial bound in the extended lower range H\geq\exp\left(C_{\varepsilon}(\log R)^{\varepsilon}\right), where C_{\varepsilon}>0 is a constant depending on \varepsilon . On the other hand, the bound ( 1.4 ) is nontrivial only when H is quite large, namely R\geq H\geq R^{2/3}p^{1/6}(\log\log R)^{2/3}F(p) with any function F(p)\to\infty as p\to\infty . Unfortunately there is a wide gap between this ranges as R^{2/3}p^{1/6}/R^{1/2}p^{-1/4}\geq p^{5/12}R^{1/6}\gg p^{1/2}. (since the we have R\geq p^{1/2} for either of the bounds ( 1.3 ) and ( 1.4 ) to be nontrivial). Closing this gap is certainly an interesting open question. Another interesting regime is H=N . In this case the bound ( 1.3 ) is always trivial, while ( 1.4 ) requires R\geq H=N\geq R^{14/15}p^{1/30}(\log R)^{4/5}(\log\log R)^{2/15}F(p) with F(p)\to\infty as p\to\infty . In a response to Question 1.1 , we have the following simple application of Theorem 1.2 . Namely, taking m=2 and applying ( 1.4 ) we immediately derive the following result. Corollary 1.4 . Let \Delta>0 be an arbitrary real number. If R\geq p^{1/2}\exp\left(2.1\log p/\log\log p\right) then, we have \displaystyle\sharp\,\left\{1\leq h\leq H:\penalty\ \left|S_{\chi,P,h}(N)\right|\geq\Delta N\right\} \displaystyle\ll\Delta^{-1}H^{3/4}N^{-1}