Equality of Bernoulli double sums for multiplicative inverses

About 9 years old · traced to

Let zz and NN be integers with gcd⁡(z,N)=1\gcd(z,N)=1. Let B1B_1 and B2B_2 denote the first and second Bernoulli polynomials, respectively, and interpret z(k−ℓ) mod Nz(k-\ell)\bmod N as the residue modulo NN. Write z−1z^{-1} for the multiplicative inverse of zz modulo NN. Equality conjecture.

∑k,ℓ=1N−1B1(k/N) B2((z(k−ℓ) mod N)/N) B1(ℓ/N)=∑k,ℓ=1N−1B1(k/N) B2((z−1(k−ℓ) mod N)/N) B1(ℓ/N).\sum_{k,\ell=1}^{N-1} B_1(k/N) \, B_2((z(k-\ell)\bmod N)/N) \, B_1(\ell/N) = \sum_{k,\ell=1}^{N-1} B_1(k/N) \, B_2((z^{-1}(k-\ell)\bmod N)/N) \, B_1(\ell/N).

The equality was verified numerically for all N≤4001N\leq 4001 and all z∈{1,…,N−1}z\in\{1,\ldots,N-1\} coprime to NN, and it is also proved for Fibonacci lattice rules in the cited context. A general proof or counterexample is not provided here.

References

Primary source

Dirk Nuyens and Ronald Cools, “The analysis of vertex modified lattice rules in a non-periodic Sobolev space”, arXiv:1709.03449 (2017).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.