Weighted sum formulas for finite multiple zeta values (Q1786692): Difference between revisions

The \textit{multiple zeta values} are the real numbers defined by \[ \zeta(k_1,\dots,k_r)=\sum_{1\le m_1<\dots<m_r}\frac{1}{m_1^{k_1}\cdots m_r^{k_r}} \] for positive integers \(k_1,\dots,k_r\) with \(k_r\ge2\). The \textit{finite multiple zeta values} are the elements of the \(\mathbb{Q}\)-algebra \[ \Biggl(\prod_{\text{\(p\): prime}}\mathbb{F}_p\Biggr)\Biggm/\Biggl(\bigoplus_{\text{\(p\): prime}}\mathbb{F}_p\Biggr) \] that are defined by \[ \zeta^{\mathcal{A}}(k_1,\dots,k_r)=\Biggl(\sum_{1\le m_1<\dots<m_r<p}\frac{1}{m_1^{k_1}\cdots m_r^{k_r}}\bmod p\Biggr)_p \] for positive integers \(k_1,\dots,k_r\). It is known that the multiple zeta values and the finite multiple zeta values often satisfy similar relations. The paper under review proves a weighted sum formula for finite multiple zeta values, thus giving an analogue of weighted sum formulas for multiple zeta values. Since the main theorem is rather complicated, we spell out two corollaries of the main theorem. If \(\boldsymbol{\alpha}=(\alpha_1,\ldots,\alpha_k)\) is a finite sequence of positive integers, then we set \(\vert\boldsymbol{\alpha}\rvert=\alpha_1+\dots+\alpha_k\) and let \(\vert\boldsymbol{\alpha}\rvert\) denote the largest integer \(m\in\{0,1,\ldots,k\}\) such that \(\alpha_i=1\) whenever \(1\le i\le m\). Then Corollary 2.3 states that if \(k\) is a positive integer and \(r\) is a nonnegative integer, then we have \[ \sum_{\vert\boldsymbol{\alpha}\rvert=k+r}w(\boldsymbol{\alpha})\zeta^{\mathcal{A}}(\alpha_1,\dots,\alpha_k)=(-1)^{r-1}\zeta^{\mathcal{A}}(\underbrace{1,\ldots,1}_{k-1},r+1); \] Corollary 2.5 states that if \(k\) is a positive integer and \(r\) is an even positive integer, then we have \[ \sum_{\vert\boldsymbol{\alpha}\rvert=k+r}2^{w(\boldsymbol{\alpha})}\zeta^{\mathcal{A}}(\alpha_1,\dots,\alpha_k)=(0)_p. \]