General Convolution Identities for Bernoulli and Euler Polynomials

K. Dilcher,C. Vignat
DOI: https://doi.org/10.48550/arXiv.1507.05356
2015-07-20
Abstract:Using general identities for difference operators, as well as a technique of symbolic computation and tools from probability theory, we derive very general kth order (k \ge 2) convolution identities for Bernoulli and Euler polynomials. This is achieved by use of an elementary result on uniformly distributed random variables. These identities depend on k positive real parameters, and as special cases we obtain numerous known and new identities for these polynomials. In particular we show that the well-known identities of Miki and Matiyasevich for Bernoulli numbers are special cases of the same general formula.
Number Theory
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to derive high - order convolution identities of Bernoulli polynomials and Euler polynomials. Specifically, the author uses general identities of difference operators, symbolic computation techniques and probability theory tools to derive very general \(k\)-order (\(k\geq2\)) convolution identities of these polynomials. These identities depend on \(k\) positive real parameters, and many known and new identities can be obtained as special cases through these general formulas. For example, the well - known Miki and Matiyasevich identities about Bernoulli numbers are proved to be special cases of this general formula. In addition, the paper also explores various special cases of these identities and their applications, further expanding previous research results. ### Formula Summary 1. **Generating function of Bernoulli numbers**: \[ \frac{z}{e^{z}-1}=\sum_{n = 0}^{\infty}B_{n}\frac{z^{n}}{n!}\quad(|z|< 2\pi) \] 2. **Euler identity**: \[ \sum_{j=0}^{n}\binom{n}{j}B_{j}B_{n - j}=-nB_{n-1}-(n - 1)B_{n}\quad(n\geq1) \] 3. **Miki identity**: \[ \sum_{j = 2}^{n-2}\frac{B_{j}B_{n-j}}{j(n - j)}-\sum_{j = 2}^{n-2}\binom{n}{j}\frac{B_{j}B_{n-j}}{j(n - j)}=2H_{n}\frac{B_{n}}{n}\quad(n\geq4) \] where \(H_{n}=1+\frac{1}{2}+\cdots+\frac{1}{n}\) is the \(n\)th harmonic number. 4. **Matiyasevich identity**: \[ (n + 2)\sum_{j=2}^{n-2}B_{j}B_{n-j}-2\sum_{j=2}^{n-2}\binom{n + 2}{j}B_{j}B_{n-j}=n(n + 1)B_{n}\quad(n\geq4) \] 5. **General convolution identity of Theorem 1**: \[ \sum_{l=0}^{n}\binom{n}{l}\frac{(a)_{l}(b)_{n-l}}{(a + b)^{n}}B_{l}(x)B_{n-l}(x)=\sum_{l=0}^{n}\binom{n}{l}\left(a\frac{(b)_{l}}{(a + b)^{l+1}}+b\frac{(a)_{l}}{(a + b)^{l+1}}\right)B_{l}B_{n-l}(x)+\frac{ab}{(a + b+1)(a + b)^{n}}B_{n-1}(x) \] 6. **Euler polynomial convolution identity of Theorem 3**: \[ \sum_{l=0}^{n}\binom{n}{l}\frac{(a)_{l}(b)_{n-l}}{(a + b)^{n}}E_{l}(x)E_{n-l}(x)=\frac{4}{n+1}B_{n+1}(x)-\frac{2}{n+1}\sum_{l=0}^{n+1}\binom{n+1}{l}\frac{(a)_{l}+(b)_{l}}{(a + b)^{l}}E_{l}(0)B_{n+1-l}(x) \] These formulas show the complex convolution relationships between Bernoulli and Euler polynomials and provide powerful tools for studying these special functions.