Abstract:We prove that the dimension of the $\mathbb{Q}$-linear span of $1,\zeta(3),\zeta(5),\ldots,\zeta(s-1)$ is at least $1.009 \cdot \log s/(1+\log 2)$ for any sufficiently large even integer $s$ and we claim that the constant $1.009$ can be improved further to $1.108$. This slightly refines a well-known result of Rivoal (2000) or Ball-Rivoal (2001). Quite unexpectedly, the proof only involves inserting the arithmetic observation of Zudilin (2001) into the original proof of Ball-Rivoal. Although our result is covered by a recent development of Fischler (2021+), our proof has the advantages of being simple and providing explicit non-vanishing small linear forms in $1$ and odd zeta values.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to improve the lower bound estimate in the Ball - Rivoal theorem regarding the linear independence of odd zeta values. Specifically, the author Li Lai has proven that for a sufficiently large even number \(s\), the dimension of the vector space over the field of rational numbers generated by 1 and odd zeta values \(\zeta(3), \zeta(5), \ldots, \zeta(s - 1)\) is at least:
\[
\dim_{\mathbb{Q}} \text{Span}_{\mathbb{Q}} (1, \zeta(3), \zeta(5), \ldots, \zeta(s - 1)) \geq \frac{1.009\cdot\log s}{1+\log 2}
\]
This slightly improves the known results of Rivoal (in 2000) or Ball - Rivoal (in 2001). Moreover, the author claims that this constant 1.009 can be further improved to 1.108.
### Specific background and problem description
1. **Arithmetic properties of special values**: The values of the Riemann zeta function at positive integers have always received extensive attention. Euler has proven that the values of the zeta function at positive even numbers are in the form of a rational number multiplied by a power of \(\pi\). However, for zeta values at positive odd numbers greater than 1 (i.e., "odd zeta values"), their properties are more mysterious. An important conjecture is that \(\pi\) and all odd zeta values are algebraically independent over the field of rational numbers.
2. **Existing breakthroughs**: Rivoal in 2000 and Ball and Rivoal in 2001 made significant breakthroughs. They proved that for a sufficiently large odd number \(s\):
\[
\dim_{\mathbb{Q}} \text{Span}_{\mathbb{Q}} (1, \zeta(3), \zeta(5), \ldots, \zeta(s)) \geq \frac{\log s}{(1 + \log 2)\log s}-o(1)
\]
3. **Zudilin's observation**: In 2001, Zudilin discovered that if different lengths of Pochhammer symbols are considered, then the coefficients of the generated linear forms have a large common factor \(\Phi_n\). This common factor played a crucial role in Zudilin's proof that at least one odd zeta value is irrational.
4. **Motivation for improvement**: Previously, it was thought that Zudilin's \(\Phi_n\) factor had little impact on the asymptotic results. However, by introducing Zudilin's method into the Ball - Rivoal framework, the author of this paper has proven that this previous expectation was inaccurate and has discovered that the \(\Phi_n\) factor does have an asymptotic impact.
### Main contributions
By combining Zudilin's \(\Phi_n\) factor method with the original proof of Ball - Rivoal, the author has established new results and proven that for a sufficiently large even number \(s\), the lower bound of the above - mentioned dimension can be reached:
\[
\dim_{\mathbb{Q}} \text{Span}_{\mathbb{Q}} (1, \zeta(3), \zeta(5), \ldots, \zeta(s - 1)) \geq \frac{1.009\cdot\log s}{1+\log 2}
\]
In addition, the author has also proposed some computational improvements, so that the constant 1.009 can be further increased to 1.108.