Abstract:We prove an inequality of the form $\gamma_n\geq C_n(\gamma_{n-1})$ giving a lower bound for the Hermite constant $\gamma_n$ in dimension $n$ in terms of $\gamma_{n-1}$. Iterating this inequality yields a new proof of the Minkowski-Hlawka bound.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is a new inequality regarding Hermite constants (denoted as \(\gamma_n\)), so as to provide a new proof and an improved lower - bound estimate for the maximum density (\(\Delta_n\)) of n - dimensional lattice packing.
Specifically, the core problem of the paper is to construct a continuously increasing function \(C_n(x)\) such that:
\[
\gamma_n \geq C_n(\gamma_{n - 1})
\]
where \(\gamma_m\) is the Hermite constant in m - dimensional space. This inequality not only gives a new lower bound for \(\gamma_n\), but also derives the Minkowski - Hlawka bound \(\Delta_n\geq\zeta(n)2^{1 - n}\), where \(\Delta_n\) represents the maximum density of n - dimensional lattice packing and \(\zeta(n)\) is the Riemann Zeta function.
In addition, the main results proposed in the paper also include:
1. **New inequality**: By comparing the maximum density, the central density and the Hermite constants in different dimensions, a series of complex inequalities are given.
2. **Improved lower - bound estimate**: In particular, for specific dimensions (such as \(n = 9, 25\)), it provides a better lower - bound estimate than the known bounds.
3. **More technical results**: For any small positive number \(\epsilon\), there exists an integer \(N\) such that for all \(n\geq N\), we have:
\[
\Delta_n\geq\frac{1-\epsilon}{2^n}\sum_{k = 1}^{\infty}e^{-k^2\pi}\sim(1-\epsilon)\cdot2^{3.1388 - n}
\]
These results together form a deeper understanding of Hermite constants and related problems, and provide a theoretical basis for further research on high - dimensional lattice packing problems.