Abstract:Higher-dimensional Dedekind sums are defined as a generalization of a recent 1-dimensional probability model of Dilcher and Girstmair to a d-dimensional cube. The analysis of the frequency distribution of marked lattice points leads to new formulae in certain special cases, and to new bounds for the classical Dedekind sums. Upper bounds for the generalized Dedekind sums are defined in terms of 1-dimensional moments. In the classical two-dimensional case, the ratio of these sums to their upper bounds are cosines of angles between certain vectors of n-dimensional cones, conjectured to have a largest spacial angle of pi/6.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to define the Higher - dimensional Dedekind sums and study their properties and bounds. Specifically, the authors generalize the one - dimensional probability model to the multi - dimensional cube. By analyzing the frequency distribution of lattice points marked on the discrete diagonal, they propose new formulas and bounds. In particular, for the classical Dedekind sum, they give a new upper bound.
The following is a detailed description of the problems that this paper attempts to solve:
1. **Define Higher - dimensional Dedekind sums**:
Starting from the classical definition of the one - dimensional Dedekind sum, the authors generalize it to the \(d\)-dimensional cube. For positive integers \(a\) and \(b_1, b_2,\ldots, b_d\), they define the generalized Dedekind sum \(S_d(a; b_1, b_2,\ldots, b_d)\). These sums can be represented by mixed moments, and the specific form is:
\[
S_d(a; b_1, b_2,\ldots, b_d)=\frac{1}{a}\sum_{m = 0}^{a - 1}\left\lfloor\frac{mb_1}{a}\right\rfloor\cdots\left\lfloor\frac{mb_d}{a}\right\rfloor
\]
2. **Study frequency distribution**:
By analyzing the frequency distribution of lattice points marked along the main diagonal in the \(d\)-dimensional cube, the authors derive new formulas in some special cases and provide several types of upper and lower bounds. For example, in the two - dimensional case, \(S_2(a; b_1, b_2)\) can be represented by the inner product of two vectors, and \(R_2(a; b_1, b_2)\) is the cosine value of the angle between these two vectors.
3. **Propose a conjecture**:
Based on experimental data, the authors propose a conjecture: for all \(a, b, c\geq3\),
\[
R_2(a; b, c)\geq R_2(5; 2, 3)=\frac{\sqrt{3}}{2}
\]
This means that all these vectors are located within a cone with a maximum angle of \(\cos^{-1}\left(\frac{\sqrt{3}}{2}\right)=\frac{\pi}{6}\).
4. **Provide high - dimensional upper bounds**:
The authors use the Cauchy - Schwartz inequality to provide upper bounds based on one - dimensional moments for the high - dimensional Dedekind sum \(S_d(a; b_1, b_2,\ldots, b_d)\). For example, for \(d = 3\),
\[
S_3(a; b_1, b_2, b_3)\leq M_2(a; b_1)^{1/2}M_4(a; b_2)^{1/4}M_4(a; b_3)^{1/4}
\]
And through symmetrization, a more concise symmetric upper bound is obtained.
In summary, the core problem of this paper is to generalize the Dedekind sum to high - dimensional space and study its properties through frequency distribution and geometric methods, thereby providing new formulas and bounds for these sums.