1. Deriving differential approximation results for $k\,$CSPs from combinatorial designs
- Author
-
Culus, Jean-François and Toulouse, Sophie
- Subjects
Mathematics - Combinatorics ,Computer Science - Computational Complexity ,90C27, 68W25, 05B30, 05B15 - Abstract
Traditionally, inapproximability results for $\mathsf{Max\,k\,CSP\!-\!q}$ have been established using balanced $t$-wise independent distributions, which are related to orthogonal arrays of strength $t$. We contribute to the field by exploring such combinatorial designs in the context of the differential approximation measure. First, we establish a connection between the average differential ratio and orthogonal arrays. We deduce a differential approximation ratio of $1/q^k$ on $(k +1)$-partite instances of $\mathsf{k\,CSP\!-\!q}$, $\Omega(1/n^{k/2})$ on Boolean instances, $\Omega(1/n)$ when $k =2$, and $\Omega(1/\nu^{k -\lceil\log_{p^\kappa} k\rceil})$ when $k\geq 3$ and $q\geq 3$, where $p^\kappa$ is the smallest prime power greater than $q$. Secondly, by considering pairs of arrays related to balanced $k$-wise independence, we establish a reduction from $\mathsf{k\,CSP\!-\!q}$ to $\mathsf{k\,CSP\!-\!k}$ (with $q>k$), with an expansion factor of $1/(q -k/2)^k$ on the differential approximation guarantee. This, combined with the result of Yuri Nesterov, implies a lower differential approximability bound of $0.429/(q -1)^2$ for $\mathsf{2\,CSP\!-\!q}$. Finally, we prove that every Hamming ball with radius $k$ provides a $\Omega(1/n^k)$-approximation of the instance diameter. Our work highlights the utility of combinatorial designs in establishing approximation results., Comment: Preliminary versions of this work have been presented or published at the ISCO 2012, ISCO 2018 and IWOCA 2018 conferences
- Published
- 2024