We prove that there exist uniform $(+,\times,/)$-circuits of size $O(n^3)$ to compute the basis generating polynomial of regular matroids on $n$ elements. By tropicalization, this implies that there exist uniform $(\max,+,-)$-circuits and ReLU neural networks of the same size for weighted basis maximization of regular matroids. As a consequence in linear programming theory, we obtain a first example where taking the difference of two extended formulations can be more efficient than the best known individual extended formulation of size $O(n^6)$ by Aprile and Fiorini. Such differences have recently been introduced as virtual extended formulations. The proof of our main result relies on a fine-tuned version of Seymour's decomposition of regular matroids which allows us to identify and maintain graphic substructures to which we can apply a local version of the star-mesh transformation.
翻译:我们证明存在规模为$O(n^3)$的均匀$(+,\times,/)$电路,用于计算$n$元正则拟阵的基生成多项式。通过热带化,这意味着存在相同规模的均匀$(\max,+,-)$电路和ReLU神经网络,用于正则拟阵的加权基最大化问题。在线性规划理论中,我们由此获得首个实例,表明取两个扩展表述的差值可能比Aprile与Fiorini提出的已知最佳$O(n^6)$规模扩展表述更高效。此类差值近期被引入为虚拟扩展表述。主要结果的证明依赖于Seymour正则拟阵分解的精细调整版本,该版本允许我们识别并保持图结构子集,从而对其应用星-网变换的局部版本。