Totally $Δ$-Modular Tree Decompositions of Graphic Matrices for Integer Programming
Caleb McFarland
TL;DR
This work introduces the totally Δ-modular treewidth (TDM-treewidth) as a parameter for matrices with two nonzero entries per row, unifying signed-graph and rooted-graph perspectives. It establishes a polynomial-time solvability framework for integer programs when the matrix has bounded TDM-treewidth and the decision variables lie in a bounded interval, with a runtime of the form d^{O(k)} n^{f(k)}. Extending beyond coefficients in {−1,0,1}, the authors connect TDM-treewidth to K-free treewidth and tame OCP-treewidth, providing a grid-minor structure theorem via rooted signed graphs and a corresponding DP-based algorithm. The paper also presents an approximate grid theorem and shows how these decompositions yield tractable IPs for a broad class of totally Δ-modular matrices, highlighting deep connections to MIS in odd-minor-closed classes and outlining several future avenues, including relaxations for {0,1}-variables and bounded-entry versus unbounded-variable regimes.
Abstract
We introduce the tree-decomposition-based parameter totally $Δ$-modular treewidth (TDM-treewidth) for matrices with two nonzero entries per row. We show how to solve integer programs whose matrices have bounded TDM-treewidth when variables are bounded. This extends previous graph-based decomposition parameters for matrices with at most two nonzero entries per row to include matrices with entries outside of $\{-1,0,1\}$. We also give an analogue of the Grid Theorem of Robertson and Seymour for matrices of bounded TDM-treewidth in the language of rooted signed graphs.
