Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Self-Supervised Learning for Sparse Matrix Reordering

About

Rearranging the rows or columns of a sparse matrix using an appropriate ordering can significantly reduce fill-ins, i.e., new nonzeros introduced during matrix factorization, decreasing memory usage and runtime. However, finding an ordering that minimizes fill-ins is NP-complete. Existing approaches, including graph-theoretic and deep learning methods, rely on surrogate objectives without theoretical guarantees. The Fill-Path Theorem reveals a direct and intrinsic relationship between fill-in generation and the sparse structure of the matrix as path triplet inequalities. Here we first employ a multigrid graph network to capture structural information for each vertex. We then derive a triplet sampling strategy based on inequalities. Finally, we introduce an end-max chain loss function to reduce the number of triplets whose predicted scores satisfy these inequalities. Experimental evaluations on the publicly available SuiteSparse matrix collection demonstrate the superiority of the proposed method in terms of both fill-in reduction and speedup in LU factorization time.

Ziwei Li, Tao Yuan, Fangfang Liu, Shuzi Niu, Huiyuan Li, Wenjia Wu• 2026

Related benchmarks

TaskDatasetResultRank
Matrix Reordering for LU FactorizationBenchmark Test Matrices 2D3D
LU Factorization Time (s)91.9
8
Matrix Reordering for LU FactorizationBenchmark Test Matrices CFD
LU Factorization Time (s)99.91
8
Matrix Reordering for LU FactorizationBenchmark Test Matrices MRP
LU Factorization Time (s)7.83
8
Matrix Reordering for LU FactorizationBenchmark Matrices SP (test)
LU Factorization Time (s)76.43
8
Matrix Reordering for LU FactorizationBenchmark Test Matrices TP
LU Factorization Time (s)525
8
Matrix Reordering for LU FactorizationBenchmark Test Matrices All
LU Factorization Time (s)95.23
8
Sparse Matrix ReorderingSuiteSparse matrix collection (test)
CFD57.87
8
Showing 7 of 7 rows

Other info

Follow for update