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

T-SKM-Net: Trainable Neural Network Framework for Linear Constraint Satisfaction via Sampling Kaczmarz-Motzkin Method

About

Neural network constraint satisfaction is crucial for safety-critical applications such as power system optimization, robotic path planning, and autonomous driving. However, existing constraint satisfaction methods face efficiency-applicability trade-offs, with hard constraint methods suffering from either high computational complexity or restrictive assumptions on constraint structures. The Sampling Kaczmarz-Motzkin (SKM) method is a randomized iterative algorithm for solving large-scale linear inequality systems with favorable convergence properties, but its argmax operations introduce non-differentiability, posing challenges for neural network applications. This work proposes the Trainable Sampling Kaczmarz-Motzkin Network (T-SKM-Net) framework and, for the first time, systematically integrates SKM-type methods into neural network constraint satisfaction. The framework transforms mixed constraint problems into pure inequality problems through null space transformation, employs SKM for iterative solving, and maps solutions back to the original constraint space, efficiently handling both equality and inequality constraints. We provide theoretical proof of post-processing effectiveness in expectation and end-to-end trainability guarantees based on unbiased gradient estimators, demonstrating that despite non-differentiable operations, the framework supports standard backpropagation. On the DCOPF case118 benchmark, our method achieves 4.27ms/item GPU serial forward inference with 0.0025% max optimality gap with post-processing mode and 5.25ms/item with 0.0008% max optimality gap with joint training mode, delivering over 25$\times$ speedup compared to the pandapower solver while maintaining zero constraint violations under given tolerance.

Haoyu Zhu, Yao Zhang, Jiashen Ren, Qingchun Hou• 2025

Related benchmarks

TaskDatasetResultRank
Constrained Quadratic ProgrammingConstrained QP 100 variables, 50 equality constraints, 30 inequality constraints
Objective Value-14.05
7
Constrained Quadratic ProgrammingConstrained QP 100 variables, 50 equality constraints, 150 inequality constraints
Objective Value-10.24
7
Constrained Quadratic ProgrammingConstrained QP 100 variables, 50 equality constraints, 50 inequality constraints
Objective Value-12.53
7
Constrained Quadratic ProgrammingConstrained QP 100 variables, 50 equality constraints, 130 inequality constraints
Objective Value-10.24
7
DC Optimal Power Flow (DCOPF)IEEE 118-bus System DC Optimal Power Flow (test)
Time (ms)4.27
6
L2 ProjectionRandom Linear Constraint Systems Dim 500
Time (s)0.017
4
L2 ProjectionRandom Linear Constraint Systems Dim 2000
Time (s)0.237
4
L2 ProjectionRandom Linear Constraint Systems Dim 3000
Time (s)0.856
4
L2 ProjectionRandom linear constraint systems Dim 4000
Time (s)2.001
2
L2 ProjectionRandom linear constraint systems Dim 8000
Time (s)13.3
2
Showing 10 of 11 rows

Other info

Follow for update