ResearchResearch paperTraining & Scaling · Reinforcement Learning1 source · Oct 19, 2026

Learning SOCP Solver Dynamics with Graph Neural Networks

We introduce SOCPNet, a recursive GNN framework that uses solver trajectories to learn an iterative SOCP solution procedure.

Second-order cone programs (SOCPs) are a fundamental class of convex optimization problems widely applied across engineering domains.
This has motivated learning-based solvers that approximate optimizer solutions with neural networks.
Recent work has increasingly studied graph neural networks (GNNs) for linear, quadratic, and second-order cone optimization, leveraging their ability to model sparse relational structure among variables, constraints, and objective functions.
Finally, we demonstrate how trained SOCPNet models can be used in a hybrid optimization pipeline by warm-starting a classical interior-point solver, reducing solver iterations without loss of accuracy.

[1]Learning SOCP Solver Dynamics with Graph Neural Networks
arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 12:37 PM
We introduce SOCPNet, a recursive GNN framework that uses solver trajectories to learn an iterative SOCP solution procedure.
Second-order cone programs (SOCPs) are a fundamental class of convex optimization problems widely applied across engineering domains.

Extractive summary: sentences quoted from the sources.

Before this