This repository was used to generated the numerical results in the paper titled "Custom Solver Generation for Quadratic Objective Second-Order Cone Programs"
To run benchmarks, first create a python virtual environment and install the requirements with
pip install -r requirements.txt
To run the Maros-Meszaros problems and generate plots run
python run_maros.py && python analyze_maros.py
To run the benchmark problems, first create an empty directory called generated_solvers/
in the root of this repository then run
python generate_solvers.py && python run_benchmarks.py
Before running the benchmark problems, the user must generate the cvxgen solvers using the problem files in the cvxgen/problems/
directory and place them in a directory called generated_solvers/
within the cvxgen/
directory.
@misc{chari2025qoco,
title = {QOCO: A Quadratic Objective Conic Optimizer with Custom Solver Generation},
author = {Chari, Govind M and A{\c{c}}{\i}kme{\c{s}}e, Beh{\c{c}}et},
year = {2025},
eprint = {2503.12658},
archiveprefix = {arXiv},
primaryclass = {math.OC},
url = {https://arxiv.org/abs/2503.12658}
}