HPR-LP-C

A C implementation of the Halpern Peaceman--Rachford (HPR) method for solving linear programming (LP) problems on GPUs.

Requirements

• NVIDIA GPU with CUDA support
• CUDA Toolkit (version 12.4 or later recommended)
• cuBLAS, cuSOLVER, cuSPARSE libraries (included with CUDA Toolkit)

Building

The build system creates libraries and executables:

make clean    # Remove build artifacts
make          # Build everything (recommended)

This creates:

• lib/libhprlp.a - Static library for C/C++ linking
• lib/libhprlp.so - Shared library for language bindings (Python, Julia, MATLAB)
• build/solve_mps_file - MPS file solver executable

Usage

Solving from MPS Files

# Show help
./build/solve_mps_file -h

# Run with default settings
./build/solve_mps_file -i data/model.mps

# Run with custom settings
./build/solve_mps_file -i data/model.mps --tol 1e-4 --time-limit 3600

Using as a Library

See the examples/ directory for standalone demos showing how to link against libhprlp.a and solve LP problems programmatically.

See examples/c(cpp)/README.md for details on using HPR-LP-C in external projects.

Command-Line Options

Option	Description	Default
-i, --input <path>	Path to input MPS file	required
--device <id>	CUDA device ID	0
--max-iter <N>	Maximum iterations	unlimited
--tol <eps>	Stopping tolerance	1e-4
--time-limit <sec>	Time limit in seconds	3600
--check-iter <N>	Convergence check interval	150
--ruiz <true/false>	Ruiz scaling	true
--pock <true/false>	Pock-Chambolle scaling	true
--bc <true/false>	Bounds/cost scaling	true
-h, --help	Show help message	-

Language Interface Installation

Python (pip only):

cd HPR-LP-C/bindings/python
python -m pip install .         # or: python -m pip install -e .

Julia Interface:

cd bindings/julia
bash install.sh

MATLAB Interface:

cd bindings/matlab
bash install.sh

See the respective README files in bindings/ for detailed usage instructions.

Language Interfaces

HPR-LP-C provides native interfaces for multiple languages:

• Python: Native Python bindings via pybind11 (see bindings/python/README.md)
• MATLAB: Native MEX interface with OOP wrapper (see bindings/matlab/README.md)
• Julia: Native Julia interface (see bindings/julia/README.md)
• C/C++: Direct API usage (see examples/c/ and examples/cpp/)

License

MIT License

Copyright (c) 2025 HPR-LP Contributors

See the LICENSE file for full details.

Reference

Kaihuang Chen, Defeng Sun, Yancheng Yuan, Guojun Zhang, and Xinyuan Zhao, “HPR-LP: An implementation of an HPR method for solving linear programming”, arXiv:2408.12179 (August 2024), Mathematical Programming Computation 17:4 (2025), doi.org/10.1007/s12532-025-00292-0.

Contributors

Core Team

• Kaihuang Chen - Developer
• ZheXuan Gu - Developer
• Defeng Sun - Principal Investigator
• Yancheng Yuan - Contributor
• Guojun Zhang - Contributor
• Xinyuan Zhao - Contributor

Acknowledgments

• Community contributors and testers

Contact

For questions, issues, or collaboration inquiries:

• Open an issue on GitHub: https://github.com/PolyU-IOR/HPR-LP-C/issues

Version

Current version: 0.1.0

Available HPR-LP Implementations

For the C implementation on GPUs, please visit the following repository:

https://github.com/PolyU-IOR/HPR-LP-C

For the Julia implementation on CPUs and GPUs, please visit the following repository:

https://github.com/PolyU-IOR/HPR-LP

For the Python implementation on GPUs, please visit the following repository:

https://github.com/PolyU-IOR/HPR-LP-Python

For the MATLAB implementation on CPUs, please visit the following repository:

https://github.com/PolyU-IOR/HPR-LP-MATLAB