Mixing

The Mixing method for maximum cut (MAXCUT) and maximum satisfiability (MAXSAT) problem.

Usage

./mixing [OPTIONS] INPUT
OPTIONS:
	-s SOLVER: type of solver
	           "-s maxcut" for maximum cut
	           "-s maxsat" for maximum SAT (default)
	-k RANK: rank of solution (default auto)
	         use "-k /2" to divide the rank by 2
	-e EPS: stopping threshold (default 1.0000e-03)
	-t MAX_ITER: maximum iteration (default 1000)
	             use "-t max" for INT_MAX
	-r N_TRIAL: number of trial in evaluation (default 1)
	-u: use unspeficied wcnf format
	-v: verbose

OUTPUTS:
	(factorized) SDP solution to INPUT.sol
	(when doing maxcut) rounded assignments to INPUT.rounding

NOTE:
	For better randomized rounding results for discrete MAXCUT / MAXSAT,
	pleases increase N_TRIAL (e.g., -r 10000).

To compile the file, please use

	$ make

More Info

This repository is by Po-Wei Wang, Wei-Cheng Chang, and J. Zico Kolter and contains the source code to reproduce the experiments in our paper The Mixing method: coordinate descent for low-rank semidefinite programming. If you find this repository helpful in your publications, please consider citing our paper.

@article{wang2017mixing,
	title = {The Mixing method: coordinate descent for low-rank semidefinite programming},
	author = {Po-Wei Wang and Wei-Cheng Chang and J. Zico Kolter},
	journal = {arXiv preprint arXiv:1706.00476},
	year = {2017}
}

For any questions and comments, please send your email to [email protected]