CurveLP

CurveLP.m is a software for Linear Programming using arc-search infeasible interior-point method.

CurveLP.m implements an arc-search infeasible interior-point algorithm for standard Linear Programming problems proposed in [Yaguang Yang] CurveLP-A MATLAB implementation of an infeasible interior-point algorithm for linear programming. Numerical Algorithms, (DOI 10.1007/s11075-016-0180-1), 1-30, 2016. It is very similar to the popular Mehrotra's algorithm except that CurveLP.m searches the optimizer along an arc while Mehrotra's algorithm searches the optimizer along a straight line [S. Wright] Primal-Dual Interior-Point Methods, SIAM, Philadelphia, 1997. To compare the performance of these two algorithms, an implementation of Mehrotra's algorithm is also included (mehrotra.M). The two algorithms use the same initial point, the same pre-process and the same post-process, the same parameters, and the same terminate criterion, which are addressed in detail in [Yaguang Yang]. Users can compare the performance of the two algorithms by themselves for their problems. Users can also check the differences of the two codes by software such as Beyond Compare. The feedback on the comparison is very welcome. Bug report should be sent to the author.