Software and Applications

Barzilai-Borwein Method

BBQ is a gradient algorithm to solve unconstrained and box-constrained optimizaiton problems.

BBQ Version 1.0 (Source Code)

Main Reference: Y. Huang, Y.H. Dai and X.W. Liu (2020), Equipping Barzilai-Borwein Method with Two Dimensional Quadratic Termination Property.

Nonlinear Conjugate Gradient Methods

CGOPT is an efficient conjugate gradient software for large-sclae unconstrained optimization.

CGOPT Version 1.0 (Source Code and Test Problem Data)

Main Reference: Y.H. Dai and C.X. Kou (2013), A Nonlinear Conjugate Gradient Algorithm with An Optimal Property and An Improved Wolfe Line Search, SIAM Journal on Optimization, 23:1, pp. 296-320.

CGOPT Version 2.0 (Source Code; Test Problem Data)

Main Reference: Z. Liu, H. Liu and Y.H. Dai (2020), An Improved Dai-Kou Conjugate Gradient Algorithm for Unconstrained Optimization, Computational Optimization and Applications, 75, pp. 145-167.

CGOPT (2.0) is an improved version of the efficient conjugate gradient software package CGOPT. The main differences between CGOPT and CGOPT (2.0) are that CGOPT (2.0) uses a quasi-Newton method to improve the orthogonality of the gradients when the loss of the orthogonality is detected.

Mixed Integer Programming and Applications

CMIP is a mixed integer programming solver based on the branch-and-cut framework. It has more than sixty thousand lines of code, covering all the basic modules including preprocessing, heuristic, cutting plane, branch, node selection, and domain propagation. The first version is announced on April, 2018.

With CMIP as a main platform, our group has successfully solved many real problems including rocket trajectory planning, unit commitment problem in power system, pooling problem, steady-state gas network problem, base station location problem, AGV dock distribution problem, heated oil pipeline problem, unsplittable network design problem, and virtual machine placement problem. The group won the OR Application Award, which was awarded by the Operations Research Society of China in 2018.

Though our theoretical studies on integer programming is going smoothly (see the list of publications for some progress), more efforts are obviously much required. A webpage of ‘‘CMIP with Applications’’ is coming soon.