GIPALS is linear programming environment that incorporates large-scale linear programs solver and easy, intuitive graphical user interface to direct specify or import and solve any type of constrained optimization problems arising in various industrial, financial and educational areas.
A constrained optimization problem is stated as a linear program that size can reach up to 15,000 decision variables and constraints. The linear program solver is based on interior-point method (Mehrotra predictor - corrector algorithm) and optimized for large sparse linear programs.
The solver exploits a sparsity of the constraint matrix by implementing state-of-art ordering algorithm to preserve the matrix sparsity and hence reduce the calculation time dramatically.
Almost every stage of the linear programming calculation can be saved in the debug files or traced for the deep analysis. The user can specify the linear program constraints in a dense form using the grids or in a sparse form using the particular constraints editor.
The dense form is suitable for small and medium linear programs with non-zeros prevail over zeros. In this form the constraints can be directly copied/pasted from/to MS Excel spreadsheets by Windows clipboard.
The sparse form is designed to specify / view / edit large linear programs with tens and hundreds of thousand variables and constraints. Any constraint or variable can be disabled / enabled from the calculation by the user at any time.
GIPALS can import linear programs from Mathematical Programming System (MPS) data format that is an industry standard for the description of a variety of linear programs.
Any linear program specified in GIPALS user interface could be exported to MPS format. The solution of the linear programs can be saved as CSV (comma-delimited spreadsheet), Tab-delimited or HTML file.
Key features of GIPALS: Simple and natural way to specify a linear program without any special mathematical knowledge Robust Interior-Point method for fast and reliable solution Support the industrial standard format of linear programs Report the solutions in widely used formats including spreadsheets and HTML.