Reviewer, 2008-07-29 08:55:00

Linear programming environment to solve constrained optimization problems arising in various industrial, financial and educational areas. Optimization problems are stated as linear programs with UNLIMITED number of variables and constraints.

Pros: GIPALS or the general interior point linear solver solves constrained optimization problems using the interior point method in linear programming. This is otherwise known as the Mehrotra predictor-corrector algorithm. The application addresses solving of constrained optimization problems faced in the industry, financial and educational areas. Constrained optimization problems are stated as linear programs with unlimited number of decision variables and constraints. The algorithm implemented here is optimized for large sparse linear programs and the algorithm orders the constraints matrix efficiently.

The interface is simple and wizard-like. The user can click on one of the main tabs to be able to set the environment for calculations in an optimization run. The tabs are Objective type (set if minimization or maximization is required), title, size (define number of variables, constraints and non-zeros involved). Results display the primal objective and dual objective values along with performance tab that shows the time taken and number of iterations to arrive at the optimization. Users can define the 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 where non-zeros prevail over zeros. Constraints can directly be imported from spreadsheets via the clipboard. Other main areas of the display variable, constraints and results in separate areas.

The sparse form has a large number of linear programs that can run into tens/hundreds of thousands. Here non-zero values would be few and far between. In this scenario linear programs can be imported from mathematical systems like mathematical programming system or the MPS, data format for describing linear programs. The solution of the linear programs can be saved as CSV (comma-delimited spreadsheet), Tab-delimited or HTML file.

Cons: No obvious problems here.

Overall: The interface handles a fairly complex problem very simply and lets it become easy to use. This is a 4 star package alright.