Linear programming. The linear programming algorithms are: primal simplex, dual simplex, and network simplex. Matrix supersparsity is uniquely exploited to minimize memory consumption; this lets you solve a larger problem in a smaller machine. An automatic presolve process reduces a model to its essential core; this dramatically reduces optimization time. Also, C-WHIZ chooses between two factorization algorithms, P4 or Markowitz, depending on the size and difficulty of the model.

C-WHIZ Presolve. Large automatically generated models often contain structures that would not be there if you could carefully screen them out. C-WHIZ does that screening for you and usually reduces the model size dramatically with an even greater reduction in solution time. The presolve looks for redundancies, degeneracies, singleton rows and columns, transfer columns, duplicate rows, non-constraining rows, infeasibilities, ... Of equal importance is the C-WHIZ Postsolve that restores all of this structure and extends the optimal basis to conform to the complete model.

Input, Output. C-WHIZ accepts standard MPS format and equation form input from many application languages. A more efficient MPSIII proprietary format is available; it is produced directly by DATAFORM and OML. A wide variety of outputs (Solution, Ranging, ...) are produced in formats to meet your needs: text, binary, and proprietary formats.

A Windows GUI provides a familiar front-end to the optimizers, it is particularly useful when you are exploring the solution controls for a new or troublesome model. The GUI gives you access to several utilities including a powerful matrix editor, MATEDIT, for viewing and even modifying your matrix.

Fact Sheet. For a two page fact sheet with more detailed information, just select C-WHIZ Facts. You may wish to print a copy for your files.