Predictor Corrector algorithms for Linear Programming
Action MOCOA, INRIA-Rocquencourt
Beta version, November 2000
Frederic BONNANS, INRIA-Rocquencourt
Mounir HADDOU, INRIA-Rocquencourt and Université d'Orléans
The software contains some functions and drivers for
solving LP problems of the form
min c'x s.t Ax=b; x>=0
by a large neihghborhood infeasible predictor_corrector algorithm.
It is based on Newton steps on the perturbed optimality system
x.*s = m * 1
Ax = b
c + A'lambda = s
x>= 0 , s>=0
The matrix A may be either full or sparse; computations are made
- This is a software based on either
for solving large scale linear programming problems.
It can be freely used for non commercial use.
- DOWNLOAD: pclp.tar.gz
- gunzip pclp.tar.gz
- tar xvf pclp.tar
- The demo macros are tpclp and tpclpm, with suffix .sci
The matlab macros are in the matlab directory.
MORE INFORMATION: see Part IV of the book