Hi Xypron,
> @ARTICLE{Gondzio94presolveanalysis,
> author = {Jacek Gondzio},
> title = {Presolve Analysis of Linear Programs Prior to Applying an
> Interior Point Method},
> journal = {INFORMS Journal on Computing},
> year = {1994},
> volume = {9},
> pages = {73--91}
> }
> http://www.maths.ed.ac.uk/~gondzio/software/presolve.ps
> gives a nice overview of preprocessing for linear programming.
I have this article in my library. Note that some transformations
described there are important only for interior point methods; for
example, the simplex method is not sensitive to dense columns.
> Most of the ideas have already been implemented in GLPK.
>
> According to my understanding the following are not yet implemented in GLPK:
> - identify linear dependent rows and columns, especially duplicates
> - improving sparsity
>
> Would you consider these preprocessing steps worthwhile to implement?
I will think how to implement new features of glpk in a more systematic
way. Currently I am working over some improvements in the branch-and-cut
solver, and it is difficult for me to switch between different tasks.
Best regards,
Andrew Makhorin
_______________________________________________
Help-glpk mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-glpk
