Changes between Initial Version and Version 1 of Ticket #18735, comment 7


Ignore:
Timestamp:
06/27/15 00:56:26 (6 years ago)
Author:
mkoeppe
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #18735, comment 7

    initial v1  
    111112  7.55           101.72            11.32    1252.91         808.73
    121213  7.21           279.08            13.57    1424.28        1019.95
    13 14  8.41           562.97            15.91                   1628.54
    14 15 13.10           550.46            18.48                   2550.94
     1314  8.41           562.97            15.91    7343.37        1628.54
     1415 13.10           550.46            18.48    3667.93        2550.94
    1515}}}
    1616
    1717As you can see, PPL is much slower than pure glp_exact, and orders of magnitudes slower than glp_simplex followed by glp_exact.
    1818
    19 However, currently when we try to reconstruct the solution from the combinatorial basis information, Sage's super slow matrix functions over the rationals get us back to the same order of magnitude as PPL.
     19However, currently when we try to reconstruct the solution from the combinatorial basis information, Sage's super slow matrix functions over the rationals get us back to roughly the same order of magnitude as PPL.
    2020
    2121It would be interesting to know how the solvers perform on the kind of LPs that you have in mind.