[Symphony-tickets] [SYMPHONY] #64: MIP 2003 benchmarks

[SYMPHONY]

#64: MIP 2003 benchmarks
-------------------------+--------------------------------------------------
 Reporter:  pchan        |       Owner:  tkr
     Type:  defect       |      Status:  new
 Priority:  normal       |   Milestone:  5.2
Component:  SPP+CUTS     |     Version:     
 Keywords:  MIPLIB 2003  |  
-------------------------+--------------------------------------------------
 From a tarball downloaded 08/02/2007, I've compiled Symphony 5.1.3 and SPP
 using
 MSVC v6. The definition of CoinInt64 is (underscore) (underscore) int64.
 See ticket #63.

 I've been running some SPP benchmarks from MIPLIB 2003
 http://miplib.zib.de/miplib2003.php

 ---------------------------------------------------------
 air04: 823 rows  8904 cols  72999 non-zeros
 Symphony/SPP/Cuts found a feasible solution with cost = 56212 after +/-
 2Hrs
 optimal solution is 56137
 ---------------------------------------------------------
 air05: 426 rows  7195 cols  52121 non-zeros
 Symphony/SPP/Cuts found a feasible solution with cost = 26495 after +/-
 2Hrs
 optimal solution is 26374
 ---------------------------------------------------------
 nw04:   36 rows 87482 cols 636666 non-zero terms
 Symphony/SPP/Cuts found no feasible solution after 8 Hrs
 optimal solution is 16862
 ---------------------------------------------------------
 These three benchmarks are reported solvable on a commercial solver within
 1Hr
 But there is no information on the CPU/memory used. So my own timings are
 approximative (I use a Pentium M at 1.4GHz 500MByte RAM)

 What borders me is that the application did not find any feasible solution
 after
 8hrs of computation. Attached is the SPP file converted from the MPS
 problem
 definition.

-- 
Ticket URL: <https://projects.coin-or.org/SYMPHONY/ticket/64>
SYMPHONY <http://projects.coin-or.org/SYMPHONY>
The SYMPHONY framework for parallel branch-and-cut algorithms.