[Dip-tickets] [Dip] #84: Compress columns appears to cause infeasibility

[Dip]

#84: Compress columns appears to cause infeasibility
-------------------+--------------------------------------------------------
Reporter:  tkr     |     Owner:  mgalati
    Type:  defect  |    Status:  new    
Priority:  major   |   Version:  trunk  
Keywords:          |  
-------------------+--------------------------------------------------------
 Revision 479, Linux with gcc 4.4, CPLEX as LP/IP solver. Solving att48
 with the TSP app with these settings

 {{{
 doPriceCut    = 1
 LimitRoundCutIters = 0
 TailoffPercent = -1
 ModelNameCore  = 2MATCH
 ModelNameRelax = SUBTOUR
 }}}

 The master LP becomes infeasible after column compression, even though no
 cuts are being generated. Note the switch from PHASE_PRICE2 back to
 PHASE_PRICE1 below.

 {{{
 Processing Node   0 algo= PRICE_AND_CUT phase= PHASE_PRICE2 c=    0 p=
 2008 LB=   6890.177 UB=  10752.290 nodeLB=  10489.780 gLB=  10489.780 gUB=
 INF lpGap=      0.025 ipGap=        INF time=      30.46
 Processing Node   0 algo= PRICE_AND_CUT phase= PHASE_PRICE2 c=    0 p=
 2009 LB=   9780.191 UB=  10752.179 nodeLB=  10489.780 gLB=  10489.780 gUB=
 INF lpGap=      0.025 ipGap=        INF time=      30.47
 Processing Node   0 algo= PRICE_AND_CUT phase= PHASE_PRICE1 c=    0 p=
 2010 LB=    -30.115 UB=      0.000 nodeLB=    -30.115 gLB=    -30.115 gUB=
 INF lpGap=      1.000 ipGap=        INF time=      30.49
 Processing Node   0 algo= PRICE_AND_CUT phase= PHASE_PRICE2 c=    0 p=
 2011 LB=  -2106.519 UB=  10752.182 nodeLB=  -2106.519 gLB=  -2106.519 gUB=
 INF lpGap=      6.104 ipGap=        INF time=      30.51
 Processing Node   0 algo= PRICE_AND_CUT phase= PHASE_PRICE2 c=    0 p=
 2012 LB=   7390.329 UB=  10752.173 nodeLB=   7390.329 gLB=   7390.329 gUB=
 INF lpGap=      0.455 ipGap=        INF time=      30.52
 }}}

-- 
Ticket URL: <https://projects.coin-or.org/Dip/ticket/84>
Dip <https://projects.coin-or.org/Dip>
An extensible software framework for implementing decompositon-based bounding algorithms for use in solving large-scale discrete optimization problems.