[Clp] OsiClp vs Clp status on infeasible problem

[Matthew Galati]

I printed out the internal CLP status after OsiClp:resolve is called - but this did not help. They all claim optimal with no error.

clp status        = 0
clp prob status   = 0
clp second status = 0


John - what is happening after the 2nd line here:

Clp0006I 209  Obj 1.38968e+08 Primal inf 2.08007e+07 (287)
Clp0001I Primal infeasible - objective value 1.38968e+08

It goes back into some solve attempt but then just gives up with very high primal infeasibility - but then returns a status of optimal.

Coin0505I Presolved problem not optimal, resolve after postsolve
Coin0511I After Postsolve, objective 1.38968e+08, infeasibilities - dual -3.06102e-06 (2), primal 8.13648e+07 (697)
Clp1001I Initial range of elements is 1 to 1787
Clp1003I Final range of elements is 0.238351 to 1
Clp0022I Absolute values of scaled rhs range from 0.699496 to 1.07021, minimum gap 1e+100
Clp0020I Absolute values of scaled objective range from 51376 to 62549
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)
Clp0006I 0  Obj 1.38968e+08 Primal inf 4.12653e+07 (697)





From: Matthew Galati [mailto:magh at lehigh.edu]
Sent: Monday, October 26, 2009 1:06 PM
To: Matthew Galati
Cc: clp at list.coin-or.org
Subject: Re: [Clp] OsiClp vs Clp status on infeasible problem

This same thing is happening in many cases now. Is there some internal CLP flag I can check to see if the internal status is set to infeasible? Perhaps there is just a bug in the transfer to the OsiClp status?




The following LP when run against CLP exe seems to correctly find the problem infeasible. However, when run inside OsiClp with resolve( ) - warm started after many other previous iterations, it declares the problem optimal (by status), despite the log which seems to show it as infeasible.

https://coral.ie.lehigh.edu/~magh/tmp/PRICE_AND_CUT_masterProb.n12.c0.p0.lp<https://coral.ie.lehigh.edu/%7Emagh/tmp/PRICE_AND_CUT_masterProb.n12.c0.p0.lp>

Log running as direct executable.

[magala at orclus71 MMKP]$ ../../../bin/clp PRICE_AND_CUT_masterProb.n12.c0.p0.lp
Coin LP version trunk, build Oct 25 2009
command line - ../../../bin/clp PRICE_AND_CUT_masterProb.n12.c0.p0.lp
Presolve 884 (-7126) rows, 53 (-8069) columns and 20791 (-78748) elements
0  Obj -67321.6 Primal inf 1.23658 (4) Dual inf 5.21256e+06 (53)
10  Obj -95950.4 Primal inf 160.616 (162)
12  Obj -93938.6 Primal inf 139.824 (204)
13  Obj -93851.8 Primal inf 131.17 (211)
16  Obj -74041.9 Primal inf 4979.62 (190)
17  Obj -30215 Primal inf 33506.3 (215)
17  Obj -30215 Primal inf 33506.3 (215)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
18  Obj -29773 Primal inf 33795.2 (211)
Primal infeasible - objective value -29773
Presolved problem not optimal, resolve after postsolve
After Postsolve, objective -29773, infeasibilities - dual 263995 (1), primal 115893 (467)
PrimalInfeasible objective -29773.04364 - 36 iterations time 0.202, Presolve 0.15


Log running within OsiClp:

Coin0506I Presolve 884 (-7126) rows, 53 (-8069) columns and 20791 (-78748) elements
Clp1001I Initial range of elements is 1 to 2581
Clp1003I Final range of elements is 0.636117 to 1
Clp0022I Absolute values of scaled rhs range from 0.774893 to 1.06895, minimum gap 1e+100
Clp0020I Absolute values of scaled objective range from 94854 to 100383
Clp0021I Absolute values of scaled bounds range from 1 to 1, minimum gap 1
Clp0006I 0  Obj -98371.9 Primal inf 1356.04 (328) Dual inf 3157.55 (17)
Clp0006I 184  Obj -97923.4 Primal inf 4676.17 (409)
Clp0006I 291  Obj 182182 Primal inf 28254 (419)
Clp0006I 292  Obj 2.37148e+06 Primal inf 246180 (416)
Clp0006I 293  Obj 3.73415e+06 Primal inf 382285 (416)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0023I Possible loop - 5 matches (1f) after 11 checks
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0006I 295  Obj 6.42176e+06 Primal inf 649494 (415)
Clp0001I Primal infeasible - objective value 6.42176e+06
Coin0505I Presolved problem not optimal, resolve after postsolve
Coin0511I After Postsolve, objective 6.42176e+06, infeasibilities - dual 0 (0), primal 1.95228e+06 (996)
Clp1001I Initial range of elements is 1 to 3215
Clp1003I Final range of elements is 0.288447 to 1
Clp0022I Absolute values of scaled rhs range from 0.622084 to 1.06895, minimum gap 1e+100
Clp0020I Absolute values of scaled objective range from 89856 to 104359
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Clp0006I 0  Obj 6.42176e+06 Primal inf 1.39856e+06 (996)
Solution update n_cols:8122       n_rows: 8010       n_iter: 295        time: 1.519
Iteration Count               : 295
isAbandoned()                 : 0
isProvenOptimal()             : 1
isProvenPrimalInfeasible()    : 0
isProvenDualInfeasible()      : 0
isPrimalObjectiveLimitReached : 0
isDualObjectiveLimitReached   : 0
isIterationLimitReached       : 0
MasterObj                     : 6.42176e+06


_______________________________________________
Clp mailing list
Clp at list.coin-or.org<mailto:Clp at list.coin-or.org>
http://list.coin-or.org/mailman/listinfo/clp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20091026/17810c97/attachment.html>