[Cbc] Bug report: Problems on CORAL models

[Emilie Danna]

Hi,

I mentioned at INFORMS that I ran into problems when running CBC on
CORAL models. Here are more details.
I had a crash for neos-555343 that seems quite consistent.
I also had non-deterministic crashes for neos-955800 and neos-506422.
These crashed do not happen at every run.
The logs are attached (bad = log with crash, good = successful log)

I've built CBC 2.5.0 on x86_64 without any non-default options.
You can see from the log that I changed some tolerances to non-default values.

I had other problems with CORAL models but, on closer inspection of
the logs, it turns out that these problems are not crashes but lack of
respect of time limit. My benchmarking script specifies a hard time
limit with the UNIX 'ulimit' command. I gave CBC a time limit of 100
seconds and my script a hard time limit of 200 seconds. So, if CBC
didn't respect the time limit, then it was killed by my script and it
didn't return any solve statistics, which I incorrectly interpreted as
a crash. My apologies for incorrectly characterizing the problem.

CBC goes over the time limit by more than 100 seconds on the following models:

neos-1140050.mps
neos-1367061.mps
neos-1407044.mps
neos-1420546.mps
neos-1429212.mps
neos-1430811.mps
neos-1603965.mps
neos-476283.mps
neos-506428.mps
neos-520729.mps
neos-565672.mps
neos-619167.mps
neos-631710.mps
neos-738098.mps
neos-799716.mps
neos-824661.mps
neos-826224.mps
neos-826812.mps
neos-827175.mps
neos-839894.mps
neos-848589.mps
neos-856059.mps
neos-872648.mps
neos-873061.mps
neos-876808.mps
neos-941313.mps
npmv07.mps
ramos3.mps

Unfortunately, I don't have log files for these models. There are
other models where CBC does over the time limit, but not as much:

For a time limit of 100 seconds, here is the "total time" reported on
the "Result - " line at the end of the log.

neos-957270.mps.log total time 110.90
neos-1126860.mps.log total time 110.93
sp98ar.mps.log total time 111.31
rlp1.mps.log total time 111.37
neos-839838.mps.log total time 111.66
neos-1171448.mps.log total time 111.73
neos-935234.mps.log total time 111.77
neos-957389.mps.log total time 111.80
neos-1122047.mps.log total time 111.95
neos-1423785.mps.log total time 112.06
neos-937446.mps.log total time 112.67
neos-1096528.mps.log total time 112.81
neos-948126.mps.log total time 113.65
neos9.mps.log total time 113.69
neos-930752.mps.log total time 114.09
neos-935348.mps.log total time 114.09
neos-1605075.mps.log total time 114.48
neos-960392.mps.log total time 114.62
neos-954925.mps.log total time 116.04
neos-937815.mps.log total time 116.23
neos-934441.mps.log total time 117.47
dano3_4.mps.log total time 118.13
neos-984165.mps.log total time 118.64
neos-631784.mps.log total time 121.19
prod2.mps.log total time 121.31
neos-935627.mps.log total time 122.79
neos-950242.mps.log total time 122.84
neos-1354092.mps.log total time 124.30
neos-578379.mps.log total time 124.45
neos-948346.mps.log total time 125.70
seymour.disj-10.mps.log total time 126.29
neos-1620770.mps.log total time 126.95
dano3_5.mps.log total time 127.04
neos-799711.mps.log total time 127.98
neos-957323.mps.log total time 133.50
neos-885086.mps.log total time 133.83
neos-956971.mps.log total time 136.27
neos-1597104.mps.log total time 141.40
neos-933966.mps.log total time 142.32
neos-824695.mps.log total time 145.89
neos-631709.mps.log total time 148.60
qap10.mps.log total time 149.00
neos-691073.mps.log total time 149.55
neos-933638.mps.log total time 152.74
neos-691058.mps.log total time 155.52
neos-702280.mps.log total time 155.77
neos8.mps.log total time 155.96
neos-631694.mps.log total time 157.59
neos-827015.mps.log total time 160.92
neos-1605061.mps.log total time 164.58
neos-934278.mps.log total time 164.77
neos-934531.mps.log total time 175.30
neos-937511.mps.log total time 175.57
neos10.mps.log total time 182.35
neos-936660.mps.log total time 189.54
neos-1215891.mps.log total time 190.40
neos-780889.mps.log total time 190.56

I attach a few log files. The most common case seems to be the cut
generation running over the time limit.

Hope this helps.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-955800-good.mps.log
Type: text/x-log
Size: 12663 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0007.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-955800-bad.mps.log
Type: text/x-log
Size: 18867 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0008.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-506422-bad.mps.log
Type: text/x-log
Size: 8428 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0009.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-506422-good.mps.log
Type: text/x-log
Size: 14573 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0010.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-555343-bad.mps.log
Type: text/x-log
Size: 12281 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0011.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-934278-timelimit.mps.log
Type: text/x-log
Size: 6358 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0012.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neos-780889-timelimit.mps.log
Type: text/x-log
Size: 3511 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20101116/ee267631/attachment-0013.bin