Opened 12 years ago

Closed 11 years ago

#21 closed defect (fixed)

does not obey of timelimit

Reported by: stefan Owned by: pbonami
Priority: minor Component: Bonmin
Version: stable/0.99 Keywords:
Cc:

Description

Hi,

when I run Bonmin B-BB (branches/0.9) with a timelimit, then it seem to check this limit not often enough. E.g., for one of the models I tried, it does 28 NLP-solves before it recognized that the time limit had been exceeded (which it was already after the first NLP solve). Also if the time limit exceeds during an Ipopt call, the Ipopt local solve is not aborted.
Looks, like the timelimit is only checked inside the Cbc part of Bonmin.

Here is the output where I've run with a timelimit of 1 second:

NLP0012I
              Num      Status      Obj             It       time
NLP0013I     1        OPT         923.794        884      2.33065
NLP0013I     2        OPT         931.072        26       0.045993
NLP0013I     3        OPT         925.063        12       0.020996
NLP0013I     4        OPT         925.759        23       0.034995
NLP0013I     5        OPT         924.206        12       0.018997
NLP0013I     6        OPT         926.207        24       0.037994
NLP0013I     7        OPT         925.209        12       0.015998
NLP0013I     8        OPT         923.963        16       0.025996
NLP0013I     9        OPT         926.13         15       0.020996
NLP0013I     10       OPT         932.797        39       0.06999
NLP0013I     11       OPT         923.794        6        0.011999
NLP0013I     12       OPT         923.794        11       0.015998
NLP0013I     13       OPT         935.598        12       0.018997
NLP0013I     14       OPT         1040.31        795      1.82472
NLP0013I     15       OPT         923.794        6        0.011998
NLP0013I     16       OPT         921.294        168      0.380942
NLP0013I     17       OPT         923.794        6        0.010998
NLP0013I     18       OPT         940.162        230      0.478927
NLP0013I     19       OPT         923.794        6        0.011998
NLP0013I     20       INFEAS      1671.53        2229     5.31719
NLP0012I
              Num      Status      Obj             It       time
NLP0013I     21       OPT         923.794        6        0.008999
NLP0013I     22       OPT         923.794        6        0.010998
NLP0013I     23       OPT         959.115        390      0.909862
NLP0013I     24       OPT         923.794        9        0.015997
NLP0013I     25       OPT         1091.91        692      1.62475
NLP0013I     26       OPT         923.794        9        0.014998
NLP0013I     27       OPT         1052.3         155      0.375943
NLP0013I     28       OPT         923.794        9        0.013998
Cbc0020I Exiting on maximum time
Cbc0005I Partial search - best objective 1e+50 (best possible 923.794), took 6 iterations and 0 nodes (11.37 seconds)
Cbc0032I Strong branching done 26 times (4918 iterations), fathomed 0 nodes and fixed 1 variables
Cbc0035I Maximum depth 0, 0 variables fixed on reduced cost
NLP0013I     29       INFEAS      942.218        716      1.9947

Best, Stefan

Change History (3)

comment:1 Changed 12 years ago by pbonami

  • Status changed from new to assigned

comment:2 Changed 12 years ago by pbonami

  • Version changed from branches/0.9 to stable/0.99

comment:3 Changed 11 years ago by pbonami

  • Resolution set to fixed
  • Status changed from assigned to closed

I fixed the issue for strong branching. For the other case (if the time limit exceeds during an Ipopt call) it is harder to do (Cbc will not recognize if internal solver has reached a time limit). Anyway don't think it is really a trouble and won't fix.

Note: See TracTickets for help on using tickets.