[Cbc] Convergence problem with miplib2010/30n20b8

Mon Jun 26 10:35:11 EDT 2017

CLP:
Problem 30n20b8 has 576 rows, 18380 columns and 109706 elements
CBS:
Cbc0038I Full problem 491 rows 8028 columns, reduced to 366 rows 423 columns

Are you sure you're comparing the same problem?

On Mon, Jun 26, 2017 at 11:30 AM, <ycollette.nospam at free.fr> wrote:

> At the beginning of the cbc run, I got this in the logs:
>
> Cbc0038I Full problem 491 rows 8028 columns, reduced to 366 rows 423
> columns
> Cbc0038I Mini branch and bound did not improve solution (1.51 seconds)
> Cbc0038I After 1.51 seconds - Feasibility pump exiting - took 0.98 seconds
> Cbc0031I 59 added rows had average density of 233.55932
> Cbc0013I At root node, 59 cuts changed objective from 43.335573 to
> 86.107549 in 20 passes
> Cbc0014I Cut generator 0 (Probing) - 283 row cuts average 172.1 elements,
> 0 column cuts (0 active)  in 1.322 seconds - new frequency is 1
> Cbc0014I Cut generator 1 (Gomory) - 15 row cuts average 743.3 elements, 0
> column cuts (0 active)  in 0.283 seconds - new frequency is -100
> Cbc0014I Cut generator 2 (Knapsack) - 0 row cuts average 0.0 elements, 0
> column cuts (0 active)  in 0.045 seconds - new frequency is -100
> Cbc0014I Cut generator 3 (Clique) - 0 row cuts average 0.0 elements, 0
> column cuts (0 active)  in 0.005 seconds - new frequency is -100
> Cbc0014I Cut generator 4 (MixedIntegerRounding2) - 50 row cuts average
> 224.3 elements, 0 column cuts (0 active)  in 0.075 seconds - new frequency
> is 1
> Cbc0014I Cut generator 5 (FlowCover) - 0 row cuts average 0.0 elements, 0
> column cuts (0 active)  in 0.092 seconds - new frequency is -100
> Cbc0014I Cut generator 6 (TwoMirCuts) - 78 row cuts average 236.0
> elements, 0 column cuts (0 active)  in 0.169 seconds - new frequency is 1
>
>
> ----- Mail original -----
> De: "ycollette nospam" <ycollette.nospam at free.fr>
> À: "cbc" <cbc at list.coin-or.org>
> Envoyé: Lundi 26 Juin 2017 16:28:31
> Objet: [Cbc] Convergence problem with miplib2010/30n20b8
>
> Hello,
>
> I am trying to run somes tests from miplib2010. And I meet some strange
> behavior.
> If I run clp on 30n20b8.mps without specific options, it converges quickly
> to a solution:
>
> At line 1 NAME           30n20b8
> At line 2 ROWS
> At line 580 COLUMNS
> At line 110291 RHS
> At line 110322 BOUNDS
> At line 128761 ENDATA
> Problem 30n20b8 has 576 rows, 18380 columns and 109706 elements
> Model was imported from /home/artelys/repository/coin/
> Cbc/build/Dependencies/.//MIPLIB2000/src/EP_MIPLIB2000/
> instances/miplib2010/30n20b8.mps.gz in 0.114319 seconds
> Presolve 493 (-83) rows, 11093 (-7287) columns and 63118 (-46588) elements
> Perturbing problem by 0.001%%of 301.52267 - largest nonzero change
> 0.00028965583 ( 9.6064362e-05%%) - largest zero change 0.00028962251
> 0  Obj 0.041886881 Primal inf 253.60538 (88)
> 78  Obj 0.044696222 Primal inf 359.4158 (159)
> 142  Obj 0.04542365 Primal inf 105.42604 (119)
> 214  Obj 0.045885186 Primal inf 83.292978 (114)
> 298  Obj 0.046155092 Primal inf 32.89186 (74)
> 380  Obj 0.048508765 Primal inf 87.130897 (64)
> 464  Obj 0.057466585 Primal inf 99.796642 (62)
> 548  Obj 0.068410705 Primal inf 276.03475 (64)
> 632  Obj 0.077664609 Primal inf 167.63095 (63)
> 716  Obj 0.08533722 Primal inf 123.90711 (47)
> 800  Obj 0.093453993 Primal inf 94.340415 (35)
> 884  Obj 1.079625 Primal inf 0.56385768 (23)
> 950  Obj 1.669254
> Optimal - objective value 1.5664076
> After Postsolve, objective 1.5664076, infeasibilities - dual 0 (0), primal
> 0 (0)
> Optimal objective 1.566407646 - 950 iterations time 0.112, Presolve 0.03
>
>
>
> Now, If I run cbc on this problem, it never stops.
> An example of the log I've got:
>
> Cbc0010I After 91600 nodes, 675 on tree, 1e+50 best solution, best
> possible 202 (1278.41 seconds)
> Cbc0010I After 91700 nodes, 626 on tree, 1e+50 best solution, best
> possible 202 (1279.79 seconds)
> Cbc0010I After 91800 nodes, 724 on tree, 1e+50 best solution, best
> possible 202 (1281.46 seconds)
> Cbc0010I After 91900 nodes, 805 on tree, 1e+50 best solution, best
> possible 202 (1283.06 seconds)
> Cbc0010I After 92000 nodes, 709 on tree, 1e+50 best solution, best
> possible 202 (1284.15 seconds)
>
>
> I feel the number of best solutions is quite strange: 1e+50 (this number
> never changes).
>
> I performed these tests on the trunk version of cbc. I will try the same
> test on an older version of cbc.
>
> Best regards,
>
> YC
> _______________________________________________
> Cbc mailing list
> Cbc at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.
> coin-2Dor.org_mailman_listinfo_cbc&d=DwICAg&c=Ngd-
> ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=kY-
> tTJAFuHScYZHjojaaKVZZOEYVVkPpGiWUPr3h80U&m=EqDbNlOmPhOlal3SI7C5XWuE8KhPe3
> sA5awi8py-lBA&s=o5HK9S3C5JZo5CHwNMumDoTjHXpYF-W3LjyehdeuF7w&e=
>
> _______________________________________________
> Cbc mailing list
> Cbc at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.
> coin-2Dor.org_mailman_listinfo_cbc&d=DwIGaQ&c=Ngd-
> ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=QLQl3jHZx6hPNKTb0I9pud73uBDZjS
> YyVOorevkAFrU&m=oRCE3ZWfhw5CzXVvq3XW5L7Hm5bKjiC3gT_F_qL8yfw&s=
> zHJEZcVYlhOOruUvsI8rMvBRloUvGYesE99_H866lr4&e=
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20170626/0b8f9ebb/attachment.html>