[SCIP] Fwd: pricing in problem variables

Gerald Gamrath gamrath at zib.de
Wed May 17 08:46:48 CEST 2017 

Hi James,

just a short addition: did you leave the verbosity level at default 
(high)? Then you could change it to full and see a display line after 
each pricing round. If you already set it to full before, we will need 
to investigate further to find the reason for the behavior you are 
experiencing.

Best,
Gerald

Am 16.05.2017 um 21:05 schrieb Gerald Gamrath:
> Hi James,
>
> sorry for the late reply. Your explanation with the cuts sounds 
> reasonable and could well cause this behavior, but is not the reason 
> here, since the cuts are only added a bit later.
> I also needed to have a look in the code to find the cause, but in the 
> end, it's quite simple: SCIP does not print a display line after each 
> pricing round, but only if nrounds % 100 == 1 (and after the pricing 
> round is finished, after cutting, etc.). So, the second line you see 
> is after a few more pricing rounds and you end up at 30.000 variables 
> at that point even if only 2000 are added each round.
>
> Best,
> Gerald
>
> Am 15.05.2017 um 11:55 schrieb James Cussens:
>> Hi folks,
>>
>> Regarding my question below. I suspect now that the reason I end up 
>> with so many new columns is that the relevant variables are in the 
>> cuts I am adding. Does that make sense?
>>
>> James
>>
>>
>> ---------- Forwarded message ----------
>> From: *James Cussens* <james.cussens at york.ac.uk 
>> <mailto:james.cussens at york.ac.uk>>
>> Date: 12 May 2017 at 15:13
>> Subject: pricing in problem variables
>> To: scip at zib.de <mailto:scip at zib.de>
>>
>>
>> I am experimenting with setting my problem variables to be 
>> non-initial and having them 'priced-in'. See  below for an example of 
>> what is going on.
>>
>> The line where we get 2000 cols makes sense, since this is the 
>> (default) value of pricing/maxvarsroot = 2000. However on the next 
>> iteration we end up with ~30k columns.
>> I don't see why we don't just end up with 4000. Can anyone illuminate me?
>>
>> Note that since I have also marked my variables as 'removable' most 
>> of them get (perhaps temporarily) removed a little later on.
>>
>> James
>>
>> .....
>> 1 constraints of type <dagcluster>
>> Presolving Time: 14.87
>>
>>  time | node  | left  |LP iter| mem |depth|mdpt |frac |cons |cols 
>> |rows |cuts |pool |confs|strbr|  dualbound   | primalbound  |  gap
>>  16.6s|     1 | 0 |     2 | 249M|   0 |   0 |   - |1275 |2000 |  23 | 
>>   0 |   0 |   0 |   0 | 0.000000e+00 |      --      |    Inf
>>  16.7s|     1 | 0 |    23 | 250M|   0 |   0 |   0 |1275 |  30k|  23 | 
>>   0 |   0 |   0 |   0 |-6.120109e+04 |      --      |    Inf
>> k16.8s|     1 | 0 |    23 | 250M|   0 |   0 |   0 |1275 |  30k|  23 | 
>>   0 |   0 |   0 |   0 |-6.120109e+04 |-8.017084e+04 |  31.00%
>>  17.1s|     1 | 0 |   210 | 261M|   0 |   0 |   0 |1275 |  30k| 149 | 
>> 126 |   8 |   0 |   0 |-6.120109e+04 |-8.017084e+04 |  31.00%
>>  17.2s|     1 | 0 |   210 | 261M|   0 |   0 |   0 |1275 |  30k| 149 | 
>> 126 |   8 |   0 |   0 |-6.396899e+04 |-8.017084e+04 |  25.33%
>>  18.1s|     1 | 0 |   444 | 288M|   0 |   0 |  58 |1275 |  30k| 288 | 
>> 265 |   9 |   0 |   0 |-6.396899e+04 |-8.017084e+04 |  25.33%
>>  18.3s|     1 | 0 |   463 | 288M|   0 |   0 |  55 |1275 |  30k| 288 | 
>> 265 |   9 |   0 |   0 |-6.566947e+04 |-8.017084e+04 |  22.08%
>> k18.4s|     1 | 0 |   463 | 288M|   0 |   0 |  55 |1275 |  30k| 288 | 
>> 265 |   9 |   0 |   0 |-6.566947e+04 |-7.959231e+04 |  21.20%
>>  19.9s|     1 | 0 |  1144 | 299M|   0 |   0 | 136 |1275 |  30k| 366 | 
>> 343 |  10 |   0 |   0 |-6.566947e+04 |-7.959231e+04 |  21.20%
>>  20.4s|     1 | 0 |  1211 | 299M|   0 |   0 | 152 |1275 |  30k| 366 | 
>> 343 |  10 |   0 |   0 |-6.640382e+04 |-7.959231e+04 |  19.86%
>>  22.1s|     1 | 0 |  1782 | 301M|   0 |   0 | 207 |1275 |  30k| 442 | 
>> 419 |  11 |   0 |   0 |-6.640382e+04 |-7.959231e+04 |  19.86%
>>  22.7s|     1 | 0 |  1851 | 301M|   0 |   0 | 221 |1275 |  30k| 442 | 
>> 419 |  11 |   0 |   0 |-6.643694e+04 |-7.959231e+04 |  19.80%
>> k22.8s|     1 | 0 |  1851 | 301M|   0 |   0 | 221 |1275 |  30k| 442 | 
>> 419 |  11 |   0 |   0 |-6.643694e+04 |-7.919237e+04 |  19.20%
>>  24.3s|     1 | 0 |  2116 | 416M|   0 |   0 | 197 |1275 | 525 | 416 | 
>> 469 |  38 |   0 |   0 |-6.643694e+04 |-7.919237e+04 |  19.20%
>> ....
>>
>>
>> -- 
>> James Cussens
>> Dept of Computer Science &
>> York Centre for Complex Systems Analysis
>> Room 326, The Hub, Deramore Lane            Tel +44 (0)1904 325371 
>> <tel:+44%201904%20325371>
>> University of York      Fax +44 (0)1904 500159 <tel:+44%201904%20500159>
>> York YO10 5GE, UK http://www.cs.york.ac.uk/~jc 
>> <http://www.cs.york.ac.uk/%7Ejc>
>> http://www.york.ac.uk/docs/disclaimer/email.htm 
>> <http://www.york.ac.uk/docs/disclaimer/email.htm>
>>
>>
>>
>> -- 
>> James Cussens
>> Dept of Computer Science &
>> York Centre for Complex Systems Analysis
>> Room 326, The Hub, Deramore Lane            Tel    +44 (0)1904 325371
>> University of York  Fax  +44 (0)1904 500159
>> York YO10 5GE, UK http://www.cs.york.ac.uk/~jc 
>> <http://www.cs.york.ac.uk/%7Ejc>
>> http://www.york.ac.uk/docs/disclaimer/email.htm
>>
>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> https://listserv.zib.de/mailman/listinfo/scip
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20170517/16bdec58/attachment.html>

More information about the Scip mailing list