Model infeasible after fixing decision variables

Dear Gurobi Community,

I am currently testing a Relax-and-Fix Algorithm in which I iteratively optimize a model for a certain period window and fix a subset of the decision variables to their optimal values. In some rare cases however this fixing results in an infeasible model. Attached you will find an exemplary IIS of one such case.

Could you tell me why the variable fixing results in the model being infeasible even though the values correspond to the found optimal values of the preceeding optimization?

\ Model paper_model_copy
\ LP format - for model browsing. Use MPS format to capture full model detail.
Minimize

Subject To
internal_capacity_arc_2134_period_30: FLOW[1622,562,563,30]
+ FLOW[1623,562,563,30] + FLOW[1624,562,563,30] + FLOW[1625,562,563,30]
+ FLOW[1626,562,563,30] + FLOW[1627,562,563,30] + FLOW[1628,562,563,30]
+ FLOW[1629,562,563,30] + FLOW[1630,562,563,30] + FLOW[1631,562,563,30]
+ FLOW[1632,562,563,30] + FLOW[1633,562,563,30] + FLOW[1634,562,563,30]
+ FLOW[1635,562,563,30] + FLOW[1636,562,563,30] + FLOW[1637,562,563,30]
+ FLOW[1638,562,563,30] + FLOW[1639,562,563,30] + FLOW[1640,562,563,30]
+ FLOW[1641,562,563,30] + FLOW[1642,562,563,30]
- 9880 SHIFT[562,563,11,30] - 10868 SHIFT[562,563,21,30]
- 19760 SHIFT[562,563,12,30] - 21736 SHIFT[562,563,22,30]
- 29640 SHIFT[562,563,13,30] - 32604 SHIFT[562,563,23,30] <= 0
Bounds
FLOW[1622,562,563,30] >= 140.7288808883422
FLOW[1623,562,563,30] >= 1009
FLOW[1624,562,563,30] >= 1213
FLOW[1625,562,563,30] >= 1138.420000010985
FLOW[1627,562,563,30] >= 1064
FLOW[1629,562,563,30] >= 1123
FLOW[1630,562,563,30] >= 847.1599999780306
FLOW[1631,562,563,30] >= 586
FLOW[1632,562,563,30] >= 3346.73712517346
FLOW[1633,562,563,30] >= 1378
FLOW[1634,562,563,30] >= 471
FLOW[1635,562,563,30] >= 973.0000000010507
FLOW[1636,562,563,30] >= 1395.794040082462
FLOW[1637,562,563,30] >= 1087
FLOW[1638,562,563,30] >= 849.1599999780296
FLOW[1639,562,563,30] >= 1096
FLOW[1640,562,563,30] >= 1099
FLOW[1641,562,563,30] >= 943
SHIFT[562,563,11,30] = 0
SHIFT[562,563,21,30] <= -2.41885e-09
SHIFT[562,563,22,30] = 0
SHIFT[562,563,13,30] <= 2.44265e-09
SHIFT[562,563,23,30] = 0
Binaries
SHIFT[562,563,11,30] SHIFT[562,563,21,30] SHIFT[562,563,12,30]
SHIFT[562,563,22,30] SHIFT[562,563,13,30] SHIFT[562,563,23,30]
End

Thank you for your help.

Best regards,

Markus

Jaromił Najman

Gurobi Staff

2022年01月19日 11:40

Hi Markus,

Does your model suffer from any numerical issues? Does the solution point have any constraint violations? Could you post the first and last ~20 lines of the log file when you are solving the non-fixed model and the fixed model?

Could you share the model and the variables which you fix? It would be great if you could provide a small executable script. You can find more on how to provide files in the Community Forum in Posting to the Community Forum.

Best regards,
Jaromił

Model infeasible after fixing decision variables

コメント

お探しのものを見つけられませんでしたか？