[Dip-tickets] [Dip] #59: (MILPBlock) "Point violates row" with large relative violation
Dip
coin-trac at coin-or.org
Fri Dec 3 01:45:18 EST 2010
#59: (MILPBlock) "Point violates row" with large relative violation
-------------------+--------------------------------------------------------
Reporter: vejlin | Type: defect
Status: new | Priority: major
Version: trunk | Keywords:
-------------------+--------------------------------------------------------
I am getting some rather large violations with the unmodified MILPBlock
example:
"Point violates row 2 -> r(2) LB= 0.0000000 ax= -1017.0750594 UB= INF
RelViol= 1017.0750594"
This happened with an unmodified MILPBlock example, the "harp2" instance
from miplib and the block file (list format):
"0 74
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 106 111"
Constraints 0-71 and 111 are column disjount Special Ordeded Sets (\sum
x_i = 1) and constraint 106 is a degenerated 0-1 Knapsack constraint
(capacity = 0) covering only a subset of the columns covered by the SOS
constraints. I am auto generating my block files so it may not be a very
good decomposition, but MILPBlock responds oddly and I cannot figure out
why.
The knapsack constraint simply fixes some variables to zero, and the
remaining SOS constraints should not be that problematic. The problem is
bounded, if nothing else then because all variables are binary.
--
Ticket URL: <https://projects.coin-or.org/Dip/ticket/59>
Dip <https://projects.coin-or.org/Dip>
An extensible software framework for implementing decompositon-based bounding algorithms for use in solving large-scale discrete optimization problems.
More information about the Dip-tickets
mailing list