[Cbc] optimal solution cut off?

[John J Forrest]

It seems that changes in 2.2 gave a different (but equally correct) 
optimal basis at continuous solution.  A CglTwomir cut was then generated 
which cut off optimal solution by 1.0e-6.

I have modified CglTwomir to make greater allowance for very small element 
values.  That seems to fix problem on my machine anyway.

John Forrest





[Cbc] optimal solution cut off?

Ashutosh Mahajan 
to:
cbc
08/28/2008 02:34 AM


Sent by:
cbc-bounces at list.coin-or.org







hi all,
it seems the latest stable version 2.2 is giving a wrong optimal solution
value for the instance neos4. other solvers and version 2.1 show 
-4.8603e10 as
the optimal but cbc-2.2 shows -4.8331e10. it returns correct solution 
value if
i turn off twomir cuts.

the logs at mittleman's benchmarking page also seem show the wrong 
solution
value obtained by cbc:
http://plato.asu.edu/ftp/milpf_logs/CBC/neos4.cbc

--
regards
Ashutosh Mahajan
http://coral.ie.lehigh.edu/~asm4

_______________________________________________
Cbc mailing list
Cbc at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/cbc

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/cbc/attachments/20080828/aa6bb730/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 821 bytes
Desc: not available
Url : http://list.coin-or.org/pipermail/cbc/attachments/20080828/aa6bb730/attachment.gif