[Clp] Infinite loop in Clp

Kim Hansen kimhanse at gmail.com
Tue May 6 12:15:59 EDT 2008


On Mon, May 5, 2008 at 5:02 PM, John J Forrest <jjforre at us.ibm.com> wrote:
> Kim,
>
> Even the run with an mps file does not look quite right.
>
> a) Refresh your version of trunk - many changes have been made.

That fixed the problem, thanks.

> b) use writeMpsNative with formatType 2 to get an mps file with exactly the
> same values
>
> If you want you can send the mps fiel to me and I can look at run.

I have attached the file in the standard format as the clp program
finds the wrong solution when I use that file:

kim at raph:~/orion/svn/trunk/solver$ clp x.mps
Coin LP version 1.06.00, build May  6 2008
command line - clp x.mps
At line 1 NAME          BLANK
At line 2 ROWS
At line 188 COLUMNS
At line 1805 RHS
At line 1896 BOUNDS
At line 2073 ENDATA
Problem BLANK has 184 rows, 257 columns and 3019 elements
Model was imported from ./x.mps in 0.004 seconds
Presolve 109 (-75) rows, 120 (-137) columns and 1921 (-1098) elements
Perturbing problem by 0.001 % of 4.49537 - largest nonzero change
1.44776e-05 (% 22.5473) - largest zero change 1.44096e-05
0  Obj -1.22719e+11 Primal inf 1.22108e+18 (69)
31  Obj 1.28446e+15 Primal inf 1.74144e+07 (39)
33  Obj 1.47474e+15 Primal inf 2.05988e+13 (39)
Primal infeasible - objective value 1.47474e+15
Presolved problem not optimal, resolve after postsolve
After Postsolve, objective 1.60904e+15, infeasibilities - dual 433739
(35), primal 1.88861e+14 (114)
PrimalInfeasible objective 1.609035234e+15 - 66 iterations time 0.022,
Presolve 0.02
kim at raph:~/orion/svn/trunk/solver$

The real solution has an objective of 7.49029e+06, glpsol and the
simple clp solver I have written agree on that.

-- 
Kim Hansen
Vadgårdsvej 3, 2.tv
2860 Søborg
Fastnet: 3956 2437 -- Mobil: 3091 2437
-------------- next part --------------
A non-text attachment was scrubbed...
Name: x.mps.bz2
Type: application/x-bzip2
Size: 12959 bytes
Desc: not available
URL: <http://list.coin-or.org/pipermail/clp/attachments/20080506/03bdac0d/attachment.bz2>


More information about the Clp mailing list