[Ipopt] (no subject)

[Stefan Vigerske]

Hi,

maybe the sparsity structure information about the jacobian or hessian 
that you pass to IPOPT misses some initializations.

Stefan

On 10/25/2012 12:39 PM, Stefan Brugger wrote:
> Hi everybody,
> I'm using Ipopt with Mumps in VS2012. If running in debug mode I get for my example a runtime error when Ipopt calls Mumps for solving a KKT system:
>
>
> ****
> Calling MUMPS-1 for symbolic factorization at cpu time 4.624 (wall 0.097).
> forrtl: severe (408): fort: (11): Subscript #1 of the array ICMP has value -842150450 which is less than the lower bound of 1
> ...
> ****
>
> If looking in the Ipopt console output (see attached log file) one sees that some indices seem uninitialized:
>
> ******* KKT SYSTEM *******
> (0) KKT[1][1] = 0.000000000000000e+000
> ...
> (21) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (22) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (23) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (24) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (25) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (26) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (27) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (28) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (29) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (30) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (31) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (32) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (33) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (34) KKT[-842150450][-842150450] = 0.000000000000000e+000
> (35) KKT[-842150450][-842150450] = 0.000000000000000e+000
> ...
> (59) KKT[10][10] = -0.000000000000000e+000
> ...
>
>
> Have you got any idea what went wrong?
>
> More detailed information about my environment:
> I compiled IPOPT 3.9.3 with Mumps with the v8-ifort Visual Studio projects(converted for VS2012) and stlport. I tested successfully the generated dll's with an example file from the Ipopt source tarball.
>
>
> Best regards,
> Stefan Brugger
>
>
>
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>