Extends MA27 insufficient memory error catching

[danieloliveira56]

Ipopt might exit without a message when MA27 memory allocation fails.

I believe Ipopt will return code -102 (Insufficient_Memory), but GAMS will convert that to solveStat 13. Then I believe there's no easy way to know what happened (would love to hear if there's a way). So would be helpful to see an error message.

Example GAMS output:

iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du alpha_pr  ls
  50  5.7126274e+01 1.09e-06 5.00e-08  -4.6 1.53e+02 -14.7 1.00e+00 1.00e+00h  1 F
  51  4.7042829e+01 6.12e-02 4.10e-03  -4.9 1.86e+04 -15.2 1.00e+00 1.00e+00h  1 S
  52  3.7594621e+01 1.41e-01 3.35e+00  -5.2 2.73e+04 -14.8 3.23e-01 1.00e+00h  1 S
  53  3.4020871e+01 1.39e-02 1.04e-03  -5.6 2.07e+04 -14.4 1.00e+00 1.00e+00h  1 
  54  3.6461406e+01 2.41e-04 5.14e-05  -5.3 2.71e+04 -14.8 1.00e+00 1.00e+00h  1 S
  55  3.3032774e+01 2.14e-05 8.50e-06  -5.6 1.39e+04 -14.4 1.00e+00 1.00e+00h  1 
  56  3.2520075e+01 3.22e-02 3.06e-02  -5.7 3.66e+04 -14.9 1.00e+00 8.91e-01h  1 S
--- Restarting execution

Thanks

[CLAassistant]

All committers have signed the CLA.

[danieloliveira56]

The CLA was failing because the commit author was different than my profile e-mail.

[svigerske]

I believe Ipopt will return code -102 (Insufficient_Memory), but GAMS will convert that to solveStat 13. Then I believe there's no easy way to know what happened (would love to hear if there's a way).

The model status in this case is actually 13 (error, no solution) and the solver status is 10 (solver error; maybe not the best choice). There is only one other case where you get this combination, which is the too-degrees-of-freedom situation.
Also there would be some message from Ipopt indicating that it ran out of memory (https://github.com/coin-or/Ipopt/blob/stable/3.14/src/Interfaces/IpIpoptApplication.cpp#L547).

The log looks as if the link just crashed, in which case you get both model and solve status 13. If you set GAMS solvelink option to 1, you may see another line from a shell saying why the process stopped. Maybe the OOM killer?