[Ipopt] Solved To An Acceptable Level with less than acceptable_iter iterations completed

[Andreas Waechter]

Hi Neil,

> I have an optimisation which solves to within acceptable tolerances
> (solved with message EXIT: Solved To Acceptable Level.) in 9
> iterations
> But the acceptable_iter option is set to 15, so how is it possible for
> the optimisation to come to an acceptable solution in 9 iterations?

You are correct, usually this exit message should not occur.  However,
Ipopt will also terminate with this exit message if it would otherwise
start the restoration phase at a point that is already deemed
acceptable.

If you would like to go to the bottom of this, I suggest to set the 
print_level to 7 (maybe only 6 for the older Ipopt version you use) and 
see if that tells you what happens in the last iteration.  If you want, 
you can also submit a ticket at the Ipopt trac page and attach the output 
file.

Also, version 3.2.1 is somewhat old, and some bugs might have been
fixed in the meantime.  Maybe this goes away if you switch to the
latest version.

Regards,

Andreas

>
> I'm using ipopt version 3.2.1 using the pardiso solver.
>
>
> Number of nonzeros in equality constraint Jacobian...:        0
> Number of nonzeros in inequality constraint Jacobian.:      778
> Number of nonzeros in Lagrangian Hessian.............:        0
>
> Total number of variables............................:      133
>                      variables with only lower bounds:        0
>                 variables with lower and upper bounds:      133
>                      variables with only upper bounds:        0
> Total number of equality constraints.................:        0
> Total number of inequality constraints...............:        6
>         inequality constraints with only lower bounds:        0
>    inequality constraints with lower and upper bounds:        0
>         inequality constraints with only upper bounds:        6
>
> iter    objective    inf_pr   inf_du lg(mu)  ||d||  lg(rg) alpha_du alpha_pr  ls
>    0 1.3447041e+004 0.00e+000 7.07e+000   0.0 0.00e+000    -  0.00e+000 0.00e+000   0
>    1 1.0066625e+004 1.38e+002 7.77e+000   1.1 1.41e+001    -  9.90e-001 9.20e-002f  2
>    2 7.4745524e+003 1.93e+001 3.13e+000   0.5 4.22e+002    -  9.67e-001 1.00e+000h  1
>    3 5.9017625e+003 1.90e+001 1.80e+000   0.3 5.16e+002    -  1.00e+000 9.68e-001h  1
>    4 5.8328967e+003 8.87e-001 3.08e+000  -1.8 2.42e+002    -  9.86e-001 1.00e+000h  1
>    5 5.7464832e+003 3.16e+002 2.68e+000  -1.4 1.88e+004    -  1.00e+000 6.50e-001F  1
>    6 5.6115824e+003 5.77e+000 4.92e-001  -2.0 1.02e+002    -  1.00e+000 1.00e+000h  1
>    7 5.6001170e+003 3.79e-002 4.51e-003  -3.7 1.25e+001    -  1.00e+000 1.00e+000h  1
>    8 5.6000010e+003 3.84e-004 3.99e-004  -5.8 1.78e+000    -  1.00e+000 1.00e+000h  1
>    9 5.6000001e+003 6.24e-005 3.64e-006  -7.0 4.42e-001    -  1.00e+000 1.00e+000h  1
>
> Number of Iterations....: 9
>
>                                    (scaled)                 (unscaled)
> Objective...............:  2.7773800591066699e+001 
5.6000000681350984e+003
> Dual infeasibility......:  3.6350855810964256e-006 
7.3293820322035638e-004
> Constraint violation....:  0.0000000000000000e+000 
0.0000000000000000e+000
> Complementarity.........:  9.1302384237744727e-008 
1.8409196691529583e-005
> Overall NLP error.......:  3.6350855810964256e-006 
7.3293820322035638e-004
>
>
> Number of objective function evaluations             = 38
> Number of objective gradient evaluations             = 11
> Number of equality constraint evaluations            = 0
> Number of inequality constraint evaluations          = 38
> Number of equality constraint Jacobian evaluations   = 0
> Number of inequality constraint Jacobian evaluations = 11
> Number of Lagrangian Hessian evaluations             = 0
> Total CPU secs in IPOPT (w/o function evaluations)   =      0.093
> Total CPU secs in NLP function evaluations           =      0.000
>
> EXIT: Solved To Acceptable Level.
>
>
> Thanks Neil._______________________________________________