#65 Failing tests related to fminsearch

[Julien Bect]

There currently two failing tests under Matlab (tested with R2016b) when fminsearch is used as last recourse for box-constrained problems.

The first one is in stk_optim_testmin_box.m and simply indicates that the solution not not satisfy the box .constraint:

>> stk_test stk_optim_testmin_box
***** assert (stk_optim_testmin_box (algo));
!!!!! test failed
Error using stk_test>eval_test_code (line 505)
Assertion failed.

The second one is in stk_param_estim, and is probably a symptom of a much more severe problem:

>> stk_test stk_param_estim
***** test  % noiseless
 zi = stk_feval (f, xi);
 param1 = stk_param_estim (model, xi, zi, param0);
 param2 = stk_param_estim (model, xi, zi, param0, [], @stk_param_relik);
 % We cannot assume a DETERMINISTIC optimization algorithm
 % (for some reason, Octave's sqp is not exactly deterministic)
 assert (stk_isequal_tolrel (param1, param2, 1e-2))
!!!!! test failed
Error using stk_cholcov (line 66)
A contains NaNs or Infs.

Perhaps should we try to provide a better default optimizers for Matlab users that have neither the Optimization toolbox nor MOSEK.

[Julien Bect]

We already have a copy of Octave's sqp inside STK. Therefore, a solution would be to make it possible to use a different qp solvers when neither MOSEK nor the Optimization toolbox are available. For instance :

• provide a copy of Octave's qp solver as well,
• recommend to install GLPKmex (http://glpkmex.sourceforge.net).

[Julien Bect]

Another option would be to rely on NLopt++.

Anyway, unless someone steps in to work on this, this is not going to happen for STK 2.6.0...

[Julien Bect]

Moved to github: https://github.com/stk-kriging/stk/issues/7. Closing this ticket.