Hi,
I am experiencing a 100% failure rate ("generic failure code") with AUGLAG
when using LBFGS as subsidiary algorithm. The same optimization problem is
solved successfully by AUGLAG+SLSQP and LBFGS (without AUGLAG). ** This
makes me thing that either I am doing something wrong or there is bug in
libopt.
** It looks like it is not necessary to enforce the inequality constraints
for this problem, although I would prefer to use AUGLAG in order to make
sure the constraints are not violated. Also, LBFGS appears to be the most
efficient algorithm.
Here you can find a plot summarizing my the tests:
http://s28.postimg.org/olqc7oi2l/libopt_fail.png
This is the nloptr call:
nloptr(x0, f, gr,
eval_g_ineq=hin,
eval_jac_g_ineq=hinjac,
lb=near0, ub=5,
opts=list(
algorithm='NLOPT_LD_AUGLAG',
xtol_rel=1e-10,
xtol_abs=near0,
maxeval=-1,
maxtime=120,
local_opts=list(
algorithm='NLOPT_LD_LBFGS',
xtol_rel=near0)),
## extra arguments to f and gr
...)
Have you got any idea of what might be wrong?
Any help appreciated.
Regards,
Ernest
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
