[Ipopt-tickets] [Ipopt] #305: windows Ipopt executables most often fail for all versions , while NEOS ipopt always succeed
Ipopt
coin-trac at coin-or.org
Mon May 28 11:33:30 EDT 2018
#305: windows Ipopt executables most often fail for all versions , while NEOS
ipopt always succeed
---------------------+------------------------
Reporter: maajdl | Owner: ipopt-team
Type: defect | Status: new
Priority: normal | Component: Ipopt
Version: 3.12 | Severity: major
Resolution: | Keywords:
---------------------+------------------------
Comment (by stefan):
David Gay from AMPL had a look and changed something in ASL to get the
provided .nl files working:
{{{
I think a failure to initialize a couple of fields was due to a
slightly bogus .nl file. Pyomo did not do some folding that AMPL
does, which may have be involved in this issue. Anyway, initializing
the fields in question (even though they should be initialized
subsequently) fixes this problem.
I had missed testing for f_OPNUM at one point in sphes.c, which
caused the trouble in sphes_setup. This might in some cases
have led to overestimating the sparsity, but did not affect
actual numerical computations.
I've updated http://ampl.com/netlib/ampl/solvers.tgz and the
affected files in http://ampl.com/netlib/ampl/solvers to address
these issues. No changes where needed to the (now preferred)
http://ampl.com/netlib/ampl/solvers2.tgz .
}}}
Knitro should also use ASL, so you see the same issue there.
I'll see that we'll get an Ipopt release out that uses a newer ASL.
In the meanwhile, you can just unpack
http://ampl.com/netlib/ampl/solvers.tgz in !ThirdParty/ASL instead of
running {{{get.ASL}}}.
--
Ticket URL: <https://projects.coin-or.org/Ipopt/ticket/305#comment:14>
Ipopt <http://projects.coin-or.org/Ipopt>
Interior-point optimizer for nonlinear programs.
More information about the Ipopt-tickets
mailing list