After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 679027 - non-linear solver does not respect constraints
non-linear solver does not respect constraints
Status: RESOLVED DUPLICATE of bug 620368
Product: Gnumeric
Classification: Applications
Component: General
git master
Other All
: Normal normal
: ---
Assigned To: Jody Goldberg
Jody Goldberg
Depends on:
Blocks:
 
 
Reported: 2012-06-28 08:49 UTC by Frédéric Parrenin
Modified: 2012-07-03 14:20 UTC
See Also:
GNOME target: ---
GNOME version: ---


Description Frédéric Parrenin 2012-06-28 08:49:59 UTC 
Steps to reproduce the problem:
- open the attached file
- open the solver
- minmize cell age!H1 by modifying cells thinning!H1:H5
- use a non-linear model
- assume non-negative values
- run the solver
=> the solution found gives a negative value in cell thinning!H4.
Comment 1 Andreas J. Guelzow 2012-06-28 15:20:26 UTC 
This is part of item #5 of bug #620368

*** This bug has been marked as a duplicate of bug 620368 ***
Comment 2 Morten Welinder 2012-06-29 01:25:14 UTC 
No attachment -- I'd be interested in such a sample.

You can work around the problem by imposing large (but not infinite
penalties) on going beyond the constraint.  Something like max(-1e6*x42,0)
for a variable not allowed to be negative.  That's not particular
elegant. but it ought to work.  In  fact, it might work better than respecting
the constraints (whenever I get around to take) because there might not
be a whole lot to do beyond throwing up a "no convergence" message.
Comment 3 Frédéric Parrenin 2012-07-03 09:18:08 UTC 
Sorry, the file is here:
http://parrenin.frederic.free.fr/download/bugzilla/dating-model.gnumeric
Comment 4 Morten Welinder 2012-07-03 13:28:26 UTC 
Something has changed since the bug report was filed.  age!H1 no longer
contains an expression and thinning!H1:H5 no longer the variables.
Comment 5 Frédéric Parrenin 2012-07-03 14:20:01 UTC 
Sorry, this is now age!L1 for the variable to minimize and age!3036:3040 for the variable to tune.