'Solver not found (gurobi)'

1,586 views
Skip to first unread message

Ali Khudhair

unread,
Apr 20, 2015, 1:28:07 PM4/20/15
to yal...@googlegroups.com
Dear respected professor,
I just installed "Gurobi" solver successfully(academic version), but when I use it :
solvesdp(LMIs,gamma_inf,sdpsettings('solver','gurobi','verbose',0))

I get the following :
solvertime: 0
          info: 'Solver not found (gurobi)'
       problem: -3
    yalmiptime: 0.106

Could you please give me some hints to overcome this problem. Thank you in advance for your time.

Regards,
Ali

Johan Löfberg

unread,
Apr 20, 2015, 1:30:28 PM4/20/15
to yal...@googlegroups.com
You've installed the wrong version (linus/win etc) or you haven't added the path to the binaries in matlab. Run


>> which gurobi -all
 

and it will be clear

Ali Khudhair

unread,
Apr 20, 2015, 1:41:33 PM4/20/15
to yal...@googlegroups.com
Thank you professor for your fast response, I added the path now, I am receiving the following:
solvertime: 0
          info: 'Solver not applicable (gurobi)'
       problem: -4
    yalmiptime: 0.111

 

Johan Löfberg

unread,
Apr 20, 2015, 1:44:21 PM4/20/15
to yal...@googlegroups.com
Well, based on the name of your constraint, it looks like you are trying to solve a semidefinite program. Gurobi is not an sdp solver.

Ali Khudhair

unread,
Apr 20, 2015, 1:49:03 PM4/20/15
to yal...@googlegroups.com
That is right, Thank you professor.

Ali Khudhair

unread,
Apr 20, 2015, 1:58:14 PM4/20/15
to yal...@googlegroups.com
Dear professor,

I am trying to synthesize output-feedback gain scheduling controller (H-infinty) using ROLMIP. When I use SeDuMi, I always get "Numerical problem" :
 yalmiptime: 0.151
    solvertime: 11.726
          info: 'Numerical problems (SeDuMi-1.3)'
       problem: 4

I have tried to switch to another solver SDPT3 but the same thing happen "Lack of Progress".  Then I tried Gurobi but as you mentioned this is not good for sdp.
Is there any suggestions to get rid of this numerical problem.

Johan Löfberg

unread,
Apr 20, 2015, 2:48:19 PM4/20/15
to yal...@googlegroups.com
Bad data or bad model.
Reply all
Reply to author
Forward
0 new messages