NaN even with example codes !

[Anurag Prabhakar]

I'm fairly new to the area of convex optimization so I might as well be doing something very silly, but even example codes likes "sudoku.m" are not giving desired results, the whole matrix is NaN. Any help would be appreciated.

[Johan Löfberg]

Since you haven't told us anything about what you saw (which solver you think you are using, solver log, diagnostic codes etc), I can just as well assume that the problem is that your cat walked over the keyboard and accidentally pressed ctrl-c and terminated the solver before completion. Solution: Remove cat from room during computations.

:-).

[Johan Löfberg]

BTW, The Sudoku optimization problem is very far from convex. It is a mixed-integer linear program.

[Anurag Prabhakar]

Damn ! I was expecting something like that :D So all that I did was download and extract the YALMIP zip and add it MatLab's path and tried to run the "sdptutorial.m" and "sudoku.m" codes to see if everything is working well. While I got all 0's and NaN's for the former, the output matrix for the sudoku was all NaN's. So there's some problem, right ?

[Johan Löfberg]

You still haven't told us what happens when you solve the optimization problem, i.e., run solvesdp (remove the ; after solvesdp(F) to show us what is returned from the command also, in case the solver doesn't run or display anything). With some diagnostics, we'll see.

I can only assume that you don't have any suitable solvers installed, or your solver installation is broken.
http://users.isy.liu.se/johanl/yalmip/pmwiki.php?n=Solvers.Solvers

OPTI Toolbox is a good start, since it installs a bunch of solvers for you. If you are in academia, cplex, gurobi and mosek all have full academic licenses