Can I custormize some tolarence rule?

[Penny Xie]

Hi,

I use cvxopt to solve a quadratic problem and it turns out there is no feasible solution. I run the same problem in MatLab, and I find the iterations will stop according to the following rule:

Optimization stopped because the total relative error has grown to more than 1e5 times its minimum value during the iterations.

Since I want my result in python the same as my result in MATLAB, is there any way I can apply this rule when I use cvxopt in python?

Thanks!

[Martin]

Are you using quadprog() in MATLAB? The warning/error message does not say anything about infeasibility. 

CVXOPT will return a certificate of infeasibility if your problem is infeasible: 

http://cvxopt.org/userguide/coneprog.html#quadratic-programming