# CVX geometric mode with matrices

Hi,
I’m trying to solve the following algorithm:

``````cvx_begin gp
variables x(6)
minimize(x(1)+x(2))
subject to
x(1)>=0;x(2) >= 0;
0.5*x'*H*x+f*x-c <=0;
x(3)^2 - x(5) == 0;
x(6)-x(3)*x(4) == 0;
cvx_end
``````

but I get the error ‘inner matrix dimensions must agree’. However H is a 6x6 matrix.The rows in H corresponds with that in f and in c. I think it has to do with the fact that x is a log-affine expression.

I’m trying to calculate the minimum som of two variables, whereby these variables are solutions for the equation: 0.5x’Hx+fx-c <=0. However I haven’t manage to resolve this problem myself.
Does anyone know a solution for solving matrices in cvx in geometric mode?

Kind regards

I am away from my office so I can’t investigate the particular error message. However I can tell you that your model doesn’t obey the rules for disciplined geometric programs, and it is very likely not solvable by CVX. I think the equality constraints can be saved but unless H is all nonnegative the inequality cannot be. Please consult the GP section of the user guide.