It is solved, thanks for pointing out this abs
.
Dear Mark
I am in trouble with this constraint:
-log(trace(A))>=s
in which A is an optimization variable and is also n*n and s is also an affine optimizing variable. Could you help me how can i reformulate this so that cvx accepts it.
Thanks in advance.
That inequality is going the wrong direction to be convex.
log(trace(A))>=s
is a convex constraint and would be accepted by CVX.
Negating the LHS makes this non-convex, because it is convex >= affine
, which is non-convex.