# No-product rule

I have a fraction constraints:

 cvx_begin
variable X(3,3) hermitian;
variable t(2);

q1 = trace(X*R1);
q2 = trace(X*R2);

j1 = trace(real(X*M1))+1;
j2 = trace(real(X*M2))+1;

maximize  geo_mean(t,1,W)
subject to
%% this what i want to do: q1/j1 >= t(1) &   q2/j2 >= t(2)
q1 >= t(1)*j1
q2 >= t(2)*j2


It seems that the two last two lines violate the no-product rule at the right-hand side

Where R1,R2,M1,M2, and W are constants.

How can i write this constraint in CVX?

The Error:

Error using cvx/times (line 262)
Disciplined convex programming error:
Invalid quadratic form(s): not a square.

Error in cvx/mtimes (line 36)
z = feval( oper, x, y );

Error in max_min_assump1_plus_WSRM_02 (line 319)
q1 >= t(1)*j1


Edited Part:

The original model was:

$$maximize \sum W_i log (q_i/j_i) \quad , i =1,2$$
$$s.t \quad some \quad constraints$$

which could be transformed to:

$$maximize \prod (q_i/j_i)^{W_i} \quad , i =1,2$$
$$s.t \quad some \quad constraints$$

Then it can be transformed according to one of Boyd papers (Qos and Fairness Constrained convex optimization of resource allocation) to:

$$maximize \prod (t_i)^{W_i} \quad , i =1,2$$
$$s.t \quad some \quad constraints$$
$$\quad q_i/j_i \ge t_i$$

my problem now is with this constraint

Is your model convex? It does not appear so.

i added some details on the original problem in the edited section

I am afraid I cannot spend time reading a paper to see how the problem is transformed to convex form. All I can tell you is that, as written, it is not convex, and not compatible with CVX.