Hi, I am trying to put this constraint in CVX: (a ln(2)/b x) <= ln(1+(y_i c/b d)) + (y_i c/(y c + b d)) - ((y_i c)^2/(y_i c + b d)) (1/y c). Here, a, b, c, d are constants. x and y are variables. I am using successive convex approximation (SCA) for this constraint. So I have initialized values for y_i and hoped to get optimal values y for each iteration and put y_i = y to update.

I am getting error:

The following constraints are not DCP:

69000.0 @ power(5000000.0 @ var89, -1.0) <= log(400001.0) + 20.0 @ (1.0 / (var88 + 5e-05)) + -4e-06 @ power(0.002000005, -1.0) @ 1.0 @ power(var88 @ 0.0001, -1.0) , because the following subexpressions are not:

|-- 1.0 / (var88 + 5e-05)

How can I write the constraint so that CVX accepts it? Also, am I using SCA method correctly?

I would really appreciate your feedback.