I have an optimization problem as described below. My problem in the second constraint (y’*b>=1) which is a non-convex. My question is how to reformulate this constrain to become convex.

```
clc
a=randi(5,2,1);
cvx_begin
variable x(2,1) nonnegative
y=exp(x);
c1=y(2)*a(2) + 1
maximize( log(y(1)*a(1)/c1) - log(1+y(2)*a(2)/2) )
subject to
sum(x) <=5
y'*b>=1
cvx_end
```