For any parameter that can be defined through the combination of a variable and an operand or by a solve, it is generally better to use the solve wherever possible. In the case of constraining the effective focal length, you may choose to use a marginal ray angle solve on the radius of the last surface to calculate it from the entrance pupil diameter and the target focal length. The time that it takes for the optimization routine to run is proportional to the square of the number of operands in the merit function, so using solves to reduce the number of variables is best practice to prevent the optimizer from getting bogged down. Plus, in these cases where you can compute the value you want directly through a solve, using optimization just isn’t necessary.