Fitting wrt to a predefined distribution

Hi,
Is it possible to define the fitness function as a Wasserstein distance between all transformed samples and some predefined distribution?

In the optimization process I only care about the resulting distribution of the y=f(x)

Thank you!