The following toy function ordinarily takes two input variables:
f = lambda u1, u2 : (u1*u2)*(u1**2+u2**2)
but can scale beyond the bivariate