i know how to solve the optimization of the variant of this equation with constant values instead of probabilistic of x1, x2 and x3 using Lagrange multipliers:
H = Z(y1,y2) - λ ((2+6x1)/y1 + 6x2/y2 -18)
but when x1, x2, and x3 are probabilistic values (in this case uniform distributions) i dont know how to tackle this problem.