[Python] How can I solve "cannot pickle local object" when trying to use parallelization?

The method below belongs to a class and it is imported by a script that indirectly calls it. It is used to solve N constraints independent LSQ problems. The parallelization part is not function, and I am finding the following error message:

Can't pickle local object 'aapy.U_matrix.<locals>.lsqSOLUTION'.
​

How can I solve it?

def U_matrix(self, V):
U = np.zeros((0, self.c))
iters, eabs, min_val = 500, 0.01, 1e-10


def lsqSOLUTION(i):
ui = cp.Variable((self.c))
objective = cp.Minimize(cp.sum_squares(V @ ui - self.Data[:, i]))
constraints = [0 <= ui, ui <= 1, cp.sum(ui)==1]
prob = cp.Problem(objective, constraints)
prob.solve(solver=cp.ECOS, max_iters=iters, abstol=eabs)
u = np.transpose(ui.value)
return u

args = range(0, self.N)
with multiprocessing.Pool() as pool:
u = pool.map(lsqSOLUTION, args)
vector = np.array(u)
U = np.vstack(U, vector)
return U


I have the same code using basic for loop and it functions.

Continue reading...