Scipy.optimize.minimize 目标 function 必须返回一个标量

Question

I try to optimize my function:

def get_ret(weights):
   weights = np.array(weights)
   ret = np.sum(log_ret.mean() * weights)*252
   vol = np.sqrt(np.dot(weights.T,np.dot(log_ret.cov()*252,weights)))
   sr = ret / vol
   return [ret,vol,sr]

def neg_sharp(weights):
   return get_ret(weights[2]) * -1

and my constraints is:

def check_sum(weights):
   return np.sum(weights) -1


cons = ({"type":"eq", "fun":check_sum})

my bonds is: bounds = ((0,1),(0,1),(0,1),(0,1))

and:

init_guess = np.array([.25,.25,.25,.25])

so i run this:

opt_res = minimize(fun=neg_sharp,x0=init_guess.flatten(),
         method="SLSQP",bounds=bounds,constraints=cons)

and got this error:

ValueError: Objective function must return a scalar

Answer 1

It seems like the problem is here return get_ret(weights)[2] * -1
get_ret returns a list, I guess you want to take one element out of this list (the 3rd one?) and multiply by -1.

Try:

def get_ret(weights):
   weights = np.array(weights)
   ret = np.sum(log_ret.mean() * weights)*252
   vol = np.sqrt(np.dot(weights.T,np.dot(log_ret.cov()*252,weights)))
   sr = ret / vol
   return [ret,vol,sr]

def neg_sharp(weights):
   return get_ret(weights)[2] * -1

Answer 2

ok. I'm sorry. this is my fault.

return get_ret(weights[2]) * -1

must change to:

return get_ret(weights)[2] * -1