ValueError:设置具有序列的数组元素.最小化 [英] ValueError: setting an array element with a sequence. scipy minimize

问题描述

""" ___ """
from scipy.optimize import minimize
import numpy as np


LENGTH = 100

def process(x):
    return x * 2 + 5

def draw(process, length):
    """ """
    y = np.random.normal(0, 10, length)
    data = [process(y_) for y_ in y]
    rnd = np.random.normal(3, 1, len(data))
    return y, rnd + data


def maximum_likelyhood(y, X):
    objective = lambda b: np.transpose(X) * (y - X * b)
    x0 = np.zeros(100)
    res =  minimize(objective, x0=x0)
    return res.x

y, X = draw(process, LENGTH)
print maximum_likelyhood(y, X)

产生一个

ValueError: setting an array element with a sequence.

有几个类似的问题，它们都指出x0不是一维数组，但这里是一维数组. (或者不是?以防万一，请解释为什么以及如何制作一维)

There is several similar problems, they all point out that x0 is not a 1D array, but here it is a 1D array. (or not? in case please explain why and how to make it 1D)

推荐答案

发生错误是因为目标函数是矢量函数(采用矢量，返回矢量)，但根据

The error occurs because the objective function is a vector function (takes in a vector, returns a vector) but according to scipy.optimize.minimize documentation it only takes scalar function (takes a vector returns a scalar.)

这篇关于ValueError:设置具有序列的数组元素.最小化的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！