使用scipy.optimize最小化二元函数时遇标量返回值错误求助

问题重现 #

你编写了用于最小化卡方值的Python代码，但运行时抛出错误：

ValueError: The user-provided objective function must return a scalar value.

完整代码及依赖 #

import numpy as np
import matplotlib.pyplot as plt
import math
from scipy.integrate import quad
import scipy.optimize as optimize

# 已定义变量
f_0 = 0.261
G_F = 1.166 * 10**(-5)
m_e = 0.511 * 10**(-3)
dB_dqsq_arr = np.array([7.2,7.14,6.7,7.56,6.44,7.17,6.67,6.33,6.2,4.32,4.25,3.4,1.17])
dBdqsq_err_arr = np.array([0.70,0.45,0.39,0.43,0.43,0.45,0.47,0.48,0.44,0.43,0.41,0.42,0.26])
q_sq_arr = np.array([1,3,5,7,9,11,13,15,17,19,21,23,25])

def f_multi_para(q_sq, alpha_par):
    return f_0 / (q_sq * alpha_par)

def dB_dqsq_model2_para(q_sq, V_ub_par, alpha_par):
    sec1 = G_F**2 * V_ub_par
    sec2 = (1 - m_e**2 / q_sq)**2
    sec3 = (q_sq * (1 + m_e**2 / q_sq) * f_multi_para(q_sq, alpha_par)**2)
    return sec1 * sec2 * sec3

def chi_sq(params):
    V_ub_par, alpha_par = params
    return np.sum(((dB_dqsq_arr - np.array([dB_dqsq_model2_para(v, V_ub_par, alpha_par) for v in q_sq_arr])) / dBdqsq_err_arr)**2)

initial_guess = [0.0037, 0.54]
result = optimize.minimize(chi_sq, initial_guess)
if result.success:
    fitted_params = result.x
    print(fitted_params)
else:
    raise ValueError(result.message)

错误原因 #

这个错误的核心是scipy.optimize.minimize要求目标函数（即chi_sq）必须返回单个数值（标量），你的代码出现问题的可能原因：

• 循环遍历计算模型值时，隐性引入了数组维度不匹配问题（虽然代码中长度一致，但循环写法容易出错）；
• 优化过程中alpha_par可能趋近于0，导致f_multi_para计算出现无穷大，部分旧版本scipy会将其判定为无效返回值；
• 循环计算效率低，容易在数组转换时产生隐性错误。

修复方法 #

方法1：改用数组广播计算（推荐） #

将模型函数改为直接支持数组输入，避免循环，既高效又能彻底消除维度问题：

def f_multi_para(q_sq, alpha_par):
    return f_0 / (q_sq * alpha_par)  # 直接处理q_sq数组

def dB_dqsq_model2_para(q_sq, V_ub_par, alpha_par):
    sec1 = G_F**2 * V_ub_par
    sec2 = (1 - m_e**2 / q_sq)**2
    sec3 = (q_sq * (1 + m_e**2 / q_sq) * f_multi_para(q_sq, alpha_par)**2)
    return sec1 * sec2 * sec3

def chi_sq(params):
    V_ub_par, alpha_par = params
    # 直接传入整个q_sq_arr数组计算模型值
    model_vals = dB_dqsq_model2_para(q_sq_arr, V_ub_par, alpha_par)
    # 数组自动广播运算，无需循环
    return np.sum(((dB_dqsq_arr - model_vals) / dBdqsq_err_arr)**2)

方法2：添加参数边界约束 #

为避免alpha_par趋近于0导致的除以0问题，给优化参数设置合理范围：

initial_guess = [0.0037, 0.54]
# 根据实际物理意义调整参数范围
bounds = [(0.001, 0.01), (0.01, 1.0)]
result = optimize.minimize(chi_sq, initial_guess, bounds=bounds)

方法3：验证目标函数返回值 #

在chi_sq函数中添加打印语句，确认返回值类型和数值：

def chi_sq(params):
    V_ub_par, alpha_par = params
    model_vals = np.array([dB_dqsq_model2_para(v, V_ub_par, alpha_par) for v in q_sq_arr])
    chi_sq_val = np.sum(((dB_dqsq_arr - model_vals) / dBdqsq_err_arr)**2)
    print(f"返回值类型：{type(chi_sq_val)}，值：{chi_sq_val}")
    return chi_sq_val

如果返回值是标量但仍报错，建议更新scipy到最新版本，旧版本可能存在兼容性问题。

修复后完整代码 #

import numpy as np
import scipy.optimize as optimize

# 已定义变量
f_0 = 0.261
G_F = 1.166 * 10**(-5)
m_e = 0.511 * 10**(-3)
dB_dqsq_arr = np.array([7.2,7.14,6.7,7.56,6.44,7.17,6.67,6.33,6.2,4.32,4.25,3.4,1.17])
dBdqsq_err_arr = np.array([0.70,0.45,0.39,0.43,0.43,0.45,0.47,0.48,0.44,0.43,0.41,0.42,0.26])
q_sq_arr = np.array([1,3,5,7,9,11,13,15,17,19,21,23,25])

def f_multi_para(q_sq, alpha_par):
    return f_0 / (q_sq * alpha_par)

def dB_dqsq_model2_para(q_sq, V_ub_par, alpha_par):
    sec1 = G_F**2 * V_ub_par
    sec2 = (1 - m_e**2 / q_sq)**2
    sec3 = (q_sq * (1 + m_e**2 / q_sq) * f_multi_para(q_sq, alpha_par)**2)
    return sec1 * sec2 * sec3

def chi_sq(params):
    V_ub_par, alpha_par = params
    model_vals = dB_dqsq_model2_para(q_sq_arr, V_ub_par, alpha_par)
    return np.sum(((dB_dqsq_arr - model_vals) / dBdqsq_err_arr)**2)

initial_guess = [0.0037, 0.54]
bounds = [(0.001, 0.01), (0.01, 1.0)]
result = optimize.minimize(chi_sq, initial_guess, bounds=bounds)

if result.success:
    fitted_params = result.x
    print(f"拟合参数：V_ub_par={fitted_params[0]:.6f}, alpha_par={fitted_params[1]:.6f}")
else:
    raise ValueError(result.message)

内容的提问来源于stack exchange，提问作者noob anomaly