如何修复错误：func返回的数组必须为一维，但得到ndim=2

Hey there, let's break down why you're hitting this error and how to fix it quickly.

The Root Cause #

The main issue here is the order of arguments in your g function doesn't match what odeint expects.

scipy.integrate.odeint requires the callback function to follow this signature:

def func(y, t, *args):
    # compute dy/dt here
    return dy_dt

But your function is defined as def g(t,y,B): — you swapped the t and y parameters!

When odeint calls your function, it passes the current state y first, then the time t. Since you have them reversed, your code is treating the scalar t as the state vector y, and then doing np.dot(t, B). A scalar multiplied by a 5x5 matrix gives another 5x5 matrix (2-dimensional), which odeint can't handle — hence the error about returning a 2D array instead of 1D.

The Fix #

Just swap the order of t and y in your g function definition. Also, let's double-check that the return value is indeed a 1D array (which it will be once the parameters are correct):

import numpy as np
from scipy.integrate import odeint

y0 = np.array([1,3,2,3,5])
b = np.array([[1],[3],[4],[2],[5]])
B = np.dot(b, b.T)

# Fixed parameter order: y comes first, then t
def g(y, t, B):
    # y is a 1D array, B is 5x5 — np.dot(y, B) gives a 1D array of shape (5,)
    return np.exp(np.dot(y, B))

t = np.linspace(0, 1, 100)
y = odeint(g, y0, t, args=(B,))

Why This Works #

Now when odeint calls g, it passes the 1D state vector y first. np.dot(y, B) computes a 1D array (shape (5,)), and np.exp preserves that 1D shape — exactly what odeint expects for the derivative of your state vector.

内容的提问来源于stack exchange，提问作者Lucas