ValueError：操作数无法与形状一起广播 (32,31) (32,29)

如何解决ValueError：操作数无法与形状一起广播 (32,31) (32,29)

import numpy as np

# Todo: divide error by respective number of instances for normalization
def costfunction(params,Y,R,num_students,num_courses,num_features,reg_param,reg_param2,OrigTheta):

# X is courses*features
# Theta is students*parameters
X = np.reshape(params[:num_courses * num_features],(num_courses,num_features),order='F')
Theta = np.reshape(params[num_courses * num_features:],(num_students,order='F')
# Take dot product of theta and x transpose to compute predicted rating
# Compute squared error
squared_error = np.power(np.dot(Theta,X.T) - Y,2)
# Contribution to squared error will come only from those ratings which
# are not missing and which have not been relocated to test data
J = (1 / 2.) * np.sum(squared_error * R)

#Add contribution of theta and x to objective funciton incorporating the regularization parameter
J = J + (reg_param / 2.) * (np.sum(np.power(Theta,2)) + np.sum(np.power(X,2)))

# Limit the value of new theta close to original theta
J = J + (reg_param2 / 2.) * (np.sum(np.power(Theta - OrigTheta,2)))

X_grad = np.dot(Theta.T,(np.dot(Theta,X.T) - Y) * R).T
Theta_grad = np.dot(((np.dot(Theta,X.T) - Y) * R),X)

X_grad = X_grad + reg_param * X
Theta_grad = Theta_grad + reg_param * Theta + reg_param2 * (Theta - OrigTheta)

grad = np.concatenate((X_grad.reshape(X_grad.size,order='F'),Theta_grad.reshape(Theta_grad.size,order='F')))

return J,grad

错误：

File "C:\Users\ranjan\Desktop\course-recommendation-master\course-recommendation- 
master\rateapp\Costfunction.py",line 12,in costfunction squared_error = np.power(np.dot(Theta,2) 
ValueError: operands Could not be broadcast together with shapes (32,31) (32,29)