I'm new to cvxpy, my variable have the same dimensions but I get a valueerror. here is my code:
from cvxpy import *
import numpy as np
import cPickle as pickle
import h5py
import pdb
with open('test pickle', "rb") as f:
cluster, example =pickle.load(f)
path = '/path/to/file.mat'
f = h5py.File(path)
feat_set = np.array(f['feat_set'])
feat_square = np.square(feat_set)
feat_norm = np.sqrt(np.sum(feat_square, axis = 0))
normalized_feat_set = feat_set/feat_norm
Y = normalized_feat_set[np.where(cluster.labels_==1)]
Y = np.transpose(Y)
print Y.shape
m,n = Y.shape
w = Variable(m)
objective = Minimize(sum_entries(np.square(sum_entries(Y, 1) - w))
the error i get is:
/usr/local/lib/python2.7/dist-packages/cvxpy/expressions/expression.py:286: UserWarning: Forming a nonconvex expression (affine)*(affine).
warnings.warn("Forming a nonconvex expression (affine)*(affine).")
Traceback (most recent call last):
File "cvx_test.py", line 30, in <module>
objective = Minimize(sum_entries(np.square(sum_entries(Y, 1) - w)))# + lamda_*norm(w,1))
File "/usr/local/lib/python2.7/dist-packages/cvxpy/expressions/expression.py", line 43, in cast_op
return binary_op(self, other)
File "/usr/local/lib/python2.7/dist-packages/cvxpy/expressions/expression.py", line 287, in __mul__
return cvxtypes.affine_prod_expr()(self, other)
File "/usr/local/lib/python2.7/dist-packages/cvxpy/atoms/affine_prod.py", line 30, in __init__
super(affine_prod, self).__init__(x, y)
File "/usr/local/lib/python2.7/dist-packages/cvxpy/atoms/atom.py", line 47, in __init__
self.validate_arguments()
File "/usr/local/lib/python2.7/dist-packages/cvxpy/atoms/affine_prod.py", line 73, in validate_arguments
u.shape.mul_shapes(self.args[0].size, self.args[1].size)
File "/usr/local/lib/python2.7/dist-packages/cvxpy/utilities/shape.py", line 55, in mul_shapes
lh_shape, rh_shape))
ValueError: Incompatible dimensions (512, 1) (512, 1)
I'm not sure what is wrong as both my variable in the minimize function have the same dimensions. can it be that because i'm using np.square it creates an error?
