def test_sumd():
data = scipy.io.loadmat('test_sumd.mat',struct_as_record=True)['X']
codebook = np.array([[3.6211, 3.1914],
[-3.6694, -3.3425]])
actual_code, dist = vq( data, codebook)
actual_dist, count = postcluster.get_summed_inner_distance( codebook, data )
# The hard-coded expected values -- pre-computed using MATLAB kmeans().
expected_code = np.array([1, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1])
expected_dist = np.array([21.9997,18.1554])
assert np.allclose( actual_code, expected_code )
assert np.allclose( actual_dist, expected_dist )
d.append( (a, b, zi) )
random.shuffle(d)
d = np.array(d)
codebook0 = np.array( [[3,3,0],
[-3,-3,0]], dtype = np.float )
codebook1 = np.array( [[3.1,3,0],
[-3,-3,0]], dtype = np.float )
codebook2 = np.array( [[3.1,4.0,0],
[-3,-3,0]], dtype = np.float )
codebook3 = np.array( [[3.1,5.0,0],
[-3,-3,0]], dtype = np.float )
IDX0,_ = vq( d, codebook0 )
IDX1 = IDX0+1 # one-based indexing
SUMD,_ = postcluster.get_summed_inner_distance( codebook0, d )
scipy.io.savemat('generated_data',
{'X':d,
'codebook0':codebook0,
'codebook1':codebook1,
'codebook2':codebook2,
'codebook3':codebook3,
'IDX0':IDX0,
'IDX1':IDX1,
'SUMD':SUMD,
}, oned_as='column')
