Closed barakovic closed 7 years ago
Fixed order of repetitions for each kernels.
Checked that the norm of the matrix A generated with 'doNormalizeKernels' = False
doNormalizeKernels' = False
norm_A1 = np.linalg.norm(A,axis=0)`
is approximately equal to the norm calculate by COMMIT generate with 'doNormalizeKernels' = True
doNormalizeKernels' = True
norm1 = np.repeat(mit.KERNELS['wmr_norm'],nF) norm2 = np.repeat(mit.KERNELS['wmh_norm'],nE) norm3 = np.repeat(mit.KERNELS['iso_norm'],nV) norm_fib = np.kron(np.ones(mit.KERNELS['wmr'].shape[0]), mit.DICTIONARY['TRK']['norm']) norm_A2 = np.hstack( (norm1*norm_fib,norm2,norm3) )
norm_A1 ≈ norm_A2
Fixed order of repetitions for each kernels.
Checked that the norm of the matrix A generated with '
doNormalizeKernels' = False
is approximately equal to the norm calculate by COMMIT generate with '
doNormalizeKernels' = True
norm_A1 ≈ norm_A2