get principal eigvec

This commit is contained in:
rasbt 2018-05-16 01:06:58 -04:00
parent 5da40c30c0
commit a908a343af

View File

@ -0,0 +1,20 @@
# Select a principal eigenvector via NumPy
# to be used as a template (copy & paste) script
import numpy as np
# set A to be your matrix
A = np.array([[1, 2, 3],
[4, 5, 6],
[7, 8, 9]])
eig_vals, eig_vecs = np.linalg.eig(A)
idx = np.absolute(eig_vals).argsort()[::-1] # decreasing order
sorted_eig_vals = eig_vals[idx]
sorted_eig_vecs = eig_vecs[:, idx]
principal_eig_vec = sorted_eig_vecs[:, 0] # eigvec with largest eigval
normalized_pr_eig_vec = np.real(principal_eig_vec / np.sum(principal_eig_vec))
print(normalized_pr_eig_vec) # eigvec that sums up to one