def is_spd(A):
"""Returns True if A is symmetric positive definite.
"""
assert A.ndim == 2, "Expecting 2D matrix."
if not is_symmetric(A):
return False
eigvals, _ = eig(A, sort=False)
return np.all(eigvals >= 0)
def projected_iteration(A, k, max_iter=1000, sort=True):
"""Sequentially find the k eigenpairs of a symmetric matrix.
Concretely, combines power iteration and deflation to find
eigenpairs in order of decreasing magnitude.
Args:
A: a square symmetric array of shape (N, N).
k (int): the number of eigenpairs to return.
sort (bool): Whether to sort by decreasing eigenvalue magnitude.
Returns:
e, v: eigenvalues and eigenvectors. The eigenvectors are
stacked column-wise.
"""
assert utils.is_symmetric(A), "[!] Matrix must be symmetric."
assert k > 0 and k <= A.shape[0], "[!] k must be between 1 and {}.".format(
A.shape[0])
eigvecs = np.zeros((A.shape[0], k))
eigvals = np.zeros(A.shape[0])
for i in range(k):
v = np.random.randn(A.shape[0])
for _ in range(max_iter):
# project out computed eigenvectors
proj_sum = np.zeros_like(v)
for j in range(i):
proj_sum += utils.projection(v, eigvecs[:, j])
v -= proj_sum
v_new = A @ v
v_new = utils.normalize(v_new)
if np.all(np.abs(v_new - v) < 1e-8):
break
v = v_new
e = single.rayleigh_quotient(A, v)
# store eigenpair
eigvecs[:, i] = np.array(v)
eigvals[i] = e
# sort by largest absolute eigenvalue
if sort:
idx = np.abs(eigvals).argsort()[::-1]
eigvecs = eigvecs[:, idx]
eigvals = eigvals[idx]
return eigvals, eigvecs
def hessenberg(A, calc_q=False):
"""Reduce a square matrix to upper Hessenberg form using Householder reflections.
If the input matrix is symmetric, the resulting Hessenberg form is
reduced to tridiagonal form.
Args:
A: A square matrix of shape (M, M).
calc_q (bool): Whether to explicitly compute the product of
similarity transform Householder matrices.
Returns:
A: The upper Hessenberg form of the matrix of shape (M, M).
Q: The product of Householder matrices of shape (M, M). Only
returned if `calc_q=True`.
"""
assert utils.is_square(A), "[!] Matrix must be square."
is_symm = utils.is_symmetric(A)
A = np.array(A)
M, _ = A.shape
vs = []
for i in range(M - 2):
a = A[i + 1:, i]
c = utils.l2_norm(a)
s = utils.sign(a[0])
e = utils.basis_vec(0, len(a), flat=True)
v = a + s * c * e
vs.append(v)
# left transform
for j in range(i, M):
A[i + 1:,
j] = A[i + 1:, j] - (2 * v.T @ A[i + 1:, j]) / (v.T @ v) * v
# right transform
for j in range(i if is_symm else 0, M):
A[j, i + 1:M] = A[j, i + 1:M] - 2 * ((A[j, i + 1:M].T @ v) /
(v.T @ v)) * v.T
if calc_q:
Q = np.eye(M)
for i in range(M):
for j, v in enumerate(reversed(vs)):
Q[M - j - 2:,
i] -= (2 * v.T @ Q[M - j - 2:, i]) / (v.T @ v) * v
return A, Q
return A
def power_iteration(A, max_iter=1000):
"""Finds the largest eigenpair of a symmetric matrix.
Args:
A: a square symmetric array of shape (N, N).
Returns:
e, v: eigenvalue and right eigenvector.
"""
assert utils.is_symmetric(A), "[!] Matrix must be symmetric."
v = np.random.randn(A.shape[0])
for i in range(max_iter):
v_new = A @ v
v_new = utils.normalize(v_new)
if np.all(np.abs(v_new - v) < 1e-8):
break
v = v_new
e = rayleigh_quotient(A, v)
return e, v
def rayleigh_quotient_iteration(A, mu, max_iter=1000):
"""Finds an eigenpair closest to an initial eigenvalue guess.
Args:
A: a square symmetric array of shape (N, N).
mu: an initial eigenvalue guess.
Returns:
e, v: eigenvalue and right eigenvector.
"""
assert utils.is_symmetric(A), "[!] Matrix must be symmetric."
v = np.random.randn(A.shape[0])
for i in range(max_iter):
v_new = solve(A - mu * np.eye(A.shape[0]), v)
v_new = utils.normalize(v_new)
if np.all(np.abs(v_new - v) < 1e-8):
break
v = v_new
mu = rayleigh_quotient(A, v)
return mu, v
def inverse_iteration(A, max_iter=1000):
"""Finds the smallest eigenpair of a symmetric matrix.
Args:
A: a square symmetric array of shape (N, N).
Returns:
e, v: eigenvalue and right eigenvector.
"""
assert utils.is_symmetric(A), "[!] Matrix must be symmetric."
v = np.random.randn(A.shape[0])
PLU = LU(A, pivoting='partial').decompose()
for i in range(max_iter):
v_new = solve(PLU, v)
v_new = utils.normalize(v_new)
if np.all(np.abs(v_new - v) < 1e-8):
break
v = v_new
e = rayleigh_quotient(A, v)
return e, v
def qr_algorithm(A, hess=True, sort=True):
"""The de-facto algorithm for finding all eigenpairs of a symmetric matrix.
Args:
A: a square symmetric array of shape (N, N).
hess (bool): Whether to compute the Hessenberg form
of the matrix before starting the QR iterations.
sort (bool): Whether to sort by decreasing eigenvalue magnitude.
Returns:
e, v: eigenvalues and eigenvectors. The eigenvectors are
stacked column-wise.
"""
assert utils.is_symmetric(A), "[!] Matrix must be symmetric."
backup = np.array(A)
if hess:
A = hessenberg(A, calc_q=False)
M = A.shape[0]
for k in range(1000):
mu = A[M - 1, M - 1] # set the shift to be the last diagonal element
Q, R = QR(A - mu * np.eye(M)).decompose()
A_new = R @ Q + mu * np.eye(M)
if np.all(np.abs(A_new - A) < 1e-8):
break
A = A_new
mus = utils.diag(A)
eigvecs = np.zeros_like(A)
eigvals = np.zeros(A.shape[0])
for i, mu in enumerate(mus):
eigval, eigvec = single.rayleigh_quotient_iteration(backup,
mu,
max_iter=1)
eigvecs[:, i] = eigvec
eigvals[i] = eigval
if sort:
idx = np.abs(eigvals).argsort()[::-1]
eigvecs = eigvecs[:, idx]
eigvals = eigvals[idx]
return eigvals, eigvecs
def test_hessenberg_symmetric_stays_symmetric(self):
M = random_symmetric(4)
hess = multi.hessenberg(M)
self.assertTrue(is_symmetric(hess))
