def diag(self):
"""
return the diagonal of the matrix
"""
if not self.processed:
self.process()
scaled_diag = izmat.izdiag(self.N, self.scaled_z_nu, self.scaled_z_nd,
self.n, self.scaled_z_u, self.scaled_z_d)
return scaled_diag * self.v
def invDiag(self):
"""
return the diagonal of the inverse covariate matrix
"""
if not self.processed:
self.process()
inv_scaled_diag = izmat.izdiag(self.N, self.scaled_z_nu,
self.scaled_z_nd, self.n,
self.scaled_z_u, self.inv_scaled_z_d)
return inv_scaled_diag / self.v
def izmat_izdiag():
import numpy as np
from limetr.special_mat import izmat
ok = True
tol = 1e-10
# setup problem
# -------------------------------------------------------------------------
k = 3
n = np.array([5, 2, 4])
m = n.size
z_list = [np.random.randn(n[i], k) for i in range(m)]
z = np.vstack(z_list)
ns = np.minimum(n, k)
nu = ns * n
nx = n
nz = n
u = np.zeros(nu.sum())
s = np.zeros(ns.sum())
izmat.zdecomp(nz, nu, ns, z, u, s)
my_y = izmat.izdiag(n.sum(), nu, ns, nx, u, s**2)
y_list = [
np.diag(np.eye(n[i]) + z_list[i].dot(z_list[i].T)) for i in range(m)
]
tr_y = np.hstack(y_list)
err = np.linalg.norm(tr_y - my_y)
if not ok:
print('err in izdiag')
print('err:', err)
return ok