def test_vec_and_unvec(self):
a = np.array([[ 5., 1., 14., 2., 42.],
[132., 2., 429., 1., 1.],
[ 1., 2., 1430., 2., 2.]])
b = npu.col(5., 132., 1., 1., 2., 2., 14., 429., 1430., 2., 1., 2., 42., 1., 2.)
npt.assert_almost_equal(npu.vec(a), b)
npt.assert_almost_equal(npu.unvec(b, 3), a)
def noise_covariance(self, time_delta):
mrf_squared = self.mean_reversion_factor_squared(time_delta)
eye_minus_mrf_squared = np.eye(
self.process_dim * self.process_dim) - mrf_squared
return npu.unvec(
np.dot(np.dot(self._transition_x_2_inverse, eye_minus_mrf_squared),
self._cov_vec), self.process_dim)
def noisecovariance(self, timedelta):
mrfsquared = self.meanreversionfactorsquared(timedelta)
eyeminusmrfsquared = np.eye(self.processdim) - mrfsquared
return npu.unvec(
np.dot(np.dot(self.__transitionx2inverse, eyeminusmrfsquared),
self.__covvec), self.processdim)
