def _sample_conditional(Xnew,
feat,
kern,
f,
*,
full_output_cov=False,
q_sqrt=None,
white=False):
"""
`sample_conditional` will return amplitude sample from the conditional distribution.
In most cases this means calculating the conditional mean m and variance v and then
returning m + sqrt(v) * eps, with eps ~ N(0, 1).
However, for some combinations of Mok and Mof more efficient sampling routines exists.
The dispatcher will make sure that we use the most efficient one.
:return: N x P (full_output_cov = False) or N x P x P (full_output_cov = True)
"""
logger.debug("sample conditional: InducingFeature Kernel")
mean, var = conditional(Xnew,
feat,
kern,
f,
full_cov=False,
full_output_cov=full_output_cov,
q_sqrt=q_sqrt,
white=white) # N x P, N x P (x P)
cov_structure = "full" if full_output_cov else "diag"
return _sample_mvn(mean, var, cov_structure)
def _sample_conditional(Xnew, X, kern, f, *, q_sqrt=None, white=False):
logger.debug("sample conditional: Kernel")
mean, var = conditional(Xnew,
X,
kern,
f,
q_sqrt=q_sqrt,
white=white,
full_cov=False) # N x P, N x P
return _sample_mvn(mean, var, "diag") # N x P