description: Use a Cholesky-like function for GaussianProcess marginal_fn.
View source on GitHub |
Use a Cholesky-like function for GaussianProcess
marginal_fn
.
tfp.experimental.distributions.marginal_fns.make_cholesky_like_marginal_fn(
cholesky_like, name='CholeskyLikeMarginalFn'
)
For use with "Cholesky-like" lower-triangular factorizations (LLT). See
make_backoff_cholesky
for one way to create such functions.
Args | |
---|---|
cholesky_like
|
A callable with the same signature as tf.linalg.cholesky.
|
name
|
Python str name prefixed to Ops created by this function.
Default value: 'CholeskyLikeMarginalFn'.
|
Returns | |
---|---|
marginal_function
|
A function that can be used with the marginal_fn
argument to GaussianProcess .
|