Function _sample_conditional

gpflow/conditionals/sample_conditionals.py:40–91

`sample_conditional` will return a 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 rou

(
    Xnew: tf.Tensor,
    inducing_variable: InducingVariables,
    kernel: Kernel,
    f: tf.Tensor,
    *,
    full_cov: bool = False,
    full_output_cov: bool = False,
    q_sqrt: Optional[tf.Tensor] = None,
    white: bool = False,
    num_samples: Optional[int] = None,
)

    "return[2]: [batch..., N, R, R] if (not full_cov) and full_output_cov",
)
def _sample_conditional(
    Xnew: tf.Tensor,
    inducing_variable: InducingVariables,
    kernel: Kernel,
    f: tf.Tensor,
    *,
    full_cov: bool = False,
    full_output_cov: bool = False,
    q_sqrt: Optional[tf.Tensor] = None,
    white: bool = False,
    num_samples: Optional[int] = None,
) -> SamplesMeanAndVariance:
    """
    `sample_conditional` will return a 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: samples, mean, cov
    """

    if full_cov and full_output_cov:
        msg = "The combination of both `full_cov` and `full_output_cov` is not permitted."
        raise NotImplementedError(msg)

    mean, cov = conditional(
        Xnew,
        inducing_variable,
        kernel,
        f,
        q_sqrt=q_sqrt,
        white=white,
        full_cov=full_cov,
        full_output_cov=full_output_cov,
    )
    if full_cov:
        # mean: [..., N, P]
        # cov: [..., P, N, N]
        mean_for_sample = tf.linalg.adjoint(mean)  # [..., P, N]
        samples = sample_mvn(
            mean_for_sample, cov, full_cov=True, num_samples=num_samples
        )  # [..., (S), P, N]
        samples = tf.linalg.adjoint(samples)  # [..., (S), N, P]
    else:
        # mean: [..., N, P]
        # cov: [..., N, P] or [..., N, P, P]
        samples = sample_mvn(
            mean, cov, full_cov=full_output_cov, num_samples=num_samples
        )  # [..., (S), N, P]

    return samples, mean, cov