OperatorPosterior#

class OperatorPosterior(means, precisions, *, alreadyinverted=False)[source]#

Posterior distribution for operator matrices.

Operator inference models are uniquely determined by operator matrices \(\Ohat\in\RR^{r\times d}\) that concatenate the entries of all operators in the model. For example, the time-continuous model

\[\ddt\qhat(t) = \chat + \Ahat\qhat(t) + \Hhat[\qhat(t)\otimes\qhat(t)]\]

is uniquely determined by the operator matrix

\[\Ohat = [~\chat~~\Ahat~~\Hhat~] \in \RR^{r \times d}.\]

Typical deterministic operator inference learns a single operator matrix \(\Ohat\) from state measurements, while probabilistic or Bayesian operator inference constructs a distribution of operator matrices, \(p(\Ohat)\). This class implements an operator matrix distribution where the rows of \(\Ohat\) are multivariate Normal (Gaussian) random variables, i.e.,

\[\begin{split}p(\ohat_{i}) = \mathcal{N}(\ohat_i\mid\bfmu_i,\bfSigma_i), \\ \bfmu_i \in \RR^{d}, \quad \bfSigma_i \in \RR^{d\times d}, \quad i = 0, \ldots, r-1,\end{split}\]

where \(\ohat_i \in \RR^{d}\) is the \(i\)-th row of \(\Ohat\).

The BayesianROM class has a posterior attribute that is an OperatorPosterior object.

Parameters:
meanslist of r (d,) ndarrays

Mean values for each row of the operator matrix.

precisionslist of r (d, d) ndarrays

INVERSE covariance matrices for each row of the operator matrix.

alreadyinvertedbool

If True, assume precisions is the collection of covariance matrices, not their inverses.

Properties:
covs#

Covariance matrices \(\bfSigma_0,\ldots,\bfSigma_{r-1}\in\RR^{d\times d}\) for the rows of the operator matrix.

means#

Mean vectors \(\bfmu_0,\ldots,\bfmu_{r-1}\in\RR^{d}\) for the rows of the operator matrix.

nrows#

Number of rows \(r\) in the data matrix. This is also the state dimension of the corresponding model.

randomvariables#

Multivariate normal random variables for the rows of the operator matrix.

Methods:

load

Load a previously saved posterior operator distribution.

rvs

Draw a random operator matrix from the posterior operator distribution.

save

Save the posterior operator distribution.