# pygpcca.GPCCA.input_distribution

property GPCCA.input_distribution: numpy.ndarray

Input probability distribution of the (micro)states.

In theory $$\eta$$ can be an arbitrary distribution as long as it is a valid probability distribution (i.e., sums up to 1). A neutral and valid choice would be the uniform distribution (default).

In case of a reversible transition matrix, the stationary distribution $$\pi$$ can (but don’t has to) be used here. In case of a non-reversible P, some initial or average distribution of the states might be chosen instead of the uniform distribution.

Vector of shape (n,) which sums to 1.

Return type

ndarray