lasdi.latent_dynamics

Submodules

lasdi.latent_dynamics.sindy

Classes

LatentDynamics

Package Contents

class lasdi.latent_dynamics.LatentDynamics(dim_, nt_)

dim

nt

ncoefs

coefs

calibrate(Z, dt, compute_loss=True, numpy=False)

calibrate coefficients of the latent dynamics and compute loss for one given time series Z.

Z is the encoder output (2D/3D tensor), with shape [time_dim, space_dim] or [n_train, time_dim, space_dim]

simulate(coefs, z0, t_grid): time-integrate with one initial condition z0 at time points t_grid, for one given set of coefficients

sample(coefs_sample, z0_sample, t_grid): Sample time series for given sample initial conditions and coefficients.

export()

load(dict_)