API Reference¶
Least-Squares estimators¶
|
Linear model minimizing the \(L^{2}\) loss. |
Kernel-based estimators¶
|
Kernel model minimizing the \(L^{2}\) loss. |
|
Nyström-accelerated Kernel model minimizing the \(L^{2}\) loss. |
|
Kernel-based estimator for the infinitesimal generator of diffusion processes using the Dirichlet-form method. |
Preprocessing utilities¶
|
A scikit-learn compatible transformer that constructs time-delay embeddings (temporal windows) from trajectory data, with a configurable stride. |
A scikit-learn compatible transformer that flattens multi-dimensional trajectories into a 2D array, and restores them to their original shape when inverted. |
Datasets¶
|
Computes the eigenvalues and eigenfunctions of the Generator of the Overdamped Langevin dynamics for the Prinz potential. |
|
Fetch the MNIST dataset and return an ordered subset interleaving samples from each digit class. |
|
Generate a trajectory from the Duffing oscillator. |
|
Generate a trajectory from a discrete-time linear dynamical system. |
|
Generate a trajectory from the logistic map with optional trigonometric noise Ostruszka et al. [1]. |
|
Generate a trajectory from the Lorenz-63 system. |
|
Generate a 1D Langevin trajectory for the "Prinz potential" Prinz et al. [1]. |
|
Generate a trajectory from a regime-switching vector autoregressive (VAR) process. |
PyTorch Integration¶
Spectral contrastive loss based originally introduced by HaoChen et al. [1], and adopted for evolution operators in Turri et al. [2], Jeong et al. [3] |
|
|
Variational Approach for learning Markov Processes (VAMP) score by Wu and Noé [1]. |
|
Single-step Dynamic Autoencoder (DAE) loss introduced by Lusch et al. [1]. |
JAX Integration¶
Spectral contrastive loss for self-supervised learning. |
|
|
Variational Approach for learning Markov Processes (VAMP) score. |
|
Single-step Dynamic Autoencoder (DAE) loss. |