TESST

TESST is an experimental GPU implementation related to ESST. It batches Hankel matrices and uses torch.svd_lowrank on a CUDA device.

Experimental

TESST requires a CUDA-enabled PyTorch installation and is not part of the package's stable algorithm API.

Parameters

window_length : int

Subsequence length in samples.

n_windows : int, optional

Number of subsequences per side. Defaults to window_length // 2.

lag : int, optional

Separation between compared matrices. Defaults to n_windows.

rank : int, default 5

Low-rank decomposition size.

scoring_step : int, default 1

Distance between scored positions.

scale : bool, default True

Min-max scale the signal before batching.

Methods

transform(time_series)

time_series : numpy.ndarray, shape (n_samples,)

One-dimensional signal.

Returns

score : numpy.ndarray, shape (n_samples,)

Experimental GPU-computed entangled score.

Minimal Example

import numpy as np
from changepoynt.algorithms.torch_esst import TESST

signal = np.sin(np.linspace(0, 20 * np.pi, 400))
detector = TESST(window_length=40)  # requires CUDA-enabled PyTorch
score = detector.transform(signal)

TESST does not currently have a separate publication; see the ESST reference for the related score family.