pysiglib.branched_sig_kernel_gram

pysiglib.branched_sig_kernel_gram#

pysiglib.branched_sig_kernel_gram computes all pairwise branched signature kernels between two batches of paths. It accepts the same kernel parameters as pysiglib.branched_sig_kernel and supports max_batch chunking for large Gram matrices.

branched_sig_kernel_gram(path1, path2, depth, dyadic_order, *, static_kernel=None, time_aug=False, lead_lag=False, end_time=1.0, n_jobs=1, max_batch=-1, return_grid=False, normalize=False)[source]#

Computes a Gram matrix of branched signature kernels.

Parameters:

  • path1 (numpy.ndarray | torch.Tensor) – First path batch, shape (*batch_shape_1, length_1, dimension).

  • path2 (numpy.ndarray | torch.Tensor) – Second path batch, shape (*batch_shape_2, length_2, dimension).

  • depth (int) – Forest depth truncation for the branched kernel.

  • dyadic_order (int | tuple) – Dyadic refinement order, or a pair of orders.

  • max_batch (int) – Maximum side length of pair chunks. -1 uses all pairs.

  • return_grid (bool) – If True, returns final-depth grids per pair.

  • normalize (bool) – If True, normalizes the Gram matrix.

Returns:

Gram matrix of branched signature kernels.

Return type:

numpy.ndarray | torch.Tensor


Citation#

If you found this library useful in your research, please consider citing the paper:

@article{shmelev2025pysiglib,
  title={pySigLib-Fast Signature-Based Computations on CPU and GPU},
  author={Shmelev, Daniil and Salvi, Cristopher},
  journal={arXiv preprint arXiv:2509.10613},
  year={2025}
}
Contents