Sig combine functions

Sig combine functions#

sig_combine#

int sig_combine_f(const float *sig1, const float *sig2, float *out, uint64_t batch_size, uint64_t dimension, uint64_t degree, bool scalar_term = true, int n_jobs = 1) noexcept#

Combines pairs of truncated signatures of the same degree and dimension.

Parameters:

  • sig1 – Pointer to the batch of first truncated signatures (row-major), size = batch_size * sig_length(dimension, degree).

  • sig2 – Pointer to the batch of second truncated signatures (row-major), size = batch_size * sig_length(dimension, degree). Must have the same batch size, degree and dimension as the first.

  • out – Pointer to the output buffer (row-major, preallocated), size = batch_size * sig_length(dimension, degree).

  • batch_size – Batch size of sig1 and sig2.

  • dimension – Dimension of the underlying paths.

  • degree – Truncation degree of the signatures.

  • n_jobs – Number of threads to run in parallel. If n_jobs = 1, the computation is run serially. If set to -1, all available threads are used. For n_jobs below -1, (max_threads + 1 + n_jobs) threads are used. For example if n_jobs = -2, all threads but one are used (default = 1).

Returns:

Status code (0 = success).

int sig_combine_d(const double *sig1, const double *sig2, double *out, uint64_t batch_size, uint64_t dimension, uint64_t degree, bool scalar_term = true, int n_jobs = 1) noexcept#

sig_combine_backprop#

int sig_combine_backprop_f(const float *sig_combined_derivs, float *sig1_deriv, float *sig2_deriv, const float *sig1, const float *sig2, uint64_t batch_size, uint64_t dimension, uint64_t degree, bool scalar_term = true, int n_jobs = 1) noexcept#

Backpropagation through the sig_combine function.

Parameters:

  • sig_combined_derivs – Pointer to the derivatives with respect to the combined signatures (row-major), size = batch_size * sig_length(dimension, degree).

  • sig1_deriv – Pointer to the output buffer for the derivatives with respect to sig1 (row-major, preallocated), size = batch_size * sig_length(dimension, degree).

  • sig2_deriv – Pointer to the output buffer for the derivatives with respect to sig2 (row-major, preallocated), size = batch_size * sig_length(dimension, degree).

  • sig1 – Pointer to the batch of first truncated signatures (row-major, precomputed), size = batch_size * sig_length(dimension, degree).

  • sig2 – Pointer to the batch of second truncated signatures (row-major, precomputed), size = batch_size * sig_length(dimension, degree). Must have the same batch size, degree and dimension as sig1.

  • batch_size – Batch size of the signatures.

  • dimension – Dimension of the underlying paths.

  • degree – Truncation degree of the signatures.

  • n_jobs – Number of threads to run in parallel. If n_jobs = 1, the computation is run serially. If set to -1, all available threads are used. For n_jobs below -1, (max_threads + 1 + n_jobs) threads are used. For example if n_jobs = -2, all threads but one are used (default = 1).

Returns:

Status code (0 = success).

int sig_combine_backprop_d(const double *sig_combined_derivs, double *sig1_deriv, double *sig2_deriv, const double *sig1, const double *sig2, uint64_t batch_size, uint64_t dimension, uint64_t degree, bool scalar_term = true, int n_jobs = 1) noexcept#

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