Sig coef functions

Sig coef functions#

sig_coef_cuda#

int sig_coef_cuda_f(const float *path, float *out, const uint64_t *multi_idx, uint64_t num_multi_idx, const uint64_t *degrees, uint64_t batch_size, uint64_t dimension, uint64_t length, bool prefixes) noexcept#

Computes signature coefficients for a batch of paths on the GPU.

Parameters:

  • path – Pointer to input batch path data (row-major, on device), size = batch_size * length * dimension.

  • out – Pointer to output buffer (on device, preallocated).

  • multi_idx – Pointer to flattened multi-indices (on device).

  • num_multi_idx – Number of multi-indices (words).

  • degrees – Pointer to degree of each multi-index (on device), size = num_multi_idx.

  • batch_size – Batch size.

  • dimension – Dimension of the paths.

  • length – Length of the paths.

  • prefixes – If true, output all prefix coefficients; if false, output only the final coefficient.

Returns:

Status code (0 = success).

int sig_coef_cuda_d(const double *path, double *out, const uint64_t *multi_idx, uint64_t num_multi_idx, const uint64_t *degrees, uint64_t batch_size, uint64_t dimension, uint64_t length, bool prefixes) noexcept#

sig_coef_backprop_cuda#

int sig_coef_backprop_cuda_f(const float *path, float *out, const float *coefs, const float *derivs, const uint64_t *multi_idx, uint64_t num_multi_idx, const uint64_t *degrees, uint64_t batch_size, uint64_t dimension, uint64_t length) noexcept#

Backpropagates through sig_coef_cuda on the GPU.

Parameters:

  • path – Pointer to input batch path data (row-major, on device), size = batch_size * length * dimension.

  • out – Pointer to output buffer (on device, preallocated), size = batch_size * length * dimension.

  • coefs – Pointer to prefix coefficients from forward pass (on device).

  • derivs – Pointer to dF/d(coefs) (on device), same size as coefs.

  • multi_idx – Pointer to flattened multi-indices (on device).

  • num_multi_idx – Number of multi-indices (words).

  • degrees – Pointer to degree of each multi-index (on device), size = num_multi_idx.

  • batch_size – Batch size.

  • dimension – Dimension of the paths.

  • length – Length of the paths.

Returns:

Status code (0 = success).

int sig_coef_backprop_cuda_d(const double *path, double *out, const double *coefs, const double *derivs, const uint64_t *multi_idx, uint64_t num_multi_idx, const uint64_t *degrees, uint64_t batch_size, uint64_t dimension, uint64_t length) 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