Skip to main content

Module spatial

Module spatial 

Source
Expand description

Operators that mix channels: each output channel is a weighted sum of input channels.

  • car: common average and common median reference (subtract the mean or the median over channels at every sample).
  • SpatialWhitening: ZCA whitening, global or over each channel’s nearest contacts.
  • SurfaceLaplacian: each channel minus the (distance-weighted) mean of its neighbours (HD-EMG, ECoG grids).
  • sparse: the [channels, channels] matrix behind whitening and the Laplacian, applied dense or as sparse rows (CAR has its own single-pass kernel).

Re-exports§

pub use car::common_median_bitonic_kernel;
pub use car::common_median_bitonic_kernel;
pub use car::common_median_kernel;
pub use car::common_median_kernel;
pub use car::direct_car_kernel;
pub use car::direct_car_kernel;
pub use car::execute_common_median;
pub use car::execute_common_median_with;
pub use car::execute_direct_car;
pub use car::execute_direct_car_with;
pub use car::CarKernel;
pub use car::CmrKernel;
pub use laplacian::SurfaceLaplacian;
pub use sparse::execute_sparse_rows_multiply;
pub use sparse::execute_spatial_matrix_multiply;
pub use sparse::DeviceSpatialMatrix;
pub use sparse::SparseRows;
pub use sparse::SPARSE_MAX_ROW_FILL;
pub use whitening::SpatialWhitening;
pub use whitening::DEFAULT_WHITENING_EPSILON;
pub use whitening::MIN_WHITENING_EPSILON;

Modules§

car
kernels
laplacian
sparse
Spatial [C, C] operators on the device, stored dense or as sparse rows depending on how many weights are nonzero.
whitening