Integrates numba into interpolation & kernel functions, to create parallelable machine code. This results in performance increases of up to 40x-50x in some interpolation operations.
To integrate numba, the following changes have been made to the code structure:
Each interpolation function is now separated in two pieces, with the user-facing function converting the arguments to work with the underlying numba-compiled function.
The structure of BaseKernel has changed, with the weight method changed to a static function, which takes a q-value and the number of dimensions.
Integrates numba into interpolation & kernel functions, to create parallelable machine code. This results in performance increases of up to 40x-50x in some interpolation operations.
To integrate numba, the following changes have been made to the code structure:
BaseKernelhas changed, with the weight method changed to a static function, which takes a q-value and the number of dimensions.