scipy.

LowLevelCallable#

class scipy.LowLevelCallable(function, user_data=None, signature=None)[source]#

Low-level callback function.

Some functions in SciPy take as arguments callback functions, which can either be python callables or low-level compiled functions. Using compiled callback functions can improve performance somewhat by avoiding wrapping data in Python objects.

Such low-level functions in SciPy are wrapped in LowLevelCallable objects, which can be constructed from function pointers obtained from ctypes, cffi, Cython, or contained in Python PyCapsule objects.

See also

Functions accepting low-level callables:

scipy.integrate.quad, scipy.ndimage.generic_filter, scipy.ndimage.generic_filter1d, scipy.ndimage.geometric_transform

Usage examples:

Extending scipy.ndimage in C, Faster integration using low-level callback functions

Parameters:

function{PyCapsule, ctypes function pointer, cffi function pointer}: Low-level callback function.
user_data{PyCapsule, ctypes void pointer, cffi void pointer}: User data to pass on to the callback function.
signaturestr, optional: Signature of the function. If omitted, determined from function, if possible.

Attributes:

function: Callback function given.
user_data: User data given.
signature: Signature of the function.

Methods

from_cython(module, name[, user_data, signature])

Create a low-level callback function from an exported Cython function.

Notes

The argument function can be one of:

PyCapsule, whose name contains the C function signature
ctypes function pointer
cffi function pointer

The signature of the low-level callback must match one of those expected by the routine it is passed to.

If constructing low-level functions from a PyCapsule, the name of the capsule must be the corresponding signature, in the format:

return_type (arg1_type, arg2_type, ...)

For example:

"void (double)"
"double (double, int *, void *)"

The context of a PyCapsule passed in as function is used as user_data, if an explicit value for user_data was not given.