sparsegrad.impl.sparsevec package

Submodules

sparsegrad.impl.sparsevec.sparsevec module

This module contains implementation details of summing sparse vectors.

sparsegrad.impl.sparsevec.sparsevec.sparsesum(terms, hstack=<function hstack>, nvalue=<function <lambda>>, wrap=<function <lambda>>, check_unique=False, return_sparse=False)

Sum sparse vectors

This is a general function, which propagates index information and numerical values. Suitable functions must be supplied as arguments to propagate other information.

terms : list of sparsevec

terms

hstack : callable(vectors)

function to use for concatenating vectors

nvalue : callable(vector)

function to use for extracting numerical value

wrap : callable(idx, v, result)

function to use for wrapping the result, with idx, v being concatenated inputs

check_unique : bool

whether to perform test for double assignments (useful when this function is used to replace item assignment)

return_sparse : bool

whether to calculate sparse results

class sparsegrad.impl.sparsevec.sparsevec.sparsevec(n, idx, v)

Bases: object

Sparse vector of length n, with nonzero entries in arrays (idx,v) where idx contains the indices of entries, and v contains the corresponding values

Module contents