Skip to content

Some mapreduce improvement #63

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Jan 17, 2022
Merged

Some mapreduce improvement #63

Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
ad3db62
fully drop `_extrema_dims`, `_extrema_itr`
N5N3 Jan 15, 2022
5f67312
Generalize sparse `mapreduce`
N5N3 Jan 15, 2022
b119e05
fix empty reduction test
N5N3 Jan 15, 2022
bec4e19
Add NaN test
N5N3 Jan 15, 2022
f471cd5
Fix CI temporarily
N5N3 Jan 15, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
fully drop _extrema_dims, _extrema_itr
  • Loading branch information
@N5N3
N5N3 committed Jan 16, 2022
commit ad3db62d7ac46b6ee3ee77f2bb34cf288d83b3ca
59 changes: 1 addition & 58 deletions src/higherorderfns.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ module HigherOrderFns

# This module provides higher order functions specialized for sparse arrays,
# particularly map[!]/broadcast[!] for SparseVectors and SparseMatrixCSCs at present.
import Base: map, map!, broadcast, copy, copyto!, _extrema_dims, _extrema_itr
import Base: map, map!, broadcast, copy, copyto!

using Base: front, tail, to_shape
using ..SparseArrays: SparseVector, SparseMatrixCSC, AbstractSparseVector, AbstractSparseMatrixCSC,
Expand All @@ -29,7 +29,6 @@ using LinearAlgebra
# (11) Define broadcast[!] methods handling combinations of scalars, sparse vectors/matrices,
# structured matrices, and one- and two-dimensional Arrays.
# (12) Define map[!] methods handling combinations of sparse and structured matrices.
# (13) Define extrema methods optimized for sparse vectors/matrices.


# (0) BroadcastStyle rules and convenience types for dispatch
Expand Down Expand Up @@ -1163,60 +1162,4 @@ map(f::Tf, A::SparseOrStructuredMatrix, Bs::Vararg{SparseOrStructuredMatrix,N})
map!(f::Tf, C::AbstractSparseMatrixCSC, A::SparseOrStructuredMatrix, Bs::Vararg{SparseOrStructuredMatrix,N}) where {Tf,N} =
(_checksameshape(C, A, Bs...); _noshapecheck_map!(f, C, _sparsifystructured(A), map(_sparsifystructured, Bs)...))


# (13) extrema methods optimized for sparse vectors/matrices.
function _extrema_itr(f, A::SparseVecOrMat)
M = length(A)
iszero(M) && throw(ArgumentError("Sparse array must have at least one element."))
N = nnz(A)
iszero(N) && return f(zero(eltype(A))), f(zero(eltype(A)))
vmin, vmax = _extrema_itr(f, nonzeros(A))
if N != M
f0 = f(zero(eltype(A)))
vmin = min(f0, vmin)
vmax = max(f0, vmax)
end
vmin, vmax
end

function _extrema_dims(f, x::SparseVector, dims)
sz = zeros(1)
for d in dims
sz[d] = 1
end
if sz == [1] && !iszero(length(x))
return [_extrema_itr(f, x)]
end
invoke(_extrema_dims, Tuple{Any, AbstractArray, Any}, f, x, dims)
end

function _extrema_dims(f, A::AbstractSparseMatrix, dims)
sz = zeros(2)
for d in dims
sz[d] = 1
end
if sz == [1, 0] && !iszero(length(A))
T = eltype(A)
B = Array{Tuple{T,T}}(undef, 1, size(A, 2))
@inbounds for col_idx in 1:size(A, 2)
col = @view A[:,col_idx]
fx = (nnz(col) == size(A, 1)) ? f(A[1,col_idx]) : f(zero(T))
B[col_idx] = (fx, fx)
for x in nonzeros(col)
fx = f(x)
if fx < B[col_idx][1]
B[col_idx] = (fx, B[col_idx][2])
elseif fx > B[col_idx][2]
B[col_idx] = (B[col_idx][1], fx)
end
end
end
return B
end
invoke(_extrema_dims, Tuple{Any, AbstractArray, Any}, f, A, dims)
end

_extrema_dims(f, A::SparseVector, ::Colon) = _extrema_itr(f, A)
_extrema_dims(f, A::AbstractSparseMatrix, ::Colon) = _extrema_itr(f, A)

end