eschnett / SparseArraysCOO.jl

Create sparse vectors and matrices conveniently and efficiently

SparseArraysCOO.jl

Create sparse vectors and matrices conveniently and efficiently

• 
• 
• 

This package SparseArraysCOO.jl provides simple convenience functions for creating sparse vectors and sparse matrices. After creating a SparseMatrixCOO matrix or a SparseVectorCOO vector, it records all assignments made to it. There are no other operations defined on these types, except converting them to the efficient SparseMatrixCSC and SparseVector types. This is no more or less efficient than the way described in the documentation of SparseArrays, but it is more convenient.

The suffix COO stands for "coordinate format". This means that the internal representation stores both indices i and j for every nonzero value.

Examples

Creating a sparse matrix:

julia> using SparseArrays, SparseArraysCOO
julia> A = SparseMatrixCOO{Float64}(4, 4)
SparseMatrixCOO{Float64}(4, 4, Int64[], Int64[], Float64[])

julia> A[1, 1] = 1
julia> A[2, 3] = 4
julia> A[4, 1] = 17
julia> A = sparse(A)
4×4 SparseMatrixCSC{Float64, Int64} with 3 stored entries:
  1.0   ⋅    ⋅    ⋅
   ⋅    ⋅   4.0   ⋅
   ⋅    ⋅    ⋅    ⋅
 17.0   ⋅    ⋅    ⋅

julia> typeof(A)
SparseMatrixCSC{Float64, Int64}

Creating a sparse vector:

julia> using SparseArrays, SparseArraysCOO
julia> x = SparseVectorCOO{Float64}(4)
SparseVectorCOO{Float64}(4, Int64[], Float64[])

julia> x[1] = 1
julia> x[2] = 4
julia> x[4] = 17
julia> x = sparse(x)
4-element SparseVector{Float64, Int64} with 3 stored entries:
  [1]  =  1.0
  [2]  =  4.0
  [4]  =  17.0

julia> typeof(x)
SparseVector{Float64, Int64}