ResizingTools helps you create resizable Array types.
SimpleRDArray is a simple implementation of the resizable dense array, which
can be created simply:
julia> M = reshape(1:9, 3, 3)
3×3 reshape(::UnitRange{Int64}, 3, 3) with eltype Int64:
1 4 7
2 5 8
3 6 9
julia> RM = SimpleRDArray(M)
3×3 SimpleRDArray{Int64, 2}:
1 4 7
2 5 8
3 6 9
julia> M == RM
trueOnce a SimpleRDArray is created, you can almost do anything with which likes a
normal Array with similar performance:
julia> @benchmark $RM * $RM
BenchmarkTools.Trial: 10000 samples with 980 evaluations.
Range (min … max): 77.283 ns … 26.663 μs ┊ GC (min … max): 0.00% … 0.00%
Time (median): 214.530 ns ┊ GC (median): 0.00%
Time (mean ± σ): 300.628 ns ± 1.292 μs ┊ GC (mean ± σ): 6.87% ± 5.58%
█▂
▁▁▁▆██▄▂▁▁▁▁▁▁▁▁▂▂▂▃▃▃▃▃▄▄▄▄▄▄▄▄▃▃▃▃▃▃▂▂▂▂▂▂▂▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁ ▂
77.3 ns Histogram: frequency by time 430 ns <
Memory estimate: 160 bytes, allocs estimate: 1.
julia> @benchmark $M * $M
BenchmarkTools.Trial: 10000 samples with 980 evaluations.
Range (min … max): 71.658 ns … 4.237 μs ┊ GC (min … max): 0.00% … 94.53%
Time (median): 144.504 ns ┊ GC (median): 0.00%
Time (mean ± σ): 161.228 ns ± 206.767 ns ┊ GC (mean ± σ): 7.40% ± 5.60%
▁▅▇█▇▅▂
▂▁▂▂▂▂▂▂▃▄▅▆███████▇▄▃▃▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂ ▃
71.7 ns Histogram: frequency by time 356 ns <
Memory estimate: 160 bytes, allocs estimate: 1.
julia> RM * RM == M * M
trueBesides, a SimpleRDArray can be resized in many ways:
julia> resize!(RM, (4, 4)) # resize RM to 4 * 4
4×4 SimpleRDArray{Int64, 2}:
1 4 7 81
2 5 8 96
3 6 9 102
4 8 66 126
julia> RM[1:3,1:3] == M
true
julia> resize!(RM, 2, 3) # resize the 2nd dimension of RM to 3
4×3 SimpleRDArray{Int64, 2}:
1 4 7
2 5 8
3 6 9
4 8 66
julia> RM[4, :] .= 0
3-element view(::SimpleRDArray{Int64, 2}, 4, :) with eltype Int64:
0
0
0
julia> resize!(RM, 1, Bool[1, 1, 0, 1]) # delete RM[3, :]
3×3 SimpleRDArray{Int64, 2}:
1 4 7
2 5 8
0 0 0To make your own resizable array, you only need is defined some interface methods, see docs for details.