The goal of RangeArrays is to provide efficient and convenient array data structures where the columns of the arrays are generated (on the fly) by Ranges.
Two different types of range matrices are currently supported:
RangeMatrix: makes a vector of ranges behave as a matrix; all ranges must be the same length.RepeatedRangeMatrix: one range is repeated multiple times at offsets specified in a vector.
In all cases, indexing is specialized such that it will return an appropriate range or RangeArray if it can.
julia> R = RangeMatrix(1:5,11:15,-2:2)
5x3 RangeArrays.RangeMatrix{Int64,Array{UnitRange{Int64},1}}:
1 11 -2
2 12 -1
3 13 0
4 14 1
5 15 2
julia> R[2:3,:]
2x3 RangeArrays.RangeMatrix{Int64,Array{UnitRange{Int64},1}}:
2 12 -1
3 13 0
julia> RR = RepeatedRangeMatrix(.1:.1:1.0, [0.0,5.0,-20.2,3.3])
10x4 RangeArrays.RepeatedRangeMatrix{Float64,FloatRange{Float64},Array{Float64,1}}:
0.1 5.1 -20.1 3.4
0.2 5.2 -20.0 3.5
0.3 5.3 -19.9 3.6
0.4 5.4 -19.8 3.7
0.5 5.5 -19.7 3.8
0.6 5.6 -19.6 3.9
0.7 5.7 -19.5 4.0
0.8 5.8 -19.4 4.1
0.9 5.9 -19.3 4.2
1.0 6.0 -19.2 4.3
julia> RR[8:-2:2, end]
4.1:-0.2:3.5There is a similar structure available in mbauman/RaggedArrays.jl, which allows for ranges of varying lengths.