# How simple get vector from ranges?

**URL:** <https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537>\
**Category:** General Usage\
**Tags:** question, performance\
**Created:** [August 24, 2017, 12:03pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537 "2017-08-24T12:03:15Z")\
**Posts on this page:** 15\
**Page:** 1

<div class="post-metadata">

**Author:** ![Martin\_Florek](https://avatars.discourse-cdn.com/v4/letter/m/f4b2a3/32.png) [@Martin\_Florek](https://discourse.julialang.org/u/Martin_Florek)\
**Post date:** [August 24, 2017, 12:03pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/1 "2017-08-24T12:03:15Z")

</div>

In Matlab I have vector and vector of ranges.

> a = [10., 9., 8., 7., 6., 5., 4., 3., 2., 1.]  
> dr = [1:2, 4:7]

If I want a new vector from ranges I will call.

> mo = a(dr)  
> mo = 10 9 7 6 5 4

In Julia I started with the function of `map()` and I got this.

```julia
jHelp = map(x -> a[x], dr)
JHelp = Array{Array{Float64,1},1}[[10., 9.], [7., 6., 5., 4.]]

```

After much effort 🙂 , I worked on the same result as Matlab with the help of `mapreduce()`.  
.

```julia
jo = mapreduce(x -> a[x], vcat, dr)
jo = [10., 9., 7., 6., 5., 4.]

```

There is some simpler alternative or this is the best possible solution?  
Thanks.

---

<div class="post-metadata">

**Author:** ![mauro3](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mauro3/32/292_2.png) [@mauro3](https://discourse.julialang.org/u/mauro3)\
**Post date:** [August 24, 2017, 12:08pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/2 "2017-08-24T12:08:04Z")

</div>

Does this help:

```julia
julia> [collect(1:4); collect(6:9)]                                                                                                                                                                                
8-element Array{Int64,1}:                                                                                                                                                                                          
 1                                                                                                                                                                                                                 
 2                                                                                                                                                                                                                 
 3                                                                                                                                                                                                                 
 4                                                                                                                                                                                                                 
 6                                                                                                                                                                                                                 
 7                                                                                                                                                                                                                 
 8                                                                                                                                                                                                                 
 9                                                                                                                                                                                                                 

```

`collect` turns the range into a vector, then the `;` inside the `[]` concatenates. That should help you make the `dr` into a plain vector which you can use to index `a`.

---

<div class="post-metadata">

**Author:** ![tkoolen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tkoolen/32/1603_2.png) [@tkoolen](https://discourse.julialang.org/u/tkoolen)\
**Post date:** [August 24, 2017, 12:11pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/3 "2017-08-24T12:11:27Z")

</div>

How about

```julia
using Base.Iterators: flatten
[x for (index, x) in enumerate(a) if index in flatten(dr)]

```

or

```julia
[a[i] for i in flatten(dr)]

```

although I quite like your `mapreduce` solution.

---

<div class="post-metadata">

**Author:** ![yuyichao](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/yuyichao/32/20_2.png) [@yuyichao](https://discourse.julialang.org/u/yuyichao)\
**Post date:** [August 24, 2017, 12:50pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/4 "2017-08-24T12:50:27Z")

</div>

> [@Martin\_Florek](#):
>
> dr = [1:2, 4:7]

```julia
julia> a = [10., 9., 8., 7., 6., 5., 4., 3., 2., 1.]
10-element Array{Float64,1}:
d 10.0
  9.0
  8.0
  7.0
  6.0
  5.0
  4.0
  3.0
  2.0
  1.0

julia> dr = [1:2; 4:7]
6-element Array{Int64,1}:
 1
 2
 4
 5
 6
 7

julia> a[dr]
6-element Array{Float64,1}:
 10.0
  9.0
  7.0
  6.0
  5.0
  4.0

```

---

<div class="post-metadata">

**Author:** ![yuyichao](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/yuyichao/32/20_2.png) [@yuyichao](https://discourse.julialang.org/u/yuyichao)\
**Post date:** [August 24, 2017, 12:57pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/5 "2017-08-24T12:57:35Z")

</div>

> [@Martin\_Florek](#):
>
> In Matlab I have vector and vector of ranges.

And note that you don’t have a vector of ranges in matlab. Ranges isn’t even a (user visible) type in matlab AFAICT.

---

<div class="post-metadata">

**Author:** ![tkoolen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tkoolen/32/1603_2.png) [@tkoolen](https://discourse.julialang.org/u/tkoolen)\
**Post date:** [August 24, 2017, 1:02pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/6 "2017-08-24T13:02:22Z")

</div>

Sure, but creating a dense index vector in this manner would result in double the allocations asymptotically compared to the `mapreduce` or `flatten` approaches.

---

<div class="post-metadata">

**Author:** ![Martin\_Florek](https://avatars.discourse-cdn.com/v4/letter/m/f4b2a3/32.png) [@Martin\_Florek](https://discourse.julialang.org/u/Martin_Florek)\
**Post date:** [August 24, 2017, 1:30pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/7 "2017-08-24T13:30:22Z")

</div>

@mauro3 and @yuyichao It seems that the semicolon could be a solution, even without the use of `collect()` function. The question remains how to create / transmit such a vector that is unnecessarily long. In the sense that I am working with big data and transfer the vector of ranges through multiple functions.

```julia
Array{UnitRange{Int64},1}[1:2, 4:7]
Array{Int64,1}[1:2; 4:7] # with semicolon ';'

```

A possible solution would be to transfer a vector as Array {UnitRange {Int64}, 1} and call it as Array {Int64,1} directly with `vcat()`. What do you think?

```julia
julia> a[vcat(dr...)]
6-element Array{Float64,1}:
 10.0
  9.0
  7.0
  6.0
  5.0
  4.0

```

Thanks and thank @yuyichao for correction about AFAICT.

---

<div class="post-metadata">

**Author:** ![mauro3](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mauro3/32/292_2.png) [@mauro3](https://discourse.julialang.org/u/mauro3)\
**Post date:** [August 24, 2017, 1:47pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/8 "2017-08-24T13:47:25Z")

</div>

`vcat(dr...)` produces the same vector `[1:2; 4:7]`, so it does not save memory. Also your `mapreduce` solution produces two vectors which then get `vcat`ed. I think @tkoolen’s second solution is best.

---

<div class="post-metadata">

**Author:** ![tkoolen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tkoolen/32/1603_2.png) [@tkoolen](https://discourse.julialang.org/u/tkoolen)\
**Post date:** [August 24, 2017, 1:49pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/9 "2017-08-24T13:49:36Z")

</div>

> [@mauro3](#):
>
> Also your `mapreduce` solution produces two vectors which then get `vcat`ed

Ah yes of course, I missed that.

---

<div class="post-metadata">

**Author:** ![mkborregaard](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkborregaard/32/556_2.png) [@mkborregaard](https://discourse.julialang.org/u/mkborregaard)\
**Post date:** [August 24, 2017, 2:03pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/10 "2017-08-24T14:03:49Z")

</div>

EDIT: Just realized this is identical to @yuyichao’s opst  
I may be misunderstanding the question completely, but is the issue not just that you use semicolon rather than comma for concatenation in Julia?

```julia
a = [10., 9, 8, 7, 6, 5, 4, 3, 2, 1]
dr = [1:2; 4:7]
a[dr]
#6-element Array{Float64,1}:
# 10.0
# 9.0
# 7.0
# 6.0
# 5.0
# 4.0

```

---

<div class="post-metadata">

**Author:** ![tkoolen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tkoolen/32/1603_2.png) [@tkoolen](https://discourse.julialang.org/u/tkoolen)\
**Post date:** [August 24, 2017, 3:07pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/11 "2017-08-24T15:07:21Z")

</div>

For fun I benchmarked the `vcat`-the-indices approach and the `flatten` approach for large `a` and `dr` (with `dr` a `Vector{UnitRange{Int}}` as in the original post, to allow for fair comparison).

```julia

using BenchmarkTools
using Base.Iterators: flatten

a = rand(10_000_000)
dr = [10 : 100, 150 : 6_000, 10_000 : 8_000_000]

function f1(a, dr)
    a[vcat(dr...)]
end

function f2(a, dr)
    [a[i] for i in flatten(dr)]
end

@benchmark f1($a, $dr)

@benchmark f2($a, $dr)

```

Results for `f1` (`vcat`-the-indices):

```julia
BenchmarkTools.Trial: 
  memory estimate: 122.01 MiB
  allocs estimate: 7
  --------------
  minimum time: 42.596 ms (21.99% GC)
  median time: 43.553 ms (24.15% GC)
  mean time: 44.995 ms (26.19% GC)
  maximum time: 110.922 ms (69.96% GC)

```

and for `f2` (`flatten`):

```julia
BenchmarkTools.Trial: 
  memory estimate: 65.00 MiB
  allocs estimate: 28
  --------------
  minimum time: 72.419 ms (0.00% GC)
  median time: 74.665 ms (0.00% GC)
  mean time: 86.345 ms (8.86% GC)
  maximum time: 233.150 ms (54.92% GC)
  --------------
  samples: 43
  evals/sample: 1

```

So about double the allocations for `f1` as I expected, but `f1` is surprisingly faster!

---

<div class="post-metadata">

**Author:** ![tkoolen](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tkoolen/32/1603_2.png) [@tkoolen](https://discourse.julialang.org/u/tkoolen)\
**Post date:** [August 24, 2017, 3:19pm UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/12 "2017-08-24T15:19:25Z")

</div>

The surprising results above are because `length` is not defined for the `Flatten` object returned by `flatten`, which means that the size of the result cannot be determined in advance. If I define:

```julia
Base.length(fl::Base.Iterators.Flatten{<:AbstractVector{<:UnitRange}}) = sum(length, fl.it)
Base.iteratorsize(::Base.Iterators.Flatten{<:AbstractVector{<:UnitRange}}) = Base.HasLength()

```

then the benchmark results for the `flatten` approach become

```julia
BenchmarkTools.Trial: 
  memory estimate: 61.00 MiB
  allocs estimate: 5
  --------------
  minimum time: 21.147 ms (1.61% GC)
  median time: 27.357 ms (20.71% GC)
  mean time: 28.169 ms (22.88% GC)
  maximum time: 90.278 ms (76.97% GC)
  --------------
  samples: 178
  evals/sample: 1

```

Maybe a more general version of the methods above could make it into Base? We’re getting a lot of mileage out of this simple question.

(As an aside, how awesome is it that you can just tinker like this in Julia!)

Edit: Julia issue: [https://github.com/JuliaLang/julia/issues/23431](https://github.com/JuliaLang/julia/issues/23431).

---

<div class="post-metadata">

**Author:** ![Martin\_Florek](https://avatars.discourse-cdn.com/v4/letter/m/f4b2a3/32.png) [@Martin\_Florek](https://discourse.julialang.org/u/Martin_Florek)\
**Post date:** [August 25, 2017, 7:04am UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/13 "2017-08-25T07:04:26Z")

</div>

Yes, it is awesome.  
Thanks, I did not even know that [Iteration utilities](https://docs.julialang.org/en/stable/stdlib/iterators/) exist. This is the [new feature](https://github.com/JuliaLang/julia/blob/release-0.6/NEWS.md) of Julia 0.6. In Julia 0.5, `mapreduce` is a good choice in term of speed, but the allocation is a little more than `f1`. For completeness, I add the benchmark for the `mapreduce` in the version 0.6.

```julia
function f3(a, dr)
    mapreduce(x -> a[x], vcat, dr)
end

julia> @benchmark f1($a, $dr)
BenchmarkTools.Trial:
  memory estimate: 122.01 MiB
  allocs estimate: 8
  --------------
  minimum time: 70.672 ms (16.28% GC)
  median time: 72.660 ms (16.37% GC)
  mean time: 76.670 ms (19.92% GC)
  maximum time: 158.054 ms (61.80% GC)
  --------------
  samples: 66
  evals/sample: 1

julia> @benchmark f2($a, $dr)
BenchmarkTools.Trial:
  memory estimate: 65.00 MiB
  allocs estimate: 28
  --------------
  minimum time: 108.974 ms (0.00% GC)
  median time: 113.256 ms (0.00% GC)
  mean time: 123.818 ms (8.49% GC)
  maximum time: 234.858 ms (49.67% GC)
  --------------
  samples: 32
  evals/sample: 1

julia> @benchmark f3($a, $dr)
BenchmarkTools.Trial:
  memory estimate: 122.05 MiB
  allocs estimate: 10
  --------------
  minimum time: 61.970 ms (18.74% GC)
  median time: 63.719 ms (18.74% GC)
  mean time: 66.872 ms (22.33% GC)
  maximum time: 144.829 ms (64.45% GC)
  --------------
  samples: 75
  evals/sample: 1

```

After adding hack.

```julia
julia> @benchmark f2($a, $dr)
BenchmarkTools.Trial:
  memory estimate: 61.00 MiB
  allocs estimate: 5
  --------------
  minimum time: 30.712 ms (0.91% GC)
  median time: 37.693 ms (15.98% GC)
  mean time: 40.333 ms (19.18% GC)
  maximum time: 118.468 ms (71.25% GC)
  --------------
  samples: 124
  evals/sample: 1

```

---

<div class="post-metadata">

**Author:** ![bionicinnovations](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/bionicinnovations/32/6065_2.png) [@bionicinnovations](https://discourse.julialang.org/u/bionicinnovations)\
**Post date:** [January 20, 2023, 8:56am UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/14 "2023-01-20T08:56:14Z")

</div>

Does this help?

```julia
julia> dr = [1:2, 4:7]
2-element Vector{UnitRange{Int64}}:
 1:2
 4:7

julia> vcat(collect.(dr)...)
6-element Vector{Int64}:
 1
 2
 4
 5
 6
 7

```

---

<div class="post-metadata">

**Author:** ![carstenbauer](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/carstenbauer/32/4981_2.png) [@carstenbauer](https://discourse.julialang.org/u/carstenbauer)\
**Post date:** [January 20, 2023, 8:59am UTC](https://discourse.julialang.org/t/how-simple-get-vector-from-ranges/5537/15 "2023-01-20T08:59:38Z")

</div>

Please note that this thread hasn’t seen any activity for 5 years and it’s generally better to avoid reviving such old threads (and implicitly ping all people involved) without good reason.