Reputation: 387
Julia: limited printing of large arrays
I produce a number of large arrays in Julia using a script file. Printing out the whole array is cumbersome but I'd like to check the first few rows make sense.
I know in the REPL there's printing which is limited by the screen size e.g.
julia> zeros(1000,10)
1000×10 Array{Float64,2}:
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
⋮ ⋮
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
But I can't find any print/show function in base Julia which mimics this for scripts, say only printing out the first 10 rows of an array or something like R's head (I would have expected showcompact to do something like this).
Is there an analogous function to R's head in Julia or do I have to write my own.
Upvotes: 14
Views: 6756
Answers (4)
Reputation: 12051
As I mentioned in a comment, the way to do this in v0.5 is to use an IOContext.
A very simple way to limit the data is to pass the :limit => true parameter to the IOContext;
julia> show(IOContext(stdout, :limit => true), v)
[0.147959 0.414018 … 0.595528 0.852563; 0.32679 0.824953 … 0.432143 0.036279; … ; 0.877398 0.661854 … 0.197207 0.15596; 0.0522946 0.508075 … 0.835359 0.705987]
But this still does not print out the way the REPL does; that's because show with two arguments uses a single-line display. To use a multiline display, pass text/plain as the second argument to show (a MIME type):
julia> show(IOContext(stdout, :limit => true), "text/plain", v)
100×100 Array{Float64,2}:
0.147959 0.414018 0.0282934 … 0.816132 0.595528 0.852563
0.32679 0.824953 0.0582351 0.822526 0.432143 0.036279
0.754989 0.724317 0.533966 0.987273 0.931932 0.973622
0.547866 0.282694 0.0295411 0.75929 0.886218 0.0813057
0.0626663 0.111795 0.625083 0.439983 0.562143 0.669046
0.712093 0.469622 0.377298 … 0.298224 0.31853 0.376066
0.774625 0.754328 0.756725 0.61113 0.76566 0.999292
0.917846 0.308363 0.489246 0.715311 0.175302 0.124059
0.310922 0.140575 0.20635 0.0280192 0.683004 0.168129
0.753361 0.755103 0.831806 0.118009 0.122374 0.281476
⋮ ⋱
0.420264 0.7614 0.748408 0.330983 0.0776789 0.309464
0.984379 0.851735 0.595121 0.534982 0.255317 0.743713
0.814505 0.765941 0.71852 0.730677 0.477631 0.0360992
0.910384 0.0747604 0.490685 0.0904559 0.0756424 0.313898
0.628416 0.0790874 0.401488 … 0.523521 0.397249 0.58112
0.578361 0.336352 0.261118 0.838256 0.387374 0.451647
0.66724 0.586342 0.378968 0.602694 0.450686 0.901279
0.877398 0.661854 0.685156 0.658952 0.197207 0.15596
0.0522946 0.508075 0.244423 0.95935 0.835359 0.705987
You can of course change how many rows are shown by passing in :displaysize to the IOContext:
julia> show(IOContext(stdout, :limit => true, :displaysize => (10, 10)), "text/plain", v)
100×100 Array{Float64,2}:
0.147959 … 0.852563
0.32679 0.036279
0.754989 0.973622
⋮ ⋱
0.877398 0.15596
0.0522946 0.705987
Overall, IOContext is very flexible. See its documentation for more details.
Upvotes: 22
Reputation: 21
I believe this thread is out of date. For me on julia v1.4.1 , the display function does not print a limited array, and
a = zeros(1000,6)
show(IOContext(STDOUT, limit=true), "text/plain", a)
yields an error. Instead I found this works
a = zeros(1000,6)
show(IOContext(stdout, :limit => true), "text/plain", a)
Upvotes: 2
Reputation: 7395
We can get the same output as REPL by using display() (is this what you are looking for...?) Also, the "head" and "tail" parts can be printed by using array sections, e.g.,
disp( x ) = ( display(x) ; println() ; println() )
A = diagm( [ i for i=1:100 ] )
disp( A )
disp( A[ 1:5, : ] ) # head
disp( A[ end-4:end, : ] ) # tail
B = [ i for i=1:100 ]
disp( B )
disp( B[ 1:5 ] ) # head
disp( B[ end-4:end ] ) # tail
$ julia test.jl
100x100 Array{Int64,2}:
1 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0
0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 6 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 9 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 10 0 0 0 0 0 0 0 0 0 0 0
⋮ ⋮ ⋮ ⋱ ⋮
0 0 0 0 0 0 0 0 0 0 0 0 92 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 93 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 94 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 95 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 96 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 97 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 98 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 99 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100
5x100 Array{Int64,2}:
1 0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0 0 0 0
0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
5x100 Array{Int64,2}:
0 0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 96 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 97 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 98 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 99 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100
100-element Array{Int64,1}:
1
2
3
4
5
6
7
8
9
10
⋮
92
93
94
95
96
97
98
99
100
5-element Array{Int64,1}:
1
2
3
4
5
5-element Array{Int64,1}:
96
97
98
99
100
Upvotes: 4
Reputation: 2096
You could easily define a head function yourself. For one and two dimensions, this is pretty straightforward:
torange(n::Integer, m) = 1:min(n, m)
torange(c::Colon, m) = (:)
function head(a::AbstractArray{TypeVar(:T), 1}, n = 10)
view(a, torange(n,size(a,1)))
end
function head(a::AbstractArray{TypeVar(:T), 2}, n1 = 10, n2 = 10)
view(a, torange(n1, size(a,1)), torange(n2, size(a,2)))
end
The torange method allows to use a : to return the full length in the corresponding dimension. For example
head(zeros(10, 10), 5, :)
5×10 SubArray{Float64,2,Array{Float64,2},Tuple{UnitRange{Int64},Colon},false}:
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
For more than two dimensions, I have chosen to repeat the last argument beginning from the third:
function head{Na, Nn}(a::AbstractArray{TypeVar(:T), Na}, n1 = 10, n2 = 10, ns::Vararg{TypeVar(:T), Nn} = 2)
nend = last(ns)
view(a, torange(n1, size(a,1)), torange(n2, size(a,2)), (torange(ns[i], size(a,i+2)) for i = 1:Nn)..., (torange(nend, size(a, i)) for i in Nn+3:Na)...)
end
For example:
head(rand(10, 10, 5, 5), 3, 3, 2) # the last two is the default value and can be omitted
3×3×2×2 SubArray{Float64,4,Array{Float64,4},Tuple{UnitRange{Int64},UnitRange{Int64},UnitRange{Int64},UnitRange{Int64}},false}:
[:, :, 1, 1] =
0.384724 0.7328 0.585211
0.738284 0.95145 0.362914
0.43928 0.94307 0.758541
[:, :, 2, 1] =
0.78603 0.588877 0.677201
0.559547 0.800559 0.488433
0.993593 0.691884 0.236595
[:, :, 1, 2] =
0.25732 0.90491 0.323905
0.300924 0.703919 0.813316
0.040522 0.776142 0.624097
[:, :, 2, 2] =
0.746677 0.153574 0.155539
0.991624 0.90167 0.0880094
0.0423263 0.0153597 0.0608328
Note that the n-dimensional version is not type-stable, but that shouldn't matter for printing.
Upvotes: 4
Related Questions
- Julia: How to pretty print an array?
- Julia Vector: Excessive Memory Usage
- Bounds Error in Julia When Working with Arrays
- Instability found in Julia 1.4.1: repeated array constructions cause next "print" or "println" call to break at random
- List of different sized arrays in Julia
- Max Array Length in Julia
- Large arrays in Julia
- A multidimensional array formatted with shell Array output in julia script/print()
- Creating a 1x1 Julia array
- Prevent Julia from writing to console when instantiating large arrays