Choosing only different vectors from a matrix

New to Julia
1

Hello users,
I have a matrix A:

 A=[1.0  7.0  4.0  8.0  9.0  4.0  5.0  0.0  10.0  10.0  1.0  7.0
     1.0   0.0  0.0  9.0   8.0  1.0  3.0  5.0    7.0   1.0  1.0  0.0
     1.0   0.0  0.0  5.0  10.0  8.0  3.0  5.0   1.0   5.0  1.0  0.0];

How can I choose only the columns with different elements?
Is there a command for this?
I this example, the columns from 1 to 10 are different and columns 11 and 12 are repeated.
Thanks a lot.

2

unique(eachcol(A)) will filter for the unique columns, however, if you just want to keep columns which are not repeated, you can count them and then filter, eg:

using DataStructures
C = counter(eachcol(A))
filter(x -> last(x) == 1, C.map)

Note that if you are doing anything numerical, floating point error may make the simple notion of equality implicit in these examples quite useless.

2 Likes
3

I have used this commands but an error appears:

UndefVarError: eachcol not defined
top-level scope at none:0

I am using Julia 1.0.4 and I have installed the package DataStructures.

4

what version of Julia are you using? eachcol was added at in v1.1

1 Like
5

I am with Julia 1.0.4.
Is there a solution for this version?
Thanks so much!

6

you can add OnlineStats and use OnlineStats.eachcol. but never tested it

7

Perhaps adding all those packages may be overkill. How about the simple, albeit verbose solution:

function in_matrix(v, m)
	inMatrix = false;
	(_, nCols) = size(m);
	for ic = 1 : nCols
		if v == @view m[:,ic]
			inMatrix = true;
			break;
		end
	end	
	return inMatrix
end


function drop_duplicates(B)
	@views begin
		(_, nCols) = size(B);
		kept = trues(nCols);
		for ic = 1 : (nCols - 1)
			if in_matrix(B[:,ic], B[:, (ic+1) : nCols])
				kept[ic] = false;
			end
		end
	end
	return B[:, kept]
end


A=[1.0  7.0  4.0  8.0  9.0  4.0  5.0  0.0  10.0  10.0  1.0  7.0
	 1.0   0.0  0.0  9.0   8.0  1.0  3.0  5.0    7.0   1.0  1.0  0.0
	 1.0   0.0  0.0  5.0  10.0  8.0  3.0  5.0   1.0   5.0  1.0  0.0];
 
# Avoid numerical issues with isequal for this example
B = round.(Int, A);

drop_duplicates(B)
8

You can always operate on eg

[A[:, i] for i in axes(A, 2)]

instead, but frankly, I would just upgrade to 1.2 unless you have a compelling reason not to.

9

You could just copy/paste the implementation of eachcol:

eachcol(A) = (view(A, :, i) for i in axes(A, 2))
1 Like