How to construct a binary matrix reporting 1 in case of equal rows of two arrays of different dimensions?

1 Ansicht (letzte 30 Tage)
Hi, I have a matrix nxk
A=[ 1 2; 3 4; 5 6; 7 8]
and a matrix B mxk (m can be > = or < n)
B=[ 2 3; 4 5; 1 2; 5 6; 10 23; 7 8]
Each row of B is different.
I want to construct a matrix C nxm in which the hi-th element is 1 if A(h,:)=B(i,:), i.e.
C=[0 0 1 0 0 0; 0 0 0 0 0 0; 0 0 0 1 0 0; 0 0 0 0 0 1];
without looping.

Akzeptierte Antwort

Ashish Gudla
Ashish Gudla am 6 Aug. 2014
Bearbeitet: Ashish Gudla am 6 Aug. 2014
You could create the 'C' matrix with all zeros and then, find the positions where its supposed to be '1' and replace it.
To find the positions where there would be '1', you can use "ismember" function with rows( see doc ) to get the lowest index of each row in A that appears in B.
[a1,~] = size(A);
[b1,~] = size(B);
[~,t2]=ismember(A,B,'rows');
t1 = 1:a1;
t1= t1(t2~=0); %ignore zero indices
t2 = t2(t2~=0);
C = zeros(a1,b1);
i = sub2ind(size(C),t1,t2');
C(i) =1;

Weitere Antworten (1)

Chris Turnes
Chris Turnes am 6 Aug. 2014
Bearbeitet: Chris Turnes am 6 Aug. 2014
One way to do this without looping would be to exploit the property that two vectors w and v are equal if and only if
Using this, you could construct each term and compare:
>> IP = A*B.';
>> An = sum(abs(A).^2, 2); % abs only necessary if data is complex
>> Bn = sum(abs(B).^2, 2);
>> C = (IP == An*ones(1, length(Bn))) & (IP == ones(length(An), 1)*Bn.');
This is a way of vectorizing the operation, but it will use much more memory than looping would.

Kategorien

Mehr zu Cell Arrays finden Sie in Help Center und File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by