Using FIND for common elements WITHOUT a loop (Vectorized?)

Hello,

I have an array A 100x1 with different datenum that

I need to find in a big array B 5000000x1 (all datenum in B are in A, just repeated many times)

and then replace w another datenum C 100x1 (same index as A)

I am using find( ) but it is taking too long. Is there a way to do this without a loop?

BB=B %making a copy of B to check after.

for i=1:size(A,1)

loc=find(B==A(i))
BB(loc)=C(i);

end

[Star Strider]

The unique function comes quickly to mind.

The third output from unique may be what you want.

[Dave]

thank you, I don't know how to take your comment as the accepted answer.

Answer 1

[Stephen23]

Use ismember and its second output argument:

>> A = randperm(5)
A =
   3   5   4   1   2
>> B = randi(5,1,9)
B =
   5   1   2   5   1   2   4   5   4
>> C = (1:5).^2
C =
    1    4    9   16   25
>> [~,X] = ismember(B,A);
>> BB = C(X)
BB =
    4   16   25    4   16   25    9    4    9

[Dave]

Thanks, is there a way to use that for a matrix instead?

%A is 4x1
A =[1, 6, 5, 5]'

%B is much larger, 6x1
B =[5 5 1 1 1  6 ]'

%A2 is 
A2=[101, 501, 601;
102, 502, 602;
103, 503, 603]

% BBB is same dim as B, but using elements from A2
%ANSWER SHOULD BE:
BBB=[501 502 101 102 103 601]'

[Stephen23]

"Thanks, is there a way to use that for a matrix instead?"

Probably. Please explain the logic of your new example (which is unrelated to your earlier question when C had the same size as A).

[Omar Salah]

Hi Stephen, I really wanted to thank you. I wanted to imporve my really slow code in which I tryz to find a 100k elements in a 320k elements vector. using is memeber the code runs in 5.4832 seconds instead of 1170 seconds when using for and parfor. Vectorization is magical for Big data.