How can I design these nested for loops so they run much faster?

Hello everyone, in my matlab script I have taken two sets of data from excel and imported them as arrays in matlab. Now, I want to eliminate all of the rows in in the second larger array that do not have the same number as any of the elements in the second column of the first array.

Doesn't make sense? Let me give you an example.

Take for example you have just imported two matrices A and B that look something like this:

    A= 26   100   1991                    B= 45 125  1984  
       26   101   1991                       45 558  1984
       27   185   1992                       45 778  1984
       27   181   1992                       46 181  1995
       28   56    1981                       47 101  1996

Then I would want to create a new matrix by sorting through A to determine which rows of B should be eliminated. When the operation is complete I should end up with a new matrix for B that is much smaller:

    new_B= 46 181 1995
           47 101 1996

This is the script I have written that does this operation:

    k=1;
    for n=1:length(A)
         x=A(n,2);
     for i=1:length(B)
         if B(i,2)== x
            new_B(k,1)=B(i,1);
            new_B(k,2)=B(i,2);
            new_B(k,3)=B(i,3);
            k=k+1;
          end
     end
  end

This turns out to be very slow when the length of A = 100,000 and the length of B= 700,000. Could someone help me optimize this piece of code? Thank you for the help :)

Answer 1

[Cedric]

 B_new = B(ismember(B(:,2), A(:,2)),:) ;

should take a fraction of second with the sizes that you mention.

Answer 2

[Yao Li]

You need to design a new algorithm. Why not pre-sort the 2nd column of A and B before the comparison? Or calculate the min and max value of the 2nd column of A and B, and remove some rows in advance.