# How to accelerate the process of `find` in large for loop ?

3 ビュー (過去 30 日間)
wei zhang 2020 年 7 月 22 日
コメント済み: wei zhang 2020 年 7 月 23 日
I have two arrays A,B. I want to find the index of every element of B in A. I'm sure that every element of B must appears in A twice. The code of mini case is as below.
A = [1 2 3 4 5 6 5 4 2 7];
B = [2 4 5];
result = zeros(length(B),2);
for i=1:length(B)
result(i,:) = find(A==B(i));
end
% result should be [2,9;4,8;5,7]
The A and B in practice is very large (8000000). It takes long time in this for loop way. Is there any way to improve the code?
I had tried to use ismember function. But it seems I could not bypass the find function.
[ia,ib] = ismember(A,B);
result = zeros(length(B),2);
for i=1:length(B)
result(i,:) = find(ib==i);
end
I know I could use parallel computing toolbox. I just want a more delicated coding way.

サインインしてコメントする。

### 採用された回答

Stephen 2020 年 7 月 22 日
>> [X,Y] = ismember(A,B);
>> [~,Z] = sort(Y(X));
>> V = find(X);
>> R = reshape(V(Z),2,[]).'
R =
2 9
4 8
5 7
##### 1 件のコメント表示非表示 なし
wei zhang 2020 年 7 月 23 日
How could you understand the appilication of these functions, sort ismember reshape unique, so well?! Your answer is so good! Even I know the meaning of the above functions, I still can't use them properly, for several years!

サインインしてコメントする。

### その他の回答 (1 件)

Mehmed Saad 2020 年 7 月 22 日
Try the following code
A = [1 2 3 4 5 6 5 4 2 7];
B = [2 4 5];
[r,c] = find(A-B'==0);
[~,ic] = sort(A(c));
c(ic)
##### 1 件のコメント表示非表示 なし
wei zhang 2020 年 7 月 23 日
Because my size of A and B are very large, the matrix A-B' is out of memory very severely, unfortunately.

サインインしてコメントする。

R2019a

### Community Treasure Hunt

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

Start Hunting!