How to obtain an array of the indices from the original array when using nchoosek

How do I use nchoosek and also efficiently generate an array that has the indices from the original array. For instance,

A = [7, 8, 9]

B = nchoosek(A, 2)

so B = [7, 8; 7, 9; 8, 9]

I want a matrix C that holds the indices in A of all the members of B:

C = [1, 2; 1, 3; 2, 3].

I can generate C using a for loop since we know how nchoosek is going to generate the combinations, but this seems very wasteful precisely because we know how nchoosek is going to generate the combinations, all the information of the indices is already there in our first call of nchoosek. How do I get C efficiently?

Answer 1

[the cyclist]

I have to admit I am not absolutely certain about what you want as input and output. Here is my best guess:

A = [7, 8, 9];
C = nchoosek(1:numel(A),2)

C = 3×2

1 2 1 3 2 3

B = A(C)

B = 3×2

7 8 7 9 8 9

[Kevin H]

Yep, knew it must be something simple. Thanks a bunch!

Answer 2

[Torsten]

You better get B from C, not C from B:

A = [7, 8, 9];
v = 1:numel(A);
C = nchoosek(v,2)

C = 3×2

1 2 1 3 2 3

B = A(C)

B = 3×2

7 8 7 9 8 9

[Kevin H]

Thank you. Same answer as the cyclist and I think he was like a minute before you so I accepted his.