Find min and max in groups

I have 2 vectors of equal dimension, A and B. To each value of A corresponds a value of B, but equal values of A can correspond to different values of B.

I need to obtain a cell array, "Min", whose cells are vectors with the indexes of the minimum values of B and another cell array, "Max", with the maximum values. The position in each cell array is indicated by the value of A.

For example, if

A = [12, 11, 12, 5, 3, 10, 5]; B = [0.1, 0.1, 0.1, 0.7, 0.4, 0.3, 0.2]

Then:

Min = {0 ,0 ,5 ,0 ,4 ,0 ,0 ,0 ,0 ,6 ,2 ,(1, 3)}; Max = {0 ,0 ,5 ,0 ,4 ,0 ,0 ,0 ,0 ,0 ,0 ,2 ,(1, 3)}

Because of:

a_1 = 12 ; b_1 = 0.1
a_3 = 12 ; b_3 = 0.1
min(0.1, 0,1) = 0.1 => C_12 = (1, 3); max(0.1, 0.1) = 0.1 => D_12 = (1, 3)
a_2 = 11 ; b_2 = 0.1
min(0.1) = 0.1 => C_11 = (2); max(0.1) = 0.1 => D_11 = 2
...

Intiutively, it seems that it could be done in batch using groups.

Would anyone know how to do it? Thanks in advance

Answer 1

[Stephen23]

A = [12, 11, 12, 5, 3, 10, 5];
B = [0.1, 0.1, 0.1, 0.7, 0.4, 0.3, 0.2];
X = 1:numel(A);
Fmax = @(x){x(max(B(x))==B(x))};
Fmin = @(x){x(min(B(x))==B(x))};
Cmax = accumarray(A(:),X(:),[],Fmax,{0});
Cmax{:}

ans = 0

ans = 0

ans = 5

ans = 0

ans = 4

ans = 0

ans = 0

ans = 0

ans = 0

ans = 6

ans = 2

ans = 2x1

1 3

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Cmin = accumarray(A(:),X(:),[],Fmin,{0});
Cmin{:}

ans = 0

ans = 0

ans = 5

ans = 0

ans = 7

ans = 0

ans = 0

ans = 0

ans = 0

ans = 6

ans = 2

ans = 2x1

1 3

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[Sergio Rojas Blanco]

Thank you very much. it works perfectly and elegantly!

You answer everything :D