Uniquetol which preserves the first occurance

I'm using the uniquetol function to filter some data by rows. Here each column gives a weight value with respect to a specific basis function, and therefore each row gives the set of weights related to a specific function:

[~,index] = uniquetol(DATA,2.5e-1,'ByRows',true);

The DATA is sorted beforehand so the 'simplest' function is in row 1, and the most 'complex' function is in row N, with the practical aim being to exclude any 'complex' functions which are within tolerence to a simpler counterpart.

For this purpose, I'm looking for a way to ensure the index values output correspond to the first occurance in the DATA matrix, rather than choosing the 'highest' or 'lowest' occurance.

I searched the file-exchange, but there are a suprisingly large variety of unique-tolerence functions so I may have missed a valid existing solution.

Answer 1

[Bruno Luong]

Use the third inputs followed by accumarray

% Dummy data
n = 100;
A = randi(10,1,n)+1e-12*rand(1,n)

A = 1×100

3.0000 9.0000 6.0000 10.0000 2.0000 7.0000 4.0000 9.0000 3.0000 9.0000 10.0000 3.0000 7.0000 6.0000 9.0000 3.0000 2.0000 6.0000 3.0000 5.0000 9.0000 8.0000 7.0000 3.0000 4.0000 4.0000 1.0000 4.0000 2.0000 4.0000

[~,~,J]=uniquetol(A);
firstidx = accumarray(J,(1:length(J))',[],@min)

firstidx = 10×1

27 5 1 7 20 3 6 22 2 4

% Check
A(firstidx)

ans = 1×10

1.0000 2.0000 3.0000 4.0000 5.0000 6.0000 7.0000 8.0000 9.0000 10.0000

[ADSW121365]

Beautiful solution, thank you!