triind2sub is a helper function for working with the output of pdist. It's analogous to the way ind2sub works with linear indices of arrays to find the corresponding subscripts. Using triind2sub avoids the time and space required by using the squareform function.
Example:

square = [0 0; 0 1; 1 1; 1 0];
% find the distances between the corners of the unit square
dists = pdist(square);
% report the pairs of points with maximum separation
[col, row] = triind2sub(size(square), find(dists == max(dists)))
% report the pairs of points with minimum separation
[col, row] = triind2sub(size(square), find(dists == min(dists)))

Without triind2sub you could instead use

[col, row] = find(squareform(dists) == max(dists))
[col, row] = find(squareform(dists) == min(dists))

to achieve similar results but this approach is more than ten times slower, produces duplicates and requires more memory.