Best way to deal with index-data pairs with multiple data values per index

Hello,

I have a large vector of index-data pairs that may contain multiple data values for any given index. The way this is represented in the data is that the index vector may contain repeated indices with the corresponding location in the data vector containing the multiple data values. For example

ind =  [ 1,  2,  4,  2, 5, 6,  4,  4,  9];
data = [32, 14, 36, 45, 3, 1, 78, 90, 69];

In this case index 2 has data values of 14 and 45.

Now, what I want to do is to assign these data values into a matrix using the index values, and if multiple data values exist for a given index I want to take the maximum. For example, if ind contains indices for a 3x3 matrix I want the following (working off of the previous example)

mat = zeros(3);
mat = input_data(ind,data,mat)

OUTPUT:

mat = [32,90,0;
       45, 3,0;
        0, 1,69]

where, as you can see, the value for index 2 is 45 since 45 is greater than 36.

What is the best way to do this quickly? My vectors are very long so I don't want to use a loop because it will decrease performance too much. I suspect I will probably need to use the unique command (and possibly sort) somehow but I can't figure out how to implement it.

Any help is greatly appreciated.

Thanks,

Andrew

Answer 1

[Azzi Abdelmalek]

ind = [1,2,4,2,5,6,4,4,9];
data = [32,14,36,45,3,1,78,90,69]
out=reshape(accumarray(ind',data',[],@(x) max(x)),3,3)

Answer 2

[Guillaume]

accumarray is perfect for this:

ind =  [ 1,  2,  4,  2, 5, 6,  4,  4,  9];
data = [32, 14, 36, 45, 3, 1, 78, 90, 69];

m = accumarray(ind', data', [], @max)

You can reshape m afterward (or possibly pass the size to accumarray)

Answer 3

[Walter Roberson]

mat = zeros(3);
maxdata = accumarray(ind(:), data(:), [numel(mat), 1], @max);
mat = reshape(maxdata, size(mat));