yousef - the easiest part of the above problem is to get a matrix of sorted indices that correspond to the data in each row sorted in descending order. Consider using the sort command (type help sort in the command window for details). With it, you can sort as follows:
Z = [2 2 82 10…]; % your 5x12 input matrix
% ignore (~) sorted data and consider the indices (Idcs) only, sorting on all rows of Z but only
% on columns 3 through the end; we sort on the second dimension (2) which is the columns in
% descending order
>>[~,Idcs] = sort(Z(:,3:end),2,'descend');
>> Idcs
Idcs =
8 1 6 7 5 10 9 3 4 2
3 1 6 8 10 4 9 2 5 7
3 4 5 9 10 2 6 8 7 1
2 8 5 1 4 9 10 6 3 7
2 4 3 10 5 1 9 6 7 8
Now you can iterate over each element in the first row of Z, checking to see how many times that element is repeated in that first column (I think that is your requirement). You can cound the number of times an element is repeated by using a combination of the length and find commands (left to you). If not repeated, then do nothing since the row of Idcs is already sorted in descending order by indices. If it is repeated, then consider the second column and check for its repetition using length and find. If not repeated, then apply your first rule (which is a little fuzzy - why consider 1:7 and what is repetition? If it is repeated, then apply your second rule which is a little fuzzy too - are you leaving the first five indices in the current row, and moving the remaining five indices to the subsequent row? (Or that row which has the repeated second column element? Is that why your final answer has only four rows when the input has five?
