Different unique rows - matlab

Question

I have a function which finds the ROWS in an array having repeating number and replaces them by the a unique row.
Here is the function:
NVAR= number of parameter in the unique row
PVBu= upper bound of the numbers in the row
lower bound is 1 by default
function [A]=nonrepeatuniquerow(A,NVAR,PVBu)
HM = zeros(1,NVAR);
p = zeros(1,PVBu);
for j = 1
p(1:PVBu) = 1:PVBu;
for k = PVBu:-1:PVBu-NVAR+1
q = ceil(k*rand);
HM(j,PVBu-k+1) = p(q);
p(q:k-1) = p(q+1:k);
end
end
b = any(~diff(sort(A,2),1,2),2) | any(A==0,2);
A(b,:) = repmat(HM,sum(b),1);
end
For example,
Assume [9 1 7] is the unique randomly generated row.
Matrix A is passed to the function 'nonrepeatuniquerow'.
A= [2 3 3
2 5 2
1 5 9
9 7 6]
And output is:
A= [9 1 7
9 1 7
1 5 9
9 7 6]
My question is how to modify the function so that it replaces the ROWS having a repeating number by DIFFERENT unique rows ?
For example the output could be as follows:
A= [7 2 3
2 1 8
1 5 9
9 7 6]
The function below only generates unique rows every time it is called.
NVAR=number of parameter in the unique row
PVBu= upper bound of the numbers in the row
lower bound is 1 by default
function [HM]=generateunique(NVAR,PVBu)
HM = zeros(1,NVAR);
p = zeros(1,PVBu);
for j = 1
p(1:PVBu) = 1:PVBu;
for k = PVBu:-1:PVBu-NVAR+1
q = ceil(k*rand);
HM(j,PVBu-k+1) = p(q);
p(q:k-1) = p(q+1:k);
end
end
end
I would greatly appreciate any help you can give me in working this problem?

Answer 1

In the function nonrepeatuniquerow, the first part is just for creating the random sequence HM. You could run through all non-unique rows in a for loop and create a separate HM for each. That would look like this:
function [A]=nonrepeatuniquerow(A,NVAR,PVBu)
b = any(~diff(sort(A,2),1,2),2) | any(A==0,2);
for ii=find(b==1).'
HM = zeros(1,NVAR);
p = zeros(1,PVBu);
for j = 1
p(1:PVBu) = 1:PVBu;
for k = PVBu:-1:PVBu-NVAR+1
q = ceil(k*rand);
HM(j,PVBu-k+1) = p(q);
p(q:k-1) = p(q+1:k);
end
end
A(ii,:) = HM;
end
end
You could also replace the generateunique function or the part which does the same in your code by the built-in randperm function. You can create a unique random sequence by
HM = randperm(PVBu,NVAR);
that would give you the following code
function A = nonrepeatuniquerow(A,NVAR,PVBu)
b = any(~diff(sort(A,2),1,2),2) | any(A==0,2);
for ii=find(b==1).'
A(ii,:) = randperm(PVBu,NVAR);
end
end