minimum not zero in row

i want to find a way to go from one row to the next one that's indicated by the position of the minimum number excluding 0s until i get to the row i wanted.

Dmat = zeros(7,7);
output = [0 0 0 0 0 0 0];
O=1;
i=O;
T=7;
u=T;
c=1;
output(1)=O;
Dmat(1,1:7)=[0 4 6 5 0 0 0];
Dmat(2,1:7)=[4 0 3 0 7 0 0];
Dmat(3,1:7)=[6 3 0 11 8 0 0];
Dmat(4,1:7)=[5 0 11 0 2 10 12];
Dmat(5,1:7)=[0 7 8 2 0 5 0];
Dmat(6,1:7)=[0 0 0 0 5 0 3];
Dmat(7,1:7)=[0 0 0 12 0 3 0];
while i~=u
  val = min(Dmat(~ismember(Dmat(i,1:7),0)));
  ind = find(val==Dmat(i,1:7));
  Dmat(1:7,i)=0;
  i=ind;
  output(c+1)=i;
  c=c+1;
end

I don't know why the function i¡~ismember is not working for the second iteration. thx.

[dpb]

What is the result intended to be? Isn't clear what "to go to the next one" really means.

However, to find the location of the min() in the row w/o counting zeros, something like

D=D(Dmat(i,:)~=0);               % select the subset
imin=find(Dmat(i,:)==min(D),1);  % find loc in original

Wrote with a temporary for clarity of operations...

Answer 1

[Jan]

The argument

Dmat(~ismember(Dmat(i,1:7),0))

considers the first column only. You want:

Dmat(c, ~ismember(Dmat(i,1:7),0))

or maybe

Dmat(~ismember(Dmat(i,1:7),0), c)

It is not getting clear to me, if you want to operate on rows or columns.

By the way: This is at least confusing:

O=1;
i=O;
T=7;
u=T;
c=1;

ismember is an overkill, if you want to exclude one value only. Easier:

val = min(Dmat(Dmat(i,1:7) ~= 0), c);

Instead of excluding the zeros dynamically, what about this:

Dmat = [0 4 6 5 0 0 0; ...
        4 0 3 0 7 0 0; ...
        6 3 0 11 8 0 0; ...
        5 0 11 0 2 10 12; ...
        0 7 8 2 0 5 0; ...
        0 0 0 0 5 0 3; ...
        0 0 0 12 0 3 0];
Dmat(Dmat == 0) = Inf;

Now you can search the minimum directly:

[~, ind] = min(Dmat(i, :))