How to avoid an Infinite Loop?

Hi,

I'm encountering an infinite loop problem. I tried to debug my code but couldn't figure it out, so I would appreciate your thoughts and ideas.

[nComponents,sizes,members] = networkComponents(NetA);
[nComponentsB,sizesB,membersB] = networkComponents(NetB);

while sizes(1)~=sizesB(1)

    for ii=1:Nodes   
        if nnz(NetA(ii,:))<1 && nnz(NetA(:,ii))<1  && ~ismember(ii,attack) 
             C(ii,:)=0; % All nodes in Layer B coupled to neighboring nodes will get failed
             indab=ismember(inda,ii);
             indab=find(indab);
             NetB(indb(indab),:)=0;
             NetB(:,indb(indab))=0;
        end
    end

% Propagation back to Layer A
    for pp=1:Nodes
        if nnz(NetB(pp,:))<1 && nnz(NetB(:,pp))<1   
           C(:,pp)=0;
           indba=ismember(indb,pp);
           indba=find(indba);
           NetA(inda(indba),:)=0;
           NetA(:,inda(indba))=0;
        end
    end
end

Where,

"NetA" and "NetB" are square symmetric matrices.

"inda" and "indb" are row vectors of the same size.

"attack" and "Nodes" are positive integers.

"sizes(1)" and "sizesB(1)" are positive integers.

[per isakson]

You increases your chances to get a prompt answer if you provide sample data so that we easier can run your code.

Answer 1

[David Goodmanson]

Waseem,

Nothing in the entire while loop alters either 'sizes' or 'sizesB'. So if sizes(1) ~= sizesB(1), that condition will not ever change, leading to an infinite while loop. For the loop to terminate, you will have to change either sizes(1) or sizesB(1) within the loop, in such a way that the two eventually are equal.

[Waseem AL Aqqad]

Thanks for your reply, David!

So do I have to call the function "networkComponents" within the while loop? as sizes(1) and sizesB(1) depend on NetA and NetB which they do change within the loop.

[David Goodmanson]

Hi Waseem,

If it's true that sizes = size(NetA) then it's easy. Just insert that statement into the while loop just after the changes are made to netA, same for sizesB and netB. If sizes has some more complicated relationship to netA then you will have to call up whatever function is going to give you that answer.

ADDED:

Additionally, of course the true-false check will have to pass in one or both of the 'if' loops in order for netA or netB to change.

[Waseem AL Aqqad]

Hi David,

Unfortunately, that didn't work. I will re-debug my whole codes tomorrow morning and let you know.

Thanks!

[Waseem AL Aqqad]

Hi David,

Besides having to pass "networkComponents" in while loop, I should have imposed one constraint to my algorithm to avoid the infinite loop (both networks NetA and NetB should have same average degree).

Say I have two row vectors with the same size but might have different sum of elements as I'm generaing them randomly.

rowA=[3	6	2	4	3	1	1	1	1	2	2	2	3	1	1	1	1	1	1	1]
rowB=[5	5	5	5	6	2	2	1	1	1	2	1	1	1	1	1	1	1	1	1]
%sum(rowA) should equal sum(rowB). so, I want to manipulate the elements of the vector 
% with the bigger sum in order to satisfy this constraint but their sizes should remain 
% the same. 
% Perhaps, the code should replace some of the ones with zeros. As no. 1 will 
% have the most frequent occurance in both vectors in each simulation run.