could anyone help me to solve the following issue.

I am having two different values of centroids generated randomly by the code to be 3 and 5 for 6 users.

As a result i am getting

centroids =3 
centroids =5

in my command window

Now i need to begin my iteration so i used for loop in my code:

for iteration=1:iterations  %iterations=3;
distances=zeros(users,centroids,1) %users=6
.
.
end.

But the code executes by taking into consideration only the centroids=5 in for loop and it does not consider centroids=3.

But for me the for loop should execute first by taking centroids=3,then followed by centroids=5 when iterations begins.

could anyone please help me on this.

Answer 1

[KSSV]

for iteration=1:iterations  %iterations=3;
    if i == 1
        centroids = 3 ;
    else
        centroids = 6 ;
    end
distances=zeros(users,centroids,1) %users=6
.
.
end.

[jaah navi]

ok.But the centroids not known before.

Could you please help with the following code:

iterations=2;
particles=2;
PPP=[1,2;
    3,4;
    5,6;
    7,8]
for particle=1:particles
    P=numel(particle);
    users=size(PPP,1);
    centroids= randi(users,1,P);
end
for T=1:iterations 
    for particle=1:particles
        distances=zeros(users,centroids,P)
    end
end

In this code the first for loop generates the random number of centroids with respect to two particles.

this was the result from the command window

centroids =     4
centroids =     3

then

distances =     0     0     0
              0     0     0
     0     0     0
     0     0     0
distances =     0     0     0
     0     0     0
     0     0     0
     0     0     0
distances =     0     0     0
     0     0     0
     0     0     0
     0     0     0
distances =     0     0     0
     0     0     0
     0     0     0
     0     0     0

But it consider only the last value of centroids which is 3 and doesnt consider 4.

Could you please help me on this.

[KSSV]

Some thing like this:

iterations=2;
particles=2;
PPP=[1,2;
    3,4;
    5,6;
    7,8]
centroids = zeros(1,particles) ; 
for particle=1:particles
    P=numel(particle);
    users=size(PPP,1);
    centroids(particle)= randi(users,1,P);
end
for T=1:iterations
    if T == 1
        centroid = centroids(1) ;
    else
        centroid = centroids(2) ;
    end
    for particle=1:particles
        distances=zeros(users,centroid,P)
    end
end

[jaah navi]

I tried with the following code :

iterations=1;
particles=2;
dimensions=2;
PPP=[1,2;
    3,4;
    5,6;
    7,8];
centroids=zeros(1,particles);
for particle=1:particles
    P=numel(particle);
    users=size(PPP,1);
    centroids(particle)= randi(users,1,P);
    fprintf('\nCentroids=%d\n',centroids(particle))
    swarm_pos = rand(centroids(particle),dimensions,P)
end

for T=1:iterations 
    for particle=1:particles
         P=numel(particle);
          distances=zeros(users,centroids(particle),P)
       for centroid=1:centroids(particle)
               distance=zeros(users,1);
            for data_vector=1:users
                distance(data_vector,1)=norm(swarm_pos(centroid,:,particle)-PPP(data_vector,:))
            end
             distances(:,centroid,particle)=distance
       end
    end
end

but i am getting error .Could you please help me to overcome it.

[Walter Roberson]

for particle=1:particles
  P=numel(particle)

particle is one specific value from the range 1:particles, so it is a scalar. numel() of a scalar is 1. So P is only ever 1.

  centroids(particle) = randi(users,1,P)

would fail if P > 1 because it would be too many values to store at the scalar location centroids(particle), so you expect P to be 1.

  swarm_pos = rand(centroids(particle),dimensions,P)

P is 1, so you are generating a 3D array with 3rd dimension 1, which is the same as a 2D array.

Later,

  for particle=1:particles
      norm(swarm_pos(centroid,:,particle))
 

particle can exceed 1, and when it does you try to access swarm_pos(centroid,:,2) but we proved above that swarm_pos never has more than 2 dimensions.

[jaah navi]

ok.could you please help me how the error can be overcomed

[jaah navi]

I done few modifications with respect to the code:

iterations=2;
particles=2;
dimensions=2;
PPP=[1,2;
    3,4;
    5,6;
    7,8];
centroids=zeros(1,particles);
swarm_pos=[];
sizeS=[];
for particle=1:particles
    P=numel(particle);
    users=size(PPP,1);
    centroids(particle)= randi(users,1,P);
    fprintf('\nCentroids=%d\n',centroids(particle))
    temp1=rand(centroids(particle),dimensions,P);
    swarm_pos =[swarm_pos; temp1];
    sizeS=[sizeS; size(temp1)];
end

for T=1:iterations 
    iteration=T;
for particle=1:particles
    P=numel(particle);
    distances=zeros(users,centroids(particle),P)
           for centroid=1:centroids(particle)
                distance=zeros(users,1);
                for data_vector=1:users
                VV=data_vector;
                SS=swarm_pos(centroid,:,P)
                MM=PPP(data_vector,:);
                NN=(SS-MM);
                NNN=norm(SS-MM);
                distance(data_vector,1)=norm(swarm_pos(centroid,:,P)-PPP(data_vector,:))
                end
                distances(:,centroid,P)=distance
          end
end
end

The code executes but the result was wrong.

Could you please help me on this.

[Walter Roberson]

I spent a while on the code, trying to figure out what the heck you could possibly mean.

The best I was able to come up with is that you have a set of "users" with associated locations, and that you have a number of entities you call "particles", and that for each of those particle entities, you have a "swarm", with the number of members in the swarm to be chosen at random based upon the total number of users you have. (Random number of members in a swarm could be justified for some swarm dynamics, but why the number of members of the swarm should depend upon the number of users is a complete mystery.) Anyhow, each member of the swarm has a current position.

For each iteration, you go through each particle, and for each swarm member for the particle, you compute the distance from each swarm member to each of the users and record that.

But the distance from the swarm member to the users does not affect anything, it just gets computed, so if you had more than one iteration, it would just calculate the same thing again. In particular, the swarm members do not move positions, and no "cost" is calculated, so there is no "goodness" available for a particular swarm position or swarm and so there is no way to evaluate whether one particle is "better" than another particle.

Anyhow, I did come up with code to implement the above... but it does seem a strange thing to do.

If you had a hypothesis that a varying number of "servers" should be evaluated in parallel to optimimally fill the needs of the users, each server having a position, and there being a formula to calculate how good the service was: if that were the case, then instead of allocating a random number of swarm members, it would make more sense to systematically allocate 1 through the maximum number of servers, in which case instead of "particles" you would talk about "replications".

[jaah navi]

But each swarm_pos are different with respect to particles.

As i am using two particles two different swarm_pos are generated,following by computing the distance.

but in my code the distance is being computed with respect to sceond particle and not with respect to first particle.

I am still working on it.

If possibe help me on this.

[Walter Roberson]

Your existing code is too broken for a meaningful correction to be made by us. Too many places where we have no idea why you would even consider doing things the way you are doing them.

That is why I spent a bunch of time trying to figure out what the intended purpose of the code was -- not what the problem was that had to be solved, but rather trying to figure out what it was that you thought you had to implement along the way to solving what-ever it is you are trying to solve. I came up with a self-consistent framework for what you might have been trying to do, and I implemented it.

If my description of the goal of your code (not the current operation of your code but what the aim of the code is) is not correct, then I am not able to assist you, as your code as posted has no consistent logic in the form it is in, and cannot be repaired except by re-thinking what the purpose of the code is and going back to start over.