I have a problem with my code, it works fine but the results it isn't correct because the code is not compatible with the equations .. description below

clc;
clear;

M = 60;
N = 36; 
K = 5; 

 x = rand(1,N);
 theta = (2*pi)*x; % generated randomly
 Phi = diag(exp(1i*theta)); %NxN 
          
 H_hermitian = randn(N,M)+i*randn(N,M); %NxM Matrix ......... (h_n)hermitian = 1xM which represnt each row of H_hermitian
 G = randn(K,N)+i*randn(K,N); %KxN ......... g = 1xN and which represnt each row of G .. g* Nx1
 H_dk = randn(K,M)+i*randn(K,M); %KxM ......... h_dk = 1xM which represnt each row of H_dk
            
              
 HT = H_dk+G*Phi*H;  % KxM .. this matrix needs to be updated after computing the new theta by updating phi.
 W = pinv(HT);  % MxK ......... w = Mx1 .. Updateing HT means updating W and that is what I need

[Jan]

Please post the code in a readable form. Remove the pile of blank lines and use the code formatting tools. Mention, why you assume, that the code is not correct. you have this valuable information already. So sharing it with the readers is the best choice. Make it as easy as possible to answer your question.

[Matthew Worker]

I did what you asked for.

I think the results of the code is not correct, because from what I can see the code is not compatible with the equations in picture 1 but I do not know how to change it.

[Jan]

If you elaborate the message "the code is not compatible with the equations", that chances to get an answer will grow. You do see some relevant information already, so why do you hesitate to share them with us?

[Matthew Worker]

Sir, the only thing that I know that this code is not correct because it is part of a bigger and code I can not write the whole code here, so because the rest of the code works just fine and gave me a good result. I know the wrong is in this part because when I added it to my code I got results that do not make sense but I do not know where exactly that is why I am asking

[Jan]

I do not see a chance to help you. We see some code and you explain, that the results do not make sense. This is not enough information to identify, what you consider as problem and therefore we cannot suggest a solution.

[Image Analyst]

original question attached (because he will delete it).

[Rena Berman]

(Answers Dev) Restored edit

Answer 1

[John D'Errico]

This is one of those questions that tend to languish unanswered on the site. Why? Because you just say that your code does not work, does not produce the correct results. So we see a complicated code, lacking much in the way of comments about what you did. Worse, those comments seem to make little sense. For example, I see:

         Alfa_i(t) = 0;
         for t2=1:K
             Alfa_i(t) = Alfa_i(t) + 1/(abs(H_dk(t2,:)*W(:,t2)));
         end
             Alfa_i(t) = Alfa_i(t) - Alfa_k(t);  % Sum of Alfa_i
       end

There is clearly a sum happening in there, but exctly what is being summed is not Alpha_i. At best, Alpha_i is the result of a sum.

And that was one of the only comments in your entire code. Oh. One other, where you tell the reader the completely obvious:

       G = G(idx,:);  % sort the whole matrix using the sort idx

I could compare your code to one I was given to maintian. In some 300 lines of code, there was only one comment. In a code block, where a matrix called C was clearly constructed, the ONLY comment was: "Create C matrix here". There was no information offered as to what C meant, or what will be done with it. Your code is almost analogous. And you want us to read that code and debug it, knowing only that it possibly does not work properly for some ungiven reason?

Then we must guess what it does wrong. Yes, it is supposed to follow the given pseudo-code. But then we must dive deeply into code written by you, to figure out where the problem is, when nobody has even said what it does wrong. And that is going to take some serious luck, for someone to wander by who has many hours to waste for probably no success.

If you want help, then I would first explain CLEARLY what answer you think it should produce. Then explain what it does produce, and why you think it is a problem. Remember, that if you go to a doctor and tell them only that your leg does not hurt, will they be able to know what is the reason you went to see them, and perhaps how to help you?

Better yet is to help yourself. Extract this code from what contains it. Use the code, by providing a simple, sample problem where you will know the answer that JUST this code fragment should perform correctly. Does it work? If not, then you can lean why. This, by the way, is why modular code is valuable. You can test every piece. NEVER throw together a huge mess of code and expect it all to work the first time. Write code in SMALL fragment. Test EVERY such code fragment. Only when they all work as you know they should do you build the larger code.

Next, learn to use the debugger! Trace through the code. Verify every line of code. Does it do what you expect it to do on the sample subproblem I suggested you formulate?

So I'm sorry that this is not really an answer, because I think no answer is really possible. You are the best person to solve this problem. You have the most incentive to want to solve it, to spend what may be a few hours tracking down the problem. I recall a saying from my misspent youth: "God helps those who help themselves." At the very least, make it possible for someone to help you. Good luck.

[Matthew Worker]

sir thank you for your response.

I think I wrote all the matrices which needed to compute these equations.

I just don't know how to do these for loops.

I tried to do it too many times in different ways but I think my code is not compatible with for loops in the picture.

I wrote a brief description in front of each line and I deleted the rest of the code