help regarding conditional if statement

How do i write the code for the if statement such that if the condition fails it maintains the previous model (m)

theta = [0:90]';
M = [1800;400;1850;0.2;0.5]';
Dm = ref_coeff(M,theta);
Ds = Dm + (0.02*randn(size(Dm)) + 0);
T = [logspace(3,0,200)]';

Mint = M;
for i = 1 : length(T)
    
  Er = ((abs(Ds - Dm)).^2)./(2*(0.02^2));
  E = sum(Er); % Initial Energy State%
  
  n1 = 25 .* randn(1,1) + 1800; 
  m1 = 1500 + n1*(2000 - 1500); % Perturbation in Model Parameters%
  
  n2 = 50 .* randn(1,1) + 400;
  m2 = 0 + n2*(1000 - 0);
  
  n3 = 50 .* randn(1,1) + 1850;
  m3 = 1200 + n3*(2200 - 1200);
  
  n4 = 0.05 .* randn(1,1) + 0.2;
  m4 = 0 + n4*(1 - 0);
  
  n5 = 0.05 .* randn(1,1) + 0.5;
  m5 = 0 + n5*(1 - 0);
  
  m = [m1;m2;m3;m4;m5]';
  
  dm = ref_coeff(m,theta);
  ds = dm + (0.02*randn(size(dm)) + 0);

  En1 = ((abs(ds - dm)).^2)./(2*(0.02)^2);
  En = sum(En1); % New Energy State%
  
  Ec = En - E;
  
  if Ec >= 0
      
   Mint = m;
   z(i,:) = Mint
        
  elseif (exp(Ec/T(i))) >= randn(0,1)
      Mint = m;

  else 
      ???????
  end
end

[Prince Igweze]

Also im not really sure if my method of storing all iterations of the model(m) is efficently correct. Help

Answer 1

[Jakob B. Nielsen]

At a glance, you could put m_old=m in the very top of your for statement, which will be the model of your most recent iteration. Then, if the bottom condition fails, you simply flip it, m=m_old.

That will of course give a problem on the very first iteration, so put in a condition that when i=1 you skip that comparison.

I dont know if this is what you want?

theta = [0:90]';
M = [1800;400;1850;0.2;0.5]';
Dm = ref_coeff(M,theta);
Ds = Dm + (0.02*randn(size(Dm)) + 0);
T = [logspace(3,0,200)]';
Mint = M;
for i = 1 : length(T)
    if i==1
        %Skip if its the first iteration, where m does not exist yet
    else
        m_old=m;
    end
    
  Er = ((abs(Ds - Dm)).^2)./(2*(0.02^2));
  E = sum(Er); % Initial Energy State%
  
  n1 = 25 .* randn(1,1) + 1800; 
  m1 = 1500 + n1*(2000 - 1500); % Perturbation in Model Parameters%
  
  n2 = 50 .* randn(1,1) + 400;
  m2 = 0 + n2*(1000 - 0);
  
  n3 = 50 .* randn(1,1) + 1850;
  m3 = 1200 + n3*(2200 - 1200);
  
  n4 = 0.05 .* randn(1,1) + 0.2;
  m4 = 0 + n4*(1 - 0);
  
  n5 = 0.05 .* randn(1,1) + 0.5;
  m5 = 0 + n5*(1 - 0);
  
  m = [m1;m2;m3;m4;m5]';
  
  dm = ref_coeff(m,theta);
  ds = dm + (0.02*randn(size(dm)) + 0);
  En1 = ((abs(ds - dm)).^2)./(2*(0.02)^2);
  En = sum(En1); % New Energy State%
  
  Ec = En - E;
  
  if Ec >= 0
      
   Mint = m;
   z(i,:) = Mint
        
  elseif (exp(Ec/T(i))) >= randn(0,1)
      Mint = m;
  else 
      if i==1
          %skip once again
      else
          m=m_old; 
      end
  end
end