is there to make this double room to run faster?

Hi guys, I am trying to reduce the cost of running my program: is there a way to reduce two pieces of code: FIRST ONE:

     for i=1:sz(2)
      for j=1:sz(3)
       rms_NTG10(ls,i,j)  = sqrt(mean(err_NTG10(:,i,j).^2));
       rms_NTG20(ls,i,j)  = sqrt(mean(err_NTG20(:,i,j).^2));
       rms_NTG30(ls,i,j)  = sqrt(mean(err_NTG30(:,i,j).^2));
       rms_NTG40(ls,i,j)  = sqrt(mean(err_NTG40(:,i,j).^2));
       rms_NTG50(ls,i,j)  = sqrt(mean(err_NTG50(:,i,j).^2));
       rms_NTG60(ls,i,j)  = sqrt(mean(err_NTG60(:,i,j).^2));
       rms_NTG70(ls,i,j)  = sqrt(mean(err_NTG70(:,i,j).^2));
       rms_NTG80(ls,i,j)  = sqrt(mean(err_NTG80(:,i,j).^2));
       rms_NTG90(ls,i,j)  = sqrt(mean(err_NTG90(:,i,j).^2));
       rms_NTG100(ls,i,j) = sqrt(mean(err_NTG100(:,i,j).^2));
      end
     end

SECOND ONE:

     for i=1:sz(2)
      for j=1:sz(3)
       accum_ar(i,j,accum_index(i,j)) = accum_ar(i,j,accum_index(i,j)) + 1;
      end
     end

Answer 1

[Walter Roberson]

In the first situation, you can run the sqrt() after the rest of the calculations, such as

rms_NTG10(ls,:,:) = sqrt(rms_NTG10(ls,:,:));

Also, you can square the errors before you start:

err_sq_NTG10 = err_NTG10.^2;

and then pull out the part of that you need for each mean ()

Reducing the number of calls to identical operations is usually faster.