Why do i keep getting the ereor "The value assigned to variable *** might be unused"

function strain_value= acquirestrain(uncomp_window,comp_window,fs)
  NFFT = length(rcv_uncomp); %Length of the RF signal

  %% PSD of Uncompressed Signal
  z_uncomp=fft(uncomp_window); 
  PSD_uncomp= (abs(z_uncomp).^2)/(NFFT*fs); %Transforming the FFT to PSD by multiplying with the length and sample frequency
 
  %Single Sided Spectrum
  
  PSD__uncomp_single= PSD_uncomp(NFFT/2+1:end); %Get the real half of the signal (ERROR occurs here)
  PSD_uncomp_single(2:end-1)=2*PSD_uncomp_single(2:end-1); %And double it to account for the imaginary half (ERROR occurs here)

  %% PSD of Compressed Signal
  z_comp=fft(comp_window); 
  PSD_comp= (abs(z_comp).^2)/(NFFT*fs); 

  %Single Sided Spectrum
  
  PSD_comp_single= PSD_comp(NFFT/2+1:end); 
  PSD_comp_single(2:end-1)=2*PSD_comp_single(2:end-1);

  % Frequecy parameters
  
  frequency_axis= fs*(0:(pizo_elements/2)-2)/pizo_elements;  
  fc1= find(PSD_uncomp,max(PSD_uncomp));
  fc2= find(PSD_comp,max(PSD_comp));
  
  fc1=frequency_axis(fc1);
  fc2=frequency_axis(fc2);
  strain_value= (fc2-fc1)/fc1;
end

Answer 1

[dpb]

The warning is given because your code never references/uses the frequency vector, f, after defining it -- there's nothing in the remaining code that makes any use of it; you could just as well not include the line that computes it.

Most of the time when someone does compute a PSD it is desired to plot it or determine the actual frequency of a spectral feature so the frequency vector values would be needed -- unless you do something more in your case, it actually isn't needed by the code as is.

We don't know what frequencyaxis is in the above code...

[Karim Aawar]

Hello!

thanks for the clarification. so i made a small mistake with naming the variables but i fixed it now so the error doesnt occur anymore at frequency_axis. However what iu dont get is when i try computing PSD__uncomp_single the first line is fine because the variable is being used but when i try to get half of the spectrum (here: PSD_uncomp_single(2:end-1)) the error ocurrs.

Any idea why?

Thanks in advance!

[Karim Aawar]

Same happens with PSD_comp_single(2:end-1)....

[dpb]

...  
PSD_comp_single= PSD_comp(NFFT/2+1:end); 
PSD_comp_single(2:end-1)=2*PSD_comp_single(2:end-1);
% Frequecy parameters

frequency_axis= fs*(0:(pizo_elements/2)-2)/pizo_elements;  
fc1= find(PSD_uncomp,max(PSD_uncomp));
fc2= find(PSD_comp,max(PSD_comp));
...

Same thing -- you don't use PSD_comp_single after you compute it; only the two other PSD_xxx variables.

I was coming back to amplify my earlier response on that point -

Just as a coding stylistic note; I think I would have kept the variable f you had originally and changed the reference or frequency_axis to f instead; it makes more sense to me that way and f for the PSD frequency is probably the most common nomenclature used.

So, I'd've probably written something like

  f = fs*(0:(pizo_elements/2)-2)/pizo_elements;
  [mxP_un,ic1]=max(PSD_uncomp_single);
  [mxP_c,ic2]=max(PSD_comp_single);
  fc1= f(ic1);
  fc2= f(ic2);
  strain_value= (fc2-fc1)/fc1;

letting max return the optional output of the index instead of using find and using the single-sided spectra to match up with the f vector