Output argument is not assigned on some execution paths for Embedded matlab function

For a single input u to the embedded matlab function using the following code:

 if u>0  
     y=1;  
 end
 if u<0
     y=-1;
 end
 if u==0
     y=0;
 end

gives error: "output argument 'y' is not assigned on some execution paths."

I don't know why the codes above are not working. Instead, I have to use if...elseif...else structure.

Answer 1

[Cedric]

Look at the following:

 >> u = [1 2 3] ;
 >> if u < 2,  fprintf( 'OK\n' ) ;  end
 >> if u > 2,  fprintf( 'OK\n' ) ;  end
 >> if u == 2, fprintf( 'OK\n' ) ;  end
 >> if u == [1 2 3],  fprintf( 'OK\n' ) ;  end
 OK
 >>

EDIT: Also, and the error might come from that..

 >> [NaN < 0, NaN > 0, NaN == 0]
 ans =
     0     0     0

Testing equalty of arrays, or arrays/scalar, requires a little more work usually. Look up:

 >> doc any
 >> doc all
 >> doc isnan
 >> doc isequalwithequalnans

EDIT: in your case with the error, you should probably manage NaN cases. If M-Lint were indicating (as a tip) that the output arg. might not be assigned, I would guess that the array/scalar tests could be a cause.

EDIT: if you know that u is a scalar (this should be tested if you wanted to create a robust function) and not NaN, just build an if u>0, .. elseif u<0, .. else.. end conditional statement. It is better than your series of statements in the sense that it doesn't test conditions from statements below one whose condition was true.

Cheers,

Cedric

[Qingbin]

The code that I wrote works well for .m file. But it doesn't work for embedded matlab function.

You are right in that sense! But I think the series of statements should also be working while they turned out to be invalid for some reason.

[Cedric]

Define an output value for NaN and/or non-scalar cases, or an error() that will tell you in which case you are falling. If I was developing a function (I don't know how that translates to embedded functions), I would do something like:

 if isnan(u),  error('NaN input not allowed.') ;  end
 if ~isscalar(u),  error('Non-scalar inputs not allowed.') ;  end

or

 if isnan(u)
    y = 0 ;       % Set y=0 for NaN inputs (..).
 end

[Qingbin]

I have added the code to test if the input is NAN or not, the error is still there. Actually for my case, the input u is just a scalar (a positive integer). But thanks for the consideration for multiple cases.

[maryam]

I have the same problem. Please share how you solved it finally.

[Qingbin]

FYI, I used if...elseif...end structure instead of if...end...if end structure.

Answer 2

[Sean de Wolski]

How about:

y = sign(u)

?

[Qingbin]

I am trying to test the assignment structure for the embedded matlab function. So it doesn't necessarily be sign (u).

Answer 3

[Dr Yu Lu]

Have you fixed this? I have similar problem

[Walter Roberson]

You need to cover all of the cases, including taking into account that the input might be a vector or that the matrix might contain nan or inf.

Or, you need to initialize first to a value that should be used if none of the conditions hold.

Or you need to use an if/elseif/else structure ending with else to catch the final possibility that nothing else matched

Or you need to use switch with an Otherwise case.

[Heawa Mehmandoust]

Hi

I have the same problem in Simulink, according to the following message:

Output argument 'segSNR' is not assigned on some execution paths.

Function 'Signal to Noise Ratio/Signal to Noise Ratio' (#114.23.29), line 1, column 24:
"segSNR"

the MATLAB code is:

function [overall_snr, segSNR] = Signal_to_Noise_Ratio(Speech, Enhanced_speech,sample_rate)

% ----------------------------------------------------------------------
% Check the length of the clean and processed speech.  Must be the same.
% ----------------------------------------------------------------------

clean_length      = length(Speech);
Enhanced_length  = length(Enhanced_speech);

if (clean_length ~= Enhanced_length)
  disp('Error: Both Speech Files must be same length.');
  return
end

% ----------------------------------------------------------------------
% Scale both clean speech and processed speech to have same dynamic
% range.  Also remove DC component from each signal
% ----------------------------------------------------------------------

%clean_speech     = clean_speech     - mean(clean_speech);
%processed_speech = processed_speech - mean(processed_speech);

%processed_speech = processed_speech.*(max(abs(clean_speech))/ max(abs(processed_speech)));

overall_snr = 10* log10( sum(Speech.^2)/sum((Speech-Enhanced_speech).^2));

% ----------------------------------------------------------------------
% Global Variables
% ----------------------------------------------------------------------

% sample_rate = 8000;       % default sample rate
% winlength   = 240;       % window length in samples
% skiprate    = 60;       % window skip in samples
winlength   = fix(30*sample_rate/1000); %240;       % window length in samples
skiprate    = floor(winlength/4);       % window skip in samples
MIN_SNR     = -10;       % minimum SNR in dB
MAX_SNR     =  35;       % maximum SNR in dB

% ----------------------------------------------------------------------
% For each frame of input speech, calculate the Segmental SNR
% ----------------------------------------------------------------------

num_frames = clean_length/skiprate-(winlength/skiprate); % number of frames
start      = 1;          % starting sample
window     = 0.5*(1 - cos(2*pi*(1:winlength)'/(winlength+1)));

for frame_count = 1: num_frames
segmental_snr = zeros(1,num_frames);
   % ----------------------------------------------------------
   % (1) Get the Frames for the test and reference speech. 
   %     Multiply by Hanning Window.
   % ----------------------------------------------------------

     clean_frame = Speech(start:start+winlength-1);
     processed_frame = Enhanced_speech(start:start+winlength-1);
     clean_frame = clean_frame.*window;
     processed_frame = processed_frame.*window;

     % ----------------------------------------------------------
     % (2) Compute the Segmental SNR
     % ----------------------------------------------------------

     signal_energy = sum(clean_frame.^2);
     noise_energy  = sum((clean_frame-processed_frame).^2);
     segmental_snr(frame_count) = 10*log10(signal_energy/(noise_energy+eps)+eps);
     segmental_snr(frame_count) = max(segmental_snr(frame_count),MIN_SNR);
     segmental_snr(frame_count) = min(segmental_snr(frame_count),MAX_SNR);

     start = start + skiprate;
     segSNR= mean( segmental_snr);
  end

Could you please tell me how I can solve this?

Thanks in advance

[Stephen23]

Cause: if num_frames is less than one then segSNR is never defined.

Solution: put a dummy value at the start of the function, e.g. segSNR = [];

Tip: you should avoid using global variables.

[Walter Roberson]

Your section

if (clean_length ~= Enhanced_length)
  disp('Error: Both Speech Files must be same length.');
  return
end

does not assign to some output variables. You can switch to error() instead of disp/return, or you can switch to assert() instead of disp/return, or you can assign outputs to all of the variables for that case.

Also, your segSNR is not assigned to unless at least one iteration of the for loop is done; the way you calculate num_frames it can come out less than 1.

Question: why are you overwriting all of segSNR on each iteration of the loop?