How to include complex numbers in fprintf function?

lambda = [1.064e-6];
R = [30];
w=[0.001]; 
q = (1./R- i* lambda./pi./w.^2).^(-1);
a=1;
p=1; 
m=1;
probe_r=linspace(0,0.003,100);
probe_theta=linspace(0,0.003,100);
rseed=[0*max(w):max(w)/30:3*max(w)];  
thetaseed=[0:360]*pi/180;
[r,theta]=meshgrid(rseed,thetaseed);
E=LaguerreGaussianE([p,m,q,lambda,a],r,theta);
V=interp2(r,theta,E,probe_r,probe_theta);

column_names = {'r', 'theta', 'V'};
fid = fopen('fidtext.txt','wt');
fprintf(fid, '%s ', column_names{:});
fprintf(fid, '\n');
block_of_data = [probe_r, probe_theta, V];
fmt = repmat('%15g ', 1, 3);
fmt(end:end+1) = '\n';
fprintf(fid, fmt, block_of_data.'); 
fclose(fid);

With the current code I have I get a .txt file of only the real numbers from my function V along with the values of probe_r and probe_theta. How do I alter this to produce both the real and complex numbers as a 3 column .txt file of r, theta and V as I am unable to see a formatSpec to include complex numbers.

Answer 1

[Star Strider]

you have to write the real and complex parts separately.

Example —

x = sqrt(-2);
fprintf(fid, '%f%+fj\n', real(x), imag(x))

0.000000+1.414214j

[Robbie McCormack]

Thanks for the quick response! Can you see where I need to enter something similar to your example in my code as I am only needing to produce the complex numbers for V? When I enter something similar to your example I'm still not receiving complex numbers.

[Star Strider]

My pleasure.

I cannot run your code. I have no idea what ‘V’ is, or what ‘block_of_data’ is.

If ‘V’ is a vector, this works:

fid = 1;
x = sqrt(randi([-5 5], 5, 1));
fprintf(fid, '%f%+fj\n', [real(x), imag(x)].')

In all likelihood, if ‘V’ is complex, ‘block_of_data’ will also be complex (although the imaginary parts of the real numbers will be 0), considerably complicating your desire to use otherwise efficient matrix arguments to fprintf.

A for loop is likely your best option:

fid = 1;
probe_r = randi(9, 5, 1);
probe_theta = randi(9, 5, 1)
V =  sqrt(randi([-5 5], 5, 1));
for k1 = 1:size(probe_r,1)
    fprintf(fid, '%.1f %.1f %f%+fj\n', probe_r(k1), probe_theta(k1), real(V(k1)), imag(V(k1)))
end

It is not inefficient if it is the only way to do what you want.

[Robbie McCormack]

Thanks a lot for your answer, much appreciated. Unfortunately, I think I'm a little out my depth. I am getting closer but instead of showing the complex number that V gives the code produces the complex number as the next real number in the array. I have no doubt its me thats making the mistakes and not yourself! Apologies for taking up your time.

column_names = {'r', 'theta', 'V'};
fid = fopen('fidtext.txt','wt');
block_of_data = [probe_r, probe_theta]
fmt = repmat('%15g ', 1, 3);
fmt(end:end+1) = '\n';
fprintf(fid, '%f%+fi\n', [real(V), imag(V)].');
fprintf(fid, fmt, block_of_data.');
fclose(fid);

[Star Strider]

When I simulated ‘block_of_data’ as two columns of random real numbers and a complex vector ‘V’, ‘block_of_data’ became a complex matrix. That is the reason I used the for loop with the individual columns of the matrix.

If you separate out the real and imaginary parts of ‘V’, this works:

block_of_data = [probe_r probe_theta real(V) imag(V)];
fprintf(fid, '%.1f %.1f %.1f%+.1fj\n', block_of_data')

Change the format string to the one you want.

[Walter Roberson]

Try

fprintf(fid, '%f%+fi\n', [real(V(:)), imag(V(:))].');

[Robbie McCormack]

You've only gone and done it!

Thank you both for your help, it is hugely appreciated.

[Star Strider]

Our pleasure!

[Walter Roberson]

@NAVNEET

V = [-1 - 1i
     -1 + 0i
     -1 + 1i
     0 - 1i
     0 - 0i
     0 + 1i
     1 - 1i
     1 + 0i
     1 + 1i];

fid = 1;
fprintf(fid, '%f%+fi\n', [real(V(:)), imag(V(:))].');

-1.000000-1.000000i
-1.000000+0.000000i
-1.000000+1.000000i
0.000000-1.000000i
0.000000+0.000000i
0.000000+1.000000i
1.000000-1.000000i
1.000000+0.000000i
1.000000+1.000000i

You can see that negative imaginary parts are automatically handled.

The key here is the %+ specification, which instructs that the appropriate sign of the value is to be inserted.

Answer 2

[Ken]

This works fine for a single complex number. If you replace x with an array, the reals come first followed by all the imaginaries.

For example:

rxSignal=[1.1+1j*2.2 3.3+1j*4.4 5.5+1j*6.6]
fileID=fopen('rxSignal.txt','w')
fprintf(fileID,'%f%+fj\n',real(rxSignal(:)),imag(rxSignal(:)));
fclose(fileID)

When you read the text file, you get:

1.100000+3.300000j

5.500000+2.200000j

4.400000+6.600000j

Here you see the reals filling the first three slots and the imaginaries filling the last three slots.

How to fix this without writing a for loop to do one at a time?

[Walter Roberson]

Yes, but I posted the solution to that over two years ago; https://www.mathworks.com/matlabcentral/answers/369740-how-to-include-complex-numbers-in-fprintf-function#comment_510120

[Ken]

But I am using your code example and getting the wrong result. Did the Matlab version change this? Please run my snippet on your machine and view text file. Thanks.

[Star Strider]

@Ken Crandall —

‘But I am using your code example ...’

No, you’re not. Note the concatenation of the two vectors in a matrix, and the transpose operator (.') in Walter’s code, however not in yours.

[Walter Roberson]

The concatenation and transpose are very important for this purpose .

[Star Strider]

Ken Crandall’s Answer moved here —

I apologize. Thanks!

[Star Strider]

No apology necessary!

It’s just that it’s important to read the code carefully and understand how it works.

[Paul Serna-Torre]

Thank you. It worked.