fprintf and fopen changing the data I'm trying to write

I'm trying to save the result of calculations as an ASCII using fopen and fprintf, but my data becomes corrupted. Snippet below, where the variables are vectors calculated above inside a for loop:

fid01=fopen('Specific_Momentum_and_KE.txt', 'a+'); 
        fprintf(fid01, '%.4f, %.4f, %.4f, %.4f\r\n',tstep, xfland, mag_land_vel, mag_land_KE); 

fclose(fid01);
end

I'm getting seemingly random strings of integers in my file, and the numbers which are supposed to be there are wrong--they don't match the output from my Matlab script. Example from the ASCII:

48.0000, 48.0000, 48.0000, 48.0000
0.5053, 0.5862, 0.5963, 0.5923
0.5722, 0.6441, 0.6507, 0.5837
0.6534, 0.6464, 0.6737, 0.6606
0.6153, 0.6864, 0.6746, 0.6344
0.6978, 0.7149, 0.7490, 0.6757

Where the heck did those 48's come from?? Also, the second column should be

0.5052
0.5862
0.5962
0.5922
0.5721

but it's like Matlab is changing my data! Very bad deal. Please help.

Answer 1

[Star Strider]

The ‘48’ are ASCII representations of the string representation of zero. The first row of your array must be a vector of character ‘0’ rather than numeric ‘0’.

q = char(48)
q2 = uint8(['0' 'O'])

[Image Analyst]

I agree with Star 100%. Look for code where you initialized the variables like this:

tstep = '0';  Assign to '0' which is ASCII 48 when printed as %f
xfland = '0';
mag_land_vel = '0';
mag_land_KE = '0';

and change it to

tstep = 0;
xfland = 0;
mag_land_vel = 0;
mag_land_KE = 0;

[Kirby Runyon]

Interesting idea (thanks!), but that doesn't seem to be the problem. One of my vectors, for instance, is just math (ballistics equation):

xfland=x+vfx/g.*(vfy+sqrt(vfy.^2+2.*g.*y));

But the bigger problem is that fprintf and/or fopen are somehow changing the numeric values of my results. See the second part of my original question, pasted here for convenience: the second column should be

0.5052
0.5862
0.5962
0.5922
0.5721

but fopen and/or fprintf are changing them to

0.5862
0.6441
0.6464
0.6464
0.7149

And no, I'm not doing any other operations that would change the numeric values. Thanks for your help!

[Star Strider]

Two possibilities:

1: Use a for loop to print your data:

for k1 = 1:size(tstep,1)
    fprintf(fid01, '%.4f, %.4f, %.4f, %.4f\r\n',tstep(k1), xfland(k1), mag_land_vel(k1), mag_land_KE(k1));
end

2: Concatenate your vectors and transpose them in the print statement:

Data = [tstep, xfland, mag_land_vel, mag_land_KE];
     ...
fprintf(fid01, '%.4f, %.4f, %.4f, %.4f\r\n', Data');

One of these should work for you.

I still believe you have a string of ‘0’ somewhere to account for the ‘48’ values.

[Kirby Runyon]

That worked! Thank you. How did you know to do that (you can reply to me directly)? Doing your 2nd trick took care of both problems--the bad data and the 48's.

[Star Strider]

My pleasure!

I knew how to do it from having worked with MATLAB for a while, and having encountered that problem myself. The documentation for fprintf is confusing (at least to me):

• fprintf(fileID,formatSpec,A1,...,An) applies the formatSpec to all elements of arrays A1,...An in column order, and writes the data to a text file. fprintf uses the encoding scheme specified in the call to fopen.

In the example that accompanies that part of the documentation, the data are in row vectors. This is in contrast to the usual column-major representation in MATLAB.

So if your data are in columns rather than rows, you have to transpose it to write it correctly. Concatenating them all into a single matrix first, then transposing it, is the easiest and most efficient way to use fprintf (and sprintf) to write data.