Fread format causes errors

Hi,

So I have a simple program that uses fread three times to read the data from a 3x3 matrix of numbers (.mat format). Each time it reads, I want it to read using a different size output matrix. However, the third time I read, it outputs a blank 1 cell matrix. This happens regardless of what I put into the size for that third instance of fread. Here's the code: if true % code " fid=fopen('numbers.mat','r');

sizeC1=[2, 2]; sizeC2=inf; sizeC3=[4, 4];

precision1='char'; precision2='double=>char';

c1=fread(fid,sizeC1,precision1); c2=fread(fid,sizeC2,precision1); c3=fread(fid,sizeC1,precision1);

fclose(fid); " end

Here's my outputs:. I'm not writing c2 because it is a giant vector, but seems to be behaving appropriately. " c1 =

    77    84
    65    76

c3 =

     []
"
On an unrelated note, how does formatting code work for these forums?

[Jan]

@Dylan: Welcome in this forum! I'm very glad to see, that you care about formatting and ask actively for help. Several thousand other users did not care about a proper formatting and some hundred did not even care after they have been asked for.

You find even questions about this forum answered here, so it is worth to search for it: http://www.mathworks.com/matlabcentral/answers/?term=code+formatting

A nice summary can be found at http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup or you can click on the "? Help" link on top of the field to input a message.

Answer 1

[Jan]

I do not understand the intention of the code.

fread() cannot read -MAT files properly, because this format contains headers and the data can be compressed etc.

'double=>char' is a strange format converter. What do you expect as output?

Reading Inf elements from a file moves the file pointer to the end. Then a following fread() cannot read any element anymore.

It is not clear where the "3x3 matrix" appear in the code.

Answer 2

[Image Analyst]

Why do you assume that a mat file has just a raw binary format and not some proprietary format? Try using load() instead of fread():

% Read everything back in to MATLAB.
storedStructure = load('numbers.mat');
% Save the structure members into individual variables for convenience.
c1 = storedStructure.c1;
c2 = storedStructure.c2;
c3 = storedStructure.c3;

Answer 3

[dpb]

Maybe it'll be easier to see w/o the obfuscation of the substitutions...what you ahove above translates to

c1=fread(fid,[2,2],'char'); 
c2=fread(fid,inf,'double=>char'); 
...

the SIZE argument of inf in the second read says fill c2 reading to EOF so, sure, anything after that will simply return empty.

BTW, if the input file really is a '.mat' format written by save then you need to use load to retrieve it as it includes more than just the data.

See the following example --

>> whos x1
Name      Size            Bytes  Class     Attributes
x1        1x10               80  double              
>> save x1                           % save it as .mat file
>> fid=fopen('x1.bin','w');          % then as stream...
>> fwrite(fid,x1,'real*8');
>> fid=fclose(fid);
>> d=dir('x1.*');                    % and resulting file sizes...
>> d(:).name, d(:).bytes             % NB: I formatted output a little
ans =     ans =     
x1.bin           80
x1.mat         2724

Note the .mat file contains the same 10 doubles is all but is 3X the size of the stream .bin file that is only the data.