Read a 55x55 matrix with matlab

Dear all,

I would like to read in a txt file (see attached document) the 55x55 matrix using matlab.

In the txt file, the matrix appears in a block of 9x6 followed by one line (which then accounts 55 values) . And there are 55 blocks like this.

In fact, this block contains averaging kernel values for the 55 pressure levels of the sounding instrument.

How can I read this gile so I have at the end a matrix called A which has a dimension of 55x55?

Thank you for your help,

Best regards

Mélanie

Answer 1

[Akira Agata]

How about the following?

A = readmatrix('Averaging_kernel_MLS_v04_0N.txt');

A = reshape(A',60,[])';
A = A(:,1:55);

[Mélanie Ghysels-Dubois]

Dear Akira,

Thank you for your response. It seems like it does not work: I do have the following message:

"Error using reshape

Product of known dimensions, 60, not divisible into total number

of elements, 6696.

Error in tri (line 39)

A = reshape(A',60,[])';"

The structure of the txt file containing the matrix is quite strange. In fact, the original file is the one attached here.

[Mélanie Ghysels-Dubois]

I am interested in reading the 55 blocks of 55 values and to have at the end the matric A reshaped such as size(A)=55x55.

55x55 should be 3025 values. I do not understand why the function readmatrix sees 6696 values.

[Mélanie Ghysels-Dubois]

Dear Akira,

I figured out that it was a mistake from my side. I was maning the matrix as another matrix used previously in the program.

Doing things properly, your suggestion works out well!

Thank you so much.

Best

Mélanie

[dpb]

Alternatively,

A=reshape(cell2mat(textscan(fid,'%f',inf,'CollectOutput',1)),55,[]).';

textscan won'r return the NaN elements that does readmatrix for the short records in the input file.

[Mélanie Ghysels-Dubois]

Yes, thank you dpb, Adding this line helped removing NaNs and to get the right matrix at the end.

Thanks a lot!

[dpb]

Both end up the same in the end...

>> B=readmatrix('Averaging_kernel_MLS_v04_0N.txt');
>> B=reshape(B.',60,[]).'; B=B(:,1:55);
>> all(A(:)==B(:))
ans =
  logical
   1
>> 

It's just perhaps a little cleaner w/ textscan. Too bad TMW hasn't added the ability to pass a filename as alternate input besides just a file handle so could avoid having to have the encapsulating fopen...fclose pair.