Read matrices that are separated by underscores in a text file.

I have a large text file that has two header lines followed by a long list of 3x3 matrices that are separated from one another by underscores. Here is a small portion of the file I am using. Don't mind that these matrices are all the same. There is a second set of matrices which are different.

I am trying to figure out how to read these into the work space iteratively to perform calculations with them. I will be doing the same with the other set of matrices which has an identical format. The calculations will be performed between the two matrices.

How can I reference and pull specific lines of the text file in a for loop, while skipping the underscore seperators?

RotmA

--------------------------

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

0.999985 0.000046 0.005519

0.000046 0.999863 -0.016557

-0.005519 0.016557 0.999848

___________________________

Answer 1

[Walter Roberson]

In the below, the assignment to S1 inside isunix() is just to load in some data since I do not have your file. In your actual code, you would use the fileread() to load data into S1 instead.

Afterwards, you would not use line numbers: you would index the mats1 cell array according to which matrix number you wanted.

filename = 'first_file.txt';


if isunix()
  S1 = sprintf('          RotmA                \n--------------------------\n0.999985 0.000046 0.005519\n0.000046 0.999863 -0.016557\n-0.005519 0.016557 0.999848\n___________________________\n0.999985 0.000046 0.005519\n0.000046 0.999863 -0.016557\n-0.005519 0.016557 0.999848\n')
else
  S1 = fileread(filename);
end

S1 = 
    '          RotmA                
     --------------------------
     0.999985 0.000046 0.005519
     0.000046 0.999863 -0.016557
     -0.005519 0.016557 0.999848
     ___________________________
     0.999985 0.000046 0.005519
     0.000046 0.999863 -0.016557
     -0.005519 0.016557 0.999848
     '

blocks = regexp(S1, '(?<mat>(-?[\d.]+\s*){8}-?[\d.]+)', 'names');
mats1 = arrayfun(@(S) sscanf(S.mat, '%f', [3 3]), blocks, 'uniform', 0)

mats1 = 1×2 cell array
    {3×3 double}    {3×3 double}

celldisp(mats1)

 
mats1{1} =
 
    1.0000    0.0000   -0.0055
    0.0000    0.9999    0.0166
    0.0055   -0.0166    0.9998

 
 
mats1{2} =
 
    1.0000    0.0000   -0.0055
    0.0000    0.9999    0.0166
    0.0055   -0.0166    0.9998

 

[Michael Costa]

This works perfectly. Much appreciated.

[Walter Roberson]

Oh, I see a correction needed:

mats1 = arrayfun(@(S) sscanf(S.mat, '%f', [3 3]).', blocks, 'uniform', 0)

Otherwise the matrices come out transposed.

[Michael Costa]

Oh nice catch. Thank you