Function returns different outputs with same inputs

Hello everyone,

Context (not relevant for Problem):

For a school project, im playing around with "rotation matrix" and "Quaternions", i already calculated the Quaternions with a different methode, and now i want to compare the results i get from rotationsmatrix2quaterionen(numeric_result) with the ones i calculated earlier. The function seems to work, rr is nearly zero (not exactly because of numeric imprecision)

Problem:

I don't understand why 'rr' and 'gh' give me different results. The calculation should be the same both times, only that one time at first i write the output in a seperate array.

I assume i made some stupid mistake, but i can't find it.

numeric_result = [    
    0.8138   -0.4063    0.4155   10.0000;
    0.4698    0.8808   -0.0590   12.5000;
   -0.3420    0.2432    0.9077   64.0000;
         0         0         0    1.0000]

numeric_result = 4×4

0.8138 -0.4063 0.4155 10.0000 0.4698 0.8808 -0.0590 12.5000 -0.3420 0.2432 0.9077 64.0000 0 0 0 1.0000

 e0 = 0.9490

e0 = 0.9490

 e1 = 0.0796

e1 = 0.0796

 e2 = 0.1996

e2 = 0.1996

 e3 = 0.2308

e3 = 0.2308

 
[e0_, e1_, e2_, e3_] = rotationsmatrix2quaterionen(numeric_result)

e0_ = 0.9490

e1_ = 0.0796

e2_ = 0.1996

e3_ = 0.2308

rr=[e0, e1, e2, e3]-[e0_, e1_, e2_, e3_]

rr = 1×4

1.0e-04 * 0.1370 -0.1127 0.4496 0.0109

% Output:
% rr = 1×4    
% 1.0e-04 *
% 
%     0.1370   -0.1127    0.4496    0.0109

gh=[e0, e1, e2, e3]-rotationsmatrix2quaterionen(numeric_result)

gh = 1×4

0.0000 -0.8694 -0.7494 -0.7182

% Output:
% gh = 1×4    
%     0.0000   -0.8694   -0.7494   -0.7182



function [e0, e1, e2, e3] = rotationsmatrix2quaterionen(RotMat)   
    % e = e0 + e1*i + e2*j + e3*k
    R=RotMat;
    e0 = 0.5*sqrt(1+R(1,1)+R(2,2)+R(3,3));
    e1 = (R(3,2)-R(2,3)) / (4*e0);
    e2 = (R(1,3)-R(3,1)) / (4*e0);
    e3 = (R(2,1)-R(1,2)) / (4*e0);
end

Answer 1

[Stephen23]

The difference is very simple:

Here you return FOUR output arguments from your function call:

[e0_, e1_, e2_, e3_] = rotationsmatrix2quaterionen(numeric_result)

Here you only return ONE output argument from your function call:

gh=[e0, e1, e2, e3]-rotationsmatrix2quaterionen(numeric_result)

[Stephen23]

The first output argument has a value of approx. 0.9490, so your second attempt simply subtracts one scalar from a vector of four values. Lets try it right now (due to floating point accuracy the result will only be similar to what you showed):

[0.9490,0.0796,0.1996,0.2308] - 0.9490

ans = 1×4

0 -0.8694 -0.7494 -0.7182

You seem to be writing your MATLAB code as if it were Python.

[Daniel]

Thank you very much, that makes sense