dedx / 3dtracker Goto Github PK

Hi,
i tried to use your method as described in your work "REGRESSIONS et TRAJECTOIRES en 3D" at http://de.scribd.com/doc/31477970/Regressions-et-trajectoires-3D#scribd. I found your python lib here and tried a couple of datasets to find a reasonable 3-D linear regression.
It works quite well, but if i insert some points on one line, it does not work. For e.g. i tried with this set:
x = np.array([2.82252, 2.82252, 2.82252, 2.82252, 2.82252])
y = np.array([-0.425, -0.425, -0.425, -0.425, -0.425])
z = np.array([-5.22366, -5.42366, -5.62366, -5.82366, -6.02366])
This will result in some rounding issues at the arctan generation (easy to fix, since you just have to check for bigger or lower values than 1 or -1), but, lateron it will crash the dm2 calculation due to a root of a negative number (all following numbers will get NaN).
So here is my question, is it possible to fix that - or do you have any idea how i can prevent that errors ? (e.g. use the points directly, since there are already on one line;))
Cheers and thanks in advance
Benedetto