I have a puzzle concerning track parameter propagation and the covriance matrices. I already contacted Lia via mail, but want to share my investigations.

We have the following situation: A track is defined with its first parameter set yielding the following cov matrix:

Start MARS cov (px,py,pz,E,x,y,z): 6x6 matrix is as follows | 0 | 1 | 2 | 3 | 4 | 5 | ------------------------------------------------------------------------------ 0 | 4.059e-05 3.814e-05 -9.897e-05 -2.728e-07 -8.99e-07 0 1 | 3.814e-05 3.77e-05 -9.529e-05 1.051e-06 -2.188e-06 0 2 | -9.897e-05 -9.529e-05 0.000248 -2.797e-06 2.073e-06 0 3 | -2.728e-07 1.051e-06 -2.797e-06 7.092e-06 5.262e-08 0 4 | -8.99e-07 -2.188e-06 2.073e-06 5.262e-08 7.082e-06 0 5 | 0 0 0 0 0 0

After propagating the track parameters to the distance of closest approach of the z axis, I get the following cov:

End MARS cov (px,py,pz,E,x,y,z): 6x6 matrix is as follows | 0 | 1 | 2 | 3 | 4 | 5 | ------------------------------------------------------------------------------ 0 | 5.94e-05 -1.932e-06 2.146e-05 -0.000288 9.157e-06 -0.0001177 1 | -1.932e-06 5.631e-05 1.295e-05 2.375e-05 -0.000268 -8.752e-05 2 | 2.146e-05 1.295e-05 0.0002708 -0.0005018 -0.0004349 -0.0003691 3 | -0.000288 2.375e-05 -0.0005018 0.002029 0.0004619 0.001021 4 | 9.157e-06 -0.000268 -0.0004349 0.0004619 0.001822 0.000857 5 | -0.0001177 -8.752e-05 -0.0003691 0.001021 0.000857 0.0007408

Looking at x,y,z (cols/rows 3,4,5) I get massively increased error values, like from sigma_x=27um to sigma_x=450um.

The residuals look really OK and even the pull distributions for the momenta are well, considering that the beam pipes material blows the cov a bit up.

I can only imagine there is some mismatch between the fortran units and Fairsoft (cm) or a too large contribution of the beam pipe.

I looking forward to some ideas.

Regards. Ralf]]>

Concerning the mismatch between fortran and fairroot units I would exclude this, since we made several tests (pulls and residuals) and we always got good results, so any mismatch should have shown up also before.

I have a couple of questions: which pdg code are you using to propagate and where are you propagating from in this specific event? What are your starting point and momentum?

Cheers,

Lia.]]>