LSLTrackRep now can produce the required 6x6 covariances.
The respective methods are:
virtual TVectorT<double> getGlobal(); // (x,y,z,px,py,pz)
virtual TMatrixT<double> getGlobalCov(); // covariances
When the Kalman Filter has been run, this should produce a valid covariance. See also GenfitTester::test_LSLtoGLOB();
However there is one issue:
Since there is always one free parameter in a track (z in the case of LSLTrackRep) the 6x6 covariance matrix is a bit strange. For example at the moment the z-row/column has only the variance and no COvariances! I do not know how to solve this.
Originally this was the reason why I suggested that we should NOT have a 6-coordinate system.
But maybe there is a solution to this which I do not see at the moment.
At least you can start palying around now. Please give me some feedback on what you get! So far I have not investigated the covs too much!