Issues for covariance matrix in forward tracking
There are two issues for covariance matrix in forward tracking:
- In codes like https://github.com/JeffersonLab/clas12-offline-software/blob/development/reconstruction/dc/src/main/java/org/jlab/rec/dc/track/fit/KFitterDoca.java#L173 for copying covariance matrix of final state. The reason is that reference variable for final state = another reference variable, which is changed later. The issue only affects output bank REC::CovMat
- In functions for transport like https://github.com/JeffersonLab/clas12-offline-software/blob/development/reconstruction/dc/src/main/java/org/jlab/rec/dc/track/fit/StateVecsDoca.java#L90, covariance matrix of initial site is replaced by covariance matrix of final site. This issue doest not affect hit-based tracking hit-base tracking, but affects time-based tracking. The issue affects tracking results for both state vector and covariance matrix.
Here are differences for hit-based particles, time-based particles, and bank REC::CovMat before and after issue fixing. As expected, there is no difference for hit-based particles, but there are differences for time-based particles and covariance matrix. diff of RECHBParticle.pdf diff of RECParticle.pdf diff of covmat.pdf