Skip to content

Track \(\chi^2/DOF\) data-MC inconsistency

For the run 1 and 2 data-MC comparison, we observed that for run 1 data, MC and run 2 MC, their track \(\chi^2/DOF\) distributions are all centered around 1, yet that of run 2 data is centered around 0.75:

Track chi2 inconsistency

Puzzled by this, we asked experts1 about this. The answer was:

  • The OT drift time was improved so the actual tracking performance is better
    • This is mostly due to better alignment and calibrations.
  • However, the OT uncertainties used in \(\chi^2\) calculation was not updated for data, and data only.

    • Now we have something like this:

      mode residual resolution
      data new old
      MC new new
    • Note that \(\chi^2\) here is computed by \(\frac{\text{residual}}{\text{resolution}}\).

    • As a result, MC does not agree with data.

Note

  • In DaVinci, the branch name is _TRACK_CHI2NDOF
  • This also affects related variables, like Ghost probability rate (_TRACK_GhostProb)
  • The implication of this is still under discussion

  1. Gregory Ciezarek and Lucia Grillo