Inspecting LHCb Hadron Trigger

In this note, we analyse in detail the tracks that are triggering LHCb Hadron High Level Trigger trigger. With the help of Monte Carlo information we introduce a classification to tag these tracks according to the origin of the hits used in its reconstruction and depending on which subdetector (Velo or Tstations) we find them. We have found that the majority of the tracks triggering Hadron High Level Trigger are ghosts, i.e., tracks with hits from different Monte Carlo particles. We also see that, even when most of the hits in a track belong to the same particle, very often the absence of one of its hits or the presence of a hit from a different particle degrades the slopes of the track, leading to a wrong trigger. We show the solution for many of these issues to be a fast-fit of the tracks within the High Level Trigger framework. This permits an improval in the quality of the tracks and provides us with a $\chi^2$ value to remove those which were wrongly triggered. Reapplying the cuts after the fast-fit and cutting in the $\chi^2$ obtained we manage to reduce the rate ~50%, keeping a very high efficiency on tracks with physics interest.