SimpleTrackingMM method - checks in CanTrack

At this time, the SimpleTrackingMM method implements a check on the number of MicroMega hits for tracking, using the CanTrack() function. CanTrack() checks on upstream and downstream MM. For Downstream MM, the code is:

bool MManager::DownStreamOK() const
{

 const unsigned int Nup = downstream.size();
 if(Nup < 1)     return false; //TODO: in theory is possible with just one
 else return true;
}

So, even if only one of the downstream MM has a valid cluster, the code will return true. It seems to me this is inconsistent with SimpleTrackingMM, that relies on the two hits after the magnet (see https://gitlab.cern.ch/P348/p348-daq/-/blob/master/p348reco/tracking.h?ref_type=heads#L47).

I checked the following:

• DownStreamOK() is called only in CanTrack()
• CanTrack() is used in SimpleTrackingMM(), PointDeflectionTracking(), PrintMM(), GenfitTrack()

It looks that PointDeflectionTracking has to do with old (2018!) visible-mode data. My proposal: change DownStreamOK() to DownStreamOK(bool requireAtLeastTwo=false), and set requireAtLeastTwo=true for SimpleTrackingMM

May I ask your opinion @bbantoob @mituzi? I know that for appropriate tracking we do not rely on this code, but this is useful in early stage of analysis.

created branch 168-simpletrackingmm-method-checks-in-cantrack to address this issue

mentioned in merge request !563 (merged)

@acelenta Indeed this code is ancient and has not been updated to match the current usage of track reconstruction in the invisible searches. As for the methods, you may modify them as you see fit

Hi @acelenta I have not used this code myself indeed as it had quite a few limitations. For me it is also fine if you modify the methods.

Eventually, we may want to incorporate the "simple" tracking that we use in muon, which does a linear fit of the hits before and after magnet separately, and then computes the angle of the resulting lines and from that the momentum.

Hi @acelenta ,

the SimpleTrackingMM() require at least 2+2 hits,

the PointDeflectionTracking() and GenfitTrack() require at least 2+1 hits,

My suggestion is to make the CanTrack approach very explicit as probably there are some ancient analysis codes which suffer from the same problem, so:

• delete current CanTrack()
• add the CanTrackSimple() with the check for 2+2 hits
• add the CanTrackDeflectionOrGenfit() with the check for 2+1 hits
• move the DownStreamOK() to the private

closed with merge request !563 (merged)

mentioned in commit cddf5345