Linear Collider Forum



Home » Analysis and Reconstruction » Tracking & Vertexing » Fitting scheme
Fitting scheme [message #1981] Tue, 11 May 2010 08:33 Go to previous message
rosemann
Messages: 41
Registered: March 2009
Location: hamburg.de
Hi,

after giving it some thought, I would like start the discussion about the current and future scheme of fitting in MarlinTPC.

Probably Martin (are you there?) will comment on this, but this represents my understanding of the current scheme:
  • The base class is "TrackFitterBase", which basically just defines an interface.

CR: I'm not too happy about the naming scheme for the /residuals/ access, but in principle I think it is OK to have something like this. It forces to implement some functions, but I can't see the disadvantage of this.

  • The actual fitter class are derived (inherited) from the above base class and are not Marlin processors.

CR: I can agree to both. We would like to call the fit iteratively. With processors it is believed to be possible (I talked to Steve about "processors calling processors"), but I don't think we should make life more complicated.

CR: As a different thought I eventually think that a much simpler "fit object" might be more natural. I would throw the fit object the LCIO::TrackerHits and deal with the result.

  • The calling of the fitter class is done by a processor.

CR: This seems obvious, since we want to do it within MarlinTPC. But I guess we might want to do the "bookkeeping" part in the processor, rather than in the fitter class.

  • The TrackFitterFactory can be used to handle different fitted tracks by their type; by using slightly peculiar "types" defined in the TrackFitterBase class.


CR: I have to admit that this is a possible solution to the problem: "how do I determine the residuals of a particular already fitted track". But it is not the most straightforward (at least to me).


---- BREAK -----

I would approach this whole problem slightly different. I would start with the description of what we want to do, then try to write it and think about how we might want to use it differently/figure out what is missing.

What do we want do?
  1. fit tracks on already found track hypotheses with associated hits
  2. calculate residuals (and/or distance) to obtain resolutions
  3. reject outliers
  4. ...


I can say that we (in Hamburg) now start(ed) to develop/implement a full fledged (with input and output errors) chi square fitter for a straight line and a helix. Hopefully more ideas will come along the way.


Cheers,
Christoph


When you have eliminated the impossible, whatever remains, however improbable, must be the truth. (Sir A.C. Doyle in Sign of Four)
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic:TPC Tracks not being produced
Next Topic:Conditons Data for MarlinTPC
Goto Forum:
  


Current Time: Thu Aug 16 13:02:38 Pacific Daylight Time 2018
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.1.
Copyright ©2001-2010 FUDforum Bulletin Board Software