BRAT updates of clustering/tracking

From: Peter H. L. Christiansen (pchristi@nbi.dk)
Date: Mon Oct 02 2000 - 08:45:21 EDT

  • Next message: Peter H. L. Christiansen: "Re: Note on tracking-efficiency, TPM1"

    I checked in my changes to the clustering algorithm and a few tracking
    changes.
    
    **IMPORTANT!!!**
    I added a user manual to the clustering on the BRAHMS Analysis note page.
    If you don't have the latest and greates version of BRAT ignore the
    comments about BrTPCPreProcess.
    
    Comments are more than welcome.
    
    Ian pointed oyt that it is the HitSearchWidth that is important and
    not the search width when you use the TrackFollowFinder with tracking method
    = 1. Major bummer by me. 
    Use HitSearchWidthX 0.4(pads) and  HitSearchWidtY(time)0.2 or keep the
    default which is 0.4 and 0.4 and then I will try to refine these
    parameters during the week. 
    ****************
    
    Now a list of changes :
    BrTPCPreProcess :
    Added a new feature to the preprocessor. It now cuts the clusters up
    removing 0 and negative ADC values and as default keeps pieces that
    are longer/equal to 3 time bins.
    Added a method TestOutput that can test if the output is reasonable
    compared to what we want. It is a good idea to look in this to see
    what the effect should be on the output and to use this option a few
    times to see that it is working and if you are changing something in 
    the module.
    Changed default to 0 timebins, so that people using the
    BrTrackFollowFinderModule ( see below ) does not unknowingly cut away
    the first time bins.
    How to use was updated in the class description.
    
    BrTPCSequence :
    Added IsEqual method that is used in BrTPCClusterFinder::FindClusters
    when I want to cahnmge the cluster ID of a cluster in the input table
    (special case where 2 clusters merge to one). The pointers have
    different addresses and so this one was needed.
    
    BrTPCCluster :
    Added a lot of info to this one like first timebin, firstpadbin,
    admax, mean time, mean pad, mean var, pad var and now nTime and Npad
    have the correct meaning.
    This is all calculated in one method Build which is called first time
    asking for the values or if something has changed. 
    
    BrTPCClusterTable :
    Added new methods. Remove without link and get cluster by ID.
    Did NOT yet correct my hack with always creating a TList
    
    BrTPCClusterFinder :
    Added a few histograms with the input (After FindClusters) and output
    data(After deconvolution and pad cut).
    In the method FindClusters the method now really adds all sequences
    that have overlap of at least 2 sequences into one cluster. 
    The clusters that come from deconvolution are treated correctly. Added
    a new category for the data (kMultiHitDec) which are clusters that are
    still to large after deconvolution.
    I have added a large number of assert statements to make sure I
    understand how the output propagates through the routine. 
    Added a method TestOutput that can test if the output is reasonable
    compared to what we want. It is a good idea to look in this to see
    what the effect should be on the output and to use this option a few
    times to see that it is working and if you are changing something in 
    the module.
    
    BrTPCClusterFinderBase :
    Added a few histograms with the hit errors.
    
    BrTPCHitCluster :
    Added a non streamed pointer to the cluster with Get/Set methods.
    
    BrTPCHitClusterFinder :
    Cluster was set, see above.
    
    BrTrackFinderBase :
    Added histograms :
    Number of hits per track, chi**2, and number of final local tracks
    See 2 next.
    
    BrTrackFollowFinder :
    Added call to BrTrackFinderBase::DefineHistograms in this modules
    DefineHistograms method.
    
    BrTrackStringFinder :
    Added call to BrTrackFinderBase::DefineHistograms in this modules
    DefineHistograms method.
    
    BrTrackFollowFinderModule :
    Added preprocessor and get methods. 
    Overwrote Book method so it calls Book for all modules.
    How to use was updated in the class description.
    
    
    Comment from Christian :
    He thinks that the Entry method of both the clusterfinder and the
    tracker should be changed into an overwritten Event Method that might
    call the base Event Method.
    
    That's all folks.
    
    Cheers,
       Peter
    
    :-) --------------------------------- )-:
    |Peter H L Christiansen aka PAN @ NBI	|
    |EMAIL  : pchristi@nbi.dk		|
    |OFFICE : Tb1 @ NBI			|
    |PHONE  : 353 25269 <- New!!		|
    |SNAIL  : Sdr. Fasanvej 14 ST 2000 F	|
    |PHONE  : 38 872042 			|
    :-D --------------------------------- \-:
    



    This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 08:56:43 EDT