Centrality methods (was Re: BRAT 2.0.35)

From: Christian Holm Christensen (cholm@hehi03.nbi.dk)
Date: Wed Sep 05 2001 - 07:18:00 EDT

  • Next message: Truls Martin Larsen: "The .h files in the "BRAT HTML Docmentation""

    Hi Hiro, 
    
    On Tue, 4 Sep 2001 20:44:52 -0500 (CDT)
    Hironori Ito <hito@students.phsx.ukans.edu> wrote
    concerning "Re: BRAT 2.0.35":
    > 	WHO TOLD YOU THAT YOU CAN DECIDE HOW WE CUT CENTRALITY!!!  
    
    After a long discussion here at NBI, invovling all personel here at
    NBI, we came to the conclussion that the SDM centrality method was not
    advantegous over the SDE centrality. 
    
    > Did Steve, Y.K, Ramiro and I (or somebody else working on mult
    > detector (I can not think anybody else) )  ever ask you to write the
    > centrality code???  
    
    No, but since you guys didn't deliever for the longest time, I
    originally made the SDM centrality modules, based on the independent
    work by Claus and I (see BAN 22).  
    
    > If so, please let me know because I am not aware of it.  If not,
    > please change it back to whatever it was.  
    
    As I said in the mail, I have not changed the BrMultCentModule, which
    is the one you and Steve did. 
    
    > After briefly reading  your analysis note, I understand you can do
    > what Steve and I did  3-4  years ago.  
    
    Well, fine, where was the code 3 or 4 years ago! Why didn't you share
    your experience back then?  I have a hard time endorsing an argument I
    haven't seen in black and white.  Saying simple, "we did that 3-4
    years ago" is not good enough. 
    
    > However, it is decided that the cut in multiplicity is easier
    > (normalization and background correction on a real data) and more
    > clear (presentation and discussion) than what you just have done.
    
    Well, that's all well and fine, except that the SDE centrality has the
    advantage of quick calibrations and quick analysis, independent of the
    multiplicity calibrations.  You are welcome, in discussions,
    presentations, etc. to refer to BAN 34. All the graphs are avaliable
    for you to look at, play with, and so on. 
    
    > Also, look at the other experiment at the RHIC.  (STAR,  PHOBOS and
    > even PHENIX though they also use ZDC)  
    
    Well, STAR and PHENIX has a much more segmented inner tracking
    systems, and they can expect a mean occupancy of <1 in those
    detectors, thus really they have much less trouble defining
    multiplicity in the individual detectors elements than we have.  I
    don't know, but my guess is that the same goes for PHOBOS. 
    
    > If this still does not convince you, well it is too bad.  
    
    I suggest you go back an read BAN 34 one more time and then come back,
    and we can have a (civil) discussion on all this. 
    
    I'd really like to hear more comments on this. 
    
    > In any case, just change it back.  Otherwise, I will, and you will
    > really not be happy because I will change everything.  
    
    As Claus wrote you, I did not change the method in BrMultCentModule.
    All I did in that module, was to remove a number of unused members,
    that is pointers BrSiCalibration, BrTileCalibration, BrSiParameters,
    and BrTileParameters objects, since they were not used and caused the
    module to do too much unneeded work.  If you take a look at it, you'll
    be satisfied that nothing is changed in the algorithm.  I can
    understand that this change may have confused you, and I should
    probably have explaind a bit more in details what I did.  
    
    What I did, was to add to modules: 
    
      BrMultCentCalModule: 
          Calibration module for BrMultSdeCentModule.  Maybe this should
          be renamed BrMultSdeCentModule, so as not to confuse the poor
          sods.  No seriously, I think I'll change it to that.  It makes
          more sense in any case.  BTW, when will you and Steve deliever
          the calibration code for BrMultCentModule (and the rest of the
          TMA and SMA calibrations for that matter?) 
    
      BrMultSdeCentModule
            ^^^
            |||
          Notice the "Sde" in the name!  It does not overwrite your
          BrMultCentodule or even affect the operation of that module! 
    
          This module implments the SDE centrality cuts, based on the
          calibrations done by BrMultCentCalModule, and as outlined in BAN 
          34.  
    
    Also, I propegated that change to 
    
      BrGlbPackage
        so that it uses 
        * two instances of BrMultSdeCentModule 
        * _as_well_as_ and instance of BrMultCentModule
        to make _3_ centrality estimates.   
    
    Finally, I removed 
    
      BrTileCentModule 
      BrSiCentModule 
    
    from the library libBratModuleCent, but the code is still present in
    BRAT (it just isn't compiled!). 
    
    As you can see, I did not touch the thing you and Steve did.  The
    centrality method used in the multiplicity paper is still avaliable
    and I haven't changed that at all.  
    
    Frankly I am a bit surprised at your reaction, and I actually think
    it's a bit misplaced.  As outlined above, and in my previous mail, the
    change only affects BrTileCentModule and BrSiCentModule, which was
    written by me (based on independent work by Claus and I) in the first
    place (see BAN 22).  Again, these changes affect none of your code,
    only code that I did in the first place.  
    
    If you had a problem with the BrTileCentModule and BrSiCentModule
    classes, you should have made yourself heard back then, but seeing
    that you didn't, I can only assume you didn't have a problem with
    those classes, and so you have little cause to scream now!  
    
    So therefore, I have a heard time imagine what you revert.  If you
    want to put back the unused members in BrMultCentModule, then fine by
    me, but be aware that it will unnessecariy worsen the performance of
    that code.  If you want to bring back the BrTileCentModule and
    BrSiCentModule, then let me know, and I can do it quickly, but do not
    remove BrMultSdeCentModule.  These classes are orthogonal and can
    easily work side by side (then we'd have 5 centrality methods in
    BRAT).  Just be aware that I will not support BrTileCentModule and
    BrSiCentModule anymore, since they are really redundant. 
    
    My guess is, that you misunderstood what I did.  If that
    misunderstanding was due to inadequate description on my part, I beg
    your pardon.  I hope this mail has put yours and other minds to rest
    on this issue. 
    
    Yours, 
    
    Christian Holm Christensen -------------------------------------------
    Address: Sankt Hansgade 23, 1. th.           Phone:  (+45) 35 35 96 91 
             DK-2200 Copenhagen N                Cell:   (+45) 28 82 16 23
             Denmark                             Office: (+45) 353  25 305 
    Email:   cholm@nbi.dk                        Web:    www.nbi.dk/~cholm
    



    This archive was generated by hypermail 2b30 : Wed Sep 05 2001 - 07:18:54 EDT