[Halld-offline] event sizes with MC tags stripped out

[Mark M. Ito]

Richard,

I agree we should do this. We should discuss this at the next offline 
meeting.

   -- Mark

On 09/10/2012 03:00 PM, Richard Jones wrote:
> Hello all,
>
> I have generated a sample of minimum bias events using bggen.  The 
> photon beam spectrum in this sample is the total photoproduction cross 
> section weighted by the coherent bremsstrahlung spectrum evaluated 
> with the standard GlueX collimator, between incident photon energies 
> 3GeV and 12GeV.  This roughly corresponds to the level-1 trigger.
>
> Here are the average event sizes at various stages in the production 
> pipeline for this sample.  Note that this is the current state of the 
> simulation tools, not what you might remember from 6 months ago.  
> These event sizes are larger than you might remember hearing a while back!
>
>  1. bggen output:
>       * with MC information: 800 bytes
>       * after stripping MC information: N/A
>  2. hdgeant output:
>       * with MC information: 70KB
>       * after stripping MC information: 22KB
>  3. mcsmear output:
>       * with MC information: 102KB
>       * after stripping MC information: 49KB
>
> During the stripping process, I removed anything that was not 
> absolutely needed to store the hit information for the event.  I also 
> removed the new tags added by David associated with saving the full 
> spectrum in the bcal, and incident particle information there, even 
> though the string "Truth" was not encoded in the tag name.  My goal 
> was to strip things down to the absolute minimum required to store the 
> hits information.
>
> Unfortunately, things are not as simple as the above table might 
> suggest.  While I was not paying attention, scads of "truth" 
> information have crept in to taint the hits information that is 
> included in the stripped files above.  For example, people working on 
> tracking in the CDC have mixed truth information into the hits tags, 
> as in the following example:
>
>  <cdcStrawHit maxOccurs="unbounded" d="float" dE="float" itrack="int" 
> ptype="int" t="float" />
>
> What is track number doing inside this hits tag, you might ask????  
> The equivalent and correct way to encode this same information in hddm 
> would be the following.
>
> <cdcStrawHit maxOccurs="unbounded" d="float" dE="float" t="float>
>      <cdcStrawHitTruthInfo minOccurs="0" itrack="int" ptype="int"/>
> </cdcStrawHit>
>
> Encoded in the new way, it is easily stripped, but the association is 
> still there in the unstripped event.  Encoded in the original way, the 
> entire tag is tainted with truth information.  If the substring 
> "Truth" occurs in the tag, it should be stripped in the final 
> analysis.  If not, it should be completely clear of truth information 
> and should be kept for the final analysis.
>
> Once we fix these problems, I will rerun the above stripping exercise 
> and we will get a better estimate for what the average event size will be.
>
> -Richard Jones
>
>
>
> _______________________________________________
> Halld-offline mailing list
> Halld-offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-offline

-- 
Mark M. Ito
Jefferson Lab (www.jlab.org)
(757)269-5295

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20120911/c188fe36/attachment-0002.html>