====== Event Format : extended LHE ======

People involved: Eric & Benj
===== Present state =====

{{ :2013:groups:tools:mc_formats.png?400 |}}

Sample size: example of ttbar dileptonic @ LHC (10,000 events)

|              ^ File size (Mo)       ^ 
^ LHE (gzip compression)   | 3.8          | 
^ STDHEP (gzip compression)   | 153 | 
^ HEPMC (gzip compression)   | 346          | 
^ simplified LHE (gzip compression)   | 5.1          | 
^ LHCO (gzip compression)   | 1.6          |
^ Delphes 2 ROOT    | 161         |
===== Motivations =====

  * defining a format for jet-clustering output (without fast-simulation detector).
 
  * defining a format which extends the LHCO content (too few information for performing sophisticated analysis).

  * the new format will take into account all Delphes 3 potential.


===== Some ideas to discuss =====

  * Using text format. ROOT is rejected ; STDHEP seems to be old.

  * Prefering extending an existing format to defining a totally new one (avoiding developers from coding writer and reader functions from scratch). Our choice is to extend the LHE Format (arXiv:hep-ph/0609017) and its structure based on XML tags. Reminder about the LHE structure:

  <LesHouchesEvents version="X">
  <header>
  ...
  </header>
  <init>
  2212     2212  0.40000000000E+04  0.40000000000E+04 0 0 10042 10042 3  1
  0.47468358499E+01  0.15068796356E-01  0.47469000000E-03   0
  ...
  </init>
  <event>
   12   0  0.4746900E-03  0.2312331E+03  0.7957747E-01  0.1132798E+00
   21   -1    0    0  501  502  0.00000000000E+00  0.00000000000E+00  0.74064204368E+02   0.74064204368E+02  0.00000000000E+00 0.  1.
   21   -1    0    0  502  503  0.00000000000E+00  0.00000000000E+00 -0.74552086368E+03  0.74552086368E+03  0.00000000000E+00 0.  1.
   -6    2    1    2    0  503  0.14952840473E+03 -0.23999735524E+02 -0.41424800778E+03  0.47441561784E+03  0.17473990778E+03 0.  0.
  -24    2    3    3    0    0  0.56722398399E+02 -0.36860071438E+02 -0.33540004381E+03  0.35186997544E+03  0.82116958530E+02 0.  0.
    6    2    1    2  501    0 -0.14952840473E+03  0.23999735524E+02 -0.25720865153E+03  0.34516945021E+03  0.17335203433E+03 0.  0.
   24    2    5    5    0    0 -0.16699616992E+03  0.38357854935E+02 -0.25987491067E+03  0.32192128147E+03  0.82093218139E+02 0.  0.
  -13    1    6    6    0    0 -0.76026472087E+02  0.53922169130E+02 -0.95737952146E+02  0.13361654188E+03  0.00000000000E+00 0.  1.
   14    1    6    6    0    0 -0.90969697833E+02 -0.15564314195E+02 -0.16413695853E+03  0.18830473960E+03  0.00000000000E+00 0. -1.
    5    1    5    5  501    0  0.17467765185E+02 -0.14358119410E+02  0.26662591415E+01  0.23248168736E+02  0.46999998093E+01 0. -1.
   11    1    4    4    0    0  0.50813684997E+02 -0.61274565657E+02 -0.22556131392E+03  0.23919554619E+03  0.00000000000E+00 0. -1.
  -12    1    4    4    0    0  0.59087134026E+01  0.24414494219E+02 -0.10983872989E+03  0.11267442925E+03  0.00000000000E+00 0.  1.
   -5    1    3    3    0  503  0.92806006335E+02  0.12860335914E+02 -0.78847963972E+02  0.12254564240E+03  0.46999998093E+01 0.  1.
  </event>
  <event>
  ...
  </LesHouchesEvents>

  * Extending the LHE format in order to store partons, hadrons and jets (reco objects) in a same file. The generation step (hard-process, shower, reco) will be specified by the status-code. Some details:

----

    * parton level: same conventions than the existing LHE. Example:

  6    2    1    2  501    0 -0.14952840473E+03  0.23999735524E+02 -0.25720865153E+03  0.34516945021E+03  0.17335203433E+03 0.  0.

    * hadron level: the conventions can be applied without too much change (maybe the meaning of the two ICOLUP variables could be discussed).

  2212    2    1    2  501    0 -0.14952840473E+03  0.23999735524E+02 -0.25720865153E+03  0.34516945021E+03  0.17335203433E+03 0.  0.

    * reco level: the conventions have to be adapted. We can keep:
      * one line per physics object.
      * ISTUP variable with specific PDG-id for reco objects. Example: 11/-11 for electrons, 13/-13 for muons, 15/-15 for hadronically-decaying taus, 22 for photons, 21 for jets, 12 for MET, -12 for MHT
      * MOTHUP variables linked the reconstructed object to the originated partons (only for some objects). Example: a electron coming from hard-process.
      * PUP variables without change.
      Other relevant variables, specific to the nature of the reconstructed objects, must be added.
     

----

  * Optional substructure in <event> block: defining a XML tag for each collection of reconstructed objects. When detector fast-simulation is applied, several configurations can be applied (for instance for lepton isolation) and several collection of the same object kind can be produced. The <event>-block substructure is designed to handle several collections of a same object kind. Example:

  <event>
  ...
  
  <jets>
  ...
  </jets>
  <jets name="substructure John Hopkins algo">
  ...
  </jets>
  <electrons>
  ...
  </electrons>
  <muons>
  ...
  </muons>
  </event>


Warning: If the LHE file supplies several collections of jets, people must obviously use only collection of jets in their analysis.