Monte Carlo

Page maintainer: Achim

Important note: This page has not yet been reviewed by Elke! Use the information below with caution.

Other pages that have not been reviewed yet can be found in the category FORREVIEW.

Intro

This page is meant to be a jumping point for information about MCs. Please feel free to add topics - with or without explanations!

Essentials

See the non-wiki MC page for lots of useful info!!!

Generating Monte Carlo

MC Generators

See the MC Generators page, as well as the DVCS Monte Carlo page

Selector files

Selector files are used to produce events within certain cuts. Generated events are saved or discarded based on the criteria in the selector file.
- In the selector file the user can specify the number of track, geometric cuts, particle type, track momentum, and parent particle.
- Running with a selector file will cause GMC to take longer (it will have to generate more events before a file of selected events is filled), however the advantage is that the rest of the MC production chain (HMC/HSG, uDSTwriter) is only run on events of interest.
- Selector files are used, for example, to generate high momentum multitrack events to fill the RICH Pmatrices
See the event selector help file for details on how to write a selector file

Radiative Corrections

Radiative corrections are calculate using the RADGEN program
RADGEN:
1. takes as input the observed kinematics of an event (in (y,Q2) )
2. (potentially) generates a radiative photon according to the probabilities for these kinematics
3. returns the true kinematics at the interaction vertex (which are different from the observed kinematics in case a photon has been generated)
RADGEN is used in the disNG and PYTHIA generators
Also see the Unfolding of Detector Smearing and QED Radiative Effects page
References:
- hep-ph/9906408v1, RADGEN 1.0, I. Akushevich, H. Boettcher, D. Ryckbosch
- hep-ph/9706516v1, reference to POLRAD 2.0
- Implementation in PYTHIA: Patricia Liebing's thesis (including detailed information about the differences in the implementation in PYTHIA and disNG, see Appendix B)

Fragmentation

The fragmentation of the partons into final state hadrons is performed using the JETSET program.
JETSET is included in PYTHIA and disNG
JETSET is an implementation of the Lund string fragmentation model

HERMES Monte Carlo (HMC)

The HMC takes generated events and runs them through a simulation of the HERMES detector.
- The HMC code has options to run with particular detector elements "off" - the detector mass is still included but the detector response is not recorded. This is useful as the code then runs much faster. Often the calorimeter and the RICH are turned off in MC productions.
  - To check if these detectors are on or off you can always open a file from the production in Pb and check to see if the desired tables are filled
HMC operation and event generation: Describes how to run the program, what modes of operation exist, and what options can be adjusted, as well as details about the generation of events.
HMC tracking and digitization: Provides detailed information on the way HMC tracks particles using GEANT, how detector hits are recorded, and how these are digitized to produce a simulated response for each inidividual detector.
HDB (Hermes Data Base) guide: Provides a complete description of the geometry database and the HDB program which manages it.

Aligned / Misaligned MC and systematic errors

The HERMES spectrometer has known misalignment which can be taken into account by running your analysis with misaligned MC (misalignment is specified in the geometry file) and extracting a systematic uncertainty ( e.g. for DVCS).

See the Misalignment page
Beam shifts/positions can be found on the Tracking Corrections and Beam position pages

HERMES Smearing Generator (HSG)

The HERMES smearing generator is used to simulate detectors smearing on generated MC events without actually running a full detector simulation via HMC.
Track momenta and angles are modified on a statistical basis
These modifications are based on a MC generated lookup table which takes into account the track momentum and the particle type (lepton or hadron)
Advantage: kinematic effects introduced by detector smearing can be studied with MC production produced much faster (~10% of the time needed for fully tracked MC)
But: it's a simplified, statistical model and should be used with care, there are no specific detector responses, no PID values, ...
Unfolding of Detector Smearing and QED Radiative Effects for final results should be done with a properly tracked MC production
References: PhD thesises of Felix Menden and Achim Hillenbrand
HSG acceptance: example code and all necessary functions and include files to make proper momentum dependent fiducial volume cut if HSG was used

HERMES ReConstruction (HRC) & uDSTwriter

As with experimental data, MC productions should be run through HRC and the uDST writer
- HSG MC productions produce HRC tables directly, so they can be piped straight to uDST

Important changes in the HERMES MC code

HTC added to r25 on Feb 18 2009
Changes made by Elke to gmcDVCS, Dec 28, 2008
New link smCluster.mcHit added April 2005
change of minimum Q2 and W2 cuts (ca. February 2007)
- before: disNG productions where usually produced with Q2true>1 GeV2 and W2true>10 GeV2 as cuts on hadron production
  ⇒ no hadrons were produced for events below these true kinematics
  ⇒ events that "smeared" into acceptance (Q2obs >1 GeV2, W2obs > 10 GeV2) had by definition no hadrons
- after: disNG cuts were lowered to Q2true > 0.5 GeV2 and W2true > 4 (minimal possible values)
- for a study and before/after comparison, look here
New "hmc 2005" version?! Is that the same as the min-cuts?

Using Monte Carlo

MC uDST tables

MC uDST docu
Contents of MC uDST tables (g1_MC.ddl)
Contents of Adamo tables in hrc data
g1MTrack definition, including icType codes
Table of lund particle codes

MC Productions

See the MC Productions page

Normalization

See the Normalizing MC page, including how to use event weights, IEVGEN, extraweight, and more.
For data normalization, including notes to how to normalize data using a known cross section from MC, see Cross section (data)

Event weights & Statistical Errors

See the Normalizing MC page
Be sure to use event weights when calculating statistical errors in MC
B. Krauss, Statistical Errors and Monte Carlo (Hermes Internal Note)

Unfolding

See Unfolding of Detector Smearing and QED Radiative Effects

More Info

Monte Carlo Category page