search

HEP-KBFI / ml-tau-reco

1 stars 3 forks source link

Beam backgrounds, gg->had #57

Closed jpata closed 11 months ago

jpata commented 1 year ago

  1. Starting with the CLIC CDR, we have the following beam-induced backgrounds: Screenshot from 2023-02-20 15-47-29 of which, let's focus on gg->hadronic.

  2. According to the CDR section 2.1.2.2 and table 2.1, at 3 TeV, we have 312 bunches per train, and 3.2 events with W_gg>2GeV.

  3. From a CLIC workshop talk, we have about about 0.3 gg->had events at 500 GeV, and again 312 bunches per train. Screenshot from 2023-02-20 15-48-38

  4. From the CLIC reco configuration, for 380 GeV, we have 0.0464 or 0.18 overlay events per bunch crossing ( NumberBackground), and 30 bunches in a train (NBunchtrain).

  5. Looking at the Marlin OverlayTiming code (which I don't manage to run), for each bunch in the train, a poisson or int-rounded value from NumberBackground is chosen to overlay the events.

Questions:

Simulation code:

veelken commented 1 year ago

The design luminosity at the FCC-ee is 7 x 10^34 cm^-2 s^-1 at a center-of-mass energy of 240 GeV for ZH and is ~1 x 10^34 cm^-2 s^-1 at a center-of-mass energy of ~360 GeV for ttbar production (page 379 of [1]).

The average bunch spacing is 994 ns at the ZH and ~3000 ns at the ttbar energy (page 283 of [1]).

The luminosity per bunch crossing at the FCC-ee is hence 7 x 10^28 cm^-2 at the ZH and 3 x 10^28 cm^-2 at the ttbar energy. The number of pileup events per bunch crossing is the product of these luminosities and the gg->hadronic cross section (which we get by running the card for the gg->hadronic production process in Pythia).

[1] https://link.springer.com/article/10.1140/epjst/e2019-900045-4

veelken commented 1 year ago

The gg->had background is mentioned on page 2 of the CLIC TDR [1]. The TDR mentions 3.2 gg->had events per bunch crossing at 3 TeV, but also mentions that one needs to sum over all bunch crossings (312) that occur within the same bunch train.

Page 4 of the CLIC TDR then mentions that the gg->had background that occurs in other bunch crossings can be reduced by about a factor 190 by using a combination of timing information, which is provided by the calorimeters as well as tracking detectors, and pT cuts.

For reference: The paper "Updated baseline for a staged compact linear collider" [2] has useful references in on the gg->had and also on the beamstrahlung background. The beamstrahlung background is due to photons , which are radiated off by the electron and positron beams when the beams pass the magnets that reduce the transverse beams size around the collision point. These photons then create e+ e- pairs that typically have low pT and are predominantly produced in the forward direction (at either theta=0 or 180 degrees).

[1] http://cds.cern.ch/record/1425915/files/CERN-2012-003.pdf [2] https://e-publishing.cern.ch/index.php/CYR/article/view/335/273

jpata commented 1 year ago

So the Pythia (and an alternative SLAC generator) cross-section of the gg->had process is available here [1], and comes to about ~550 nb.

A rough calculation for the number of gg->had events per bx gives 3 x 10^28 cm^-2 * 550 nb -> 0.0165 evs/bx

Being conservative, it would be 0.02 evs/bx * 312 bx/train -> ~6 gg->had events per signal.

What confuses me is that no CLIC study paper actually reports the specific number of overlay events used, as far as I've seen. But I think we could just ask e.g. the authors of [2] for more details on their configuration.

[1] https://cds.cern.ch/record/1443518/files/LCD-2011-020.pdf [2] https://cds.cern.ch/record/2843406/files/CLICdp-Pub-2022-002.pdf