New term suggestions (minimum-bias streams)

[ghost]

The following list of terms should be defined:

• [ ] NoBias
• [ ] MicroBias
• [ ] FULL.DST

Is NoBias a good source of minimum-bias data? Or, is that just MiniBias? How much minimum-bias data is typically collected per run/fill? How is MicroBias different from MiniBias? What should I think when I see FULL.DST or RAW.DST in the bookkeeping?

[alexpearce]

I'm not sure if you're planning on answering this one yourself, but here's some information either way that might help. I've only heard the 'MinBias' abbreviation used on LHCb rather than 'MiniBias', so I'll use that.

---

Each of {No,Min,Micro}Bias exists to provide analysts with data that is as unbiased as possible, in the sense that the set of triggers feeding into the datasets have very few selection requirements, or none at all.

• NoBias: No selection requirements at all, other than requiring a bunch-bunch crossing. These are selected randomly by ODIN and pass through L0, HLT1, and HLT2 without any additional requirements. Usually taken with a large prescale such that the output rate of NoBias events out of HLT2 is low, on the order of 1--10 Hz. Sometimes the rate is increased significantly out of L0 for special data-taking runs. The rate is controlled from a panel in the run control system.
• MinBias and MicroBias: Both meaning roughly the same thing, the selection requirements imposed on these datasets in the trigger are 'minimal'. Examples include requiring that at least one track was reconstructed in the VELO, or that the number of hits in the SPD are above some noise-like threshold.

As always when talking about selections, you have to be careful what stage of the trigger you're talking about. Someone may be discussing using 'NoBias data' but only in the sense that it is unbiased out of L0, whereas another person may mean it to be unbiased out of HLT2.

Hopefully this answers the questions, but to be explicit:

• Is NoBias a good source of minimum-bias data? Or, is that just MiniBias?
  • Depends what 'good' means. NoBias datasets are generally pretty small, but they are as unbiased as you can get.
• How much minimum-bias data is typically collected per run/fill?
  • The acceptance rate of NoBias during normal physics running is typically on the order of 1--10 Hz.
• How is MicroBias different from MiniBias?
  • I would interpret both terms to mean the same thing: there is a very minimal set of requirements in the triggers that selected the data. What exactly these requirements are is defined in the relevant set of trigger lines (which is analysis-dependent), and these may include either 'Min' or 'Micro' in their names.
• What should I think when I see FULL.DST or RAW.DST in the bookkeeping?
  • This is largely unrelated to the *Bias discussion, relating more to how data is partitioned, or 'streamed' in the jargon, from the trigger. In summary:
    • The RAW files (I don't think RAW.DST exists) contain the raw data directly from the trigger. What events are in these files depends on the stream you're looking at. Events in which a NoBias trigger fired are sent to the NoBias stream. Most 'physics triggers' are sent to either of the Full or Turbo streams.
    • The FULL.DST files contain the output of the Brunel processing of the raw data.

[ghost]

Thank you: that is more or less what I wanted to know. I have tried to draft a pull request.

[eduardo-rodrigues]

You should not close this until all is sorted out with an accepted pull request - at all not the case. Reopening.

[henryiii]

@jordan-rothenheimer If you say something like closes #30 in your PR description, the issue automatically gets closed by GitHub when the PR is merged.

[eduardo-rodrigues]

(Removed MiniBias from the list.)