Data Samples¶

ATLAS data comes in many flavors. Because the complete data is so large, ATLAS makes a number of different samples that are slimmed down in various ways. The consequences of this can be far reaching: if a derivation removes attributes necessary to perform a calibration for muons, then accessing calibrated muons will not be possible downstream.

In general, we’ll be accessing DAOD_PHYS datasets (more information about them can be found here). These are complete, with all objects, though they have some rather severe object cuts that makes them useless for a number of physics final states (for example, no tracks less than \(p_T\) 10 GeV are included).

R21 (Run 2)¶

For Run 2, software release R21, we use the following datasets:

Name	Events	Rucio Dataset Name
`ds_Zee`	69,734,200	mc16_13TeV.361106.PowhegPythia8EvtGen_AZNLOCTEQ6L1_Zee.deriv.DAOD_PHYS.e3601_e5984_s3126_s3136_r10724_r10726_p4164
`ds_Zmumu`	292,915,100	mc16_13TeV.361107.PowhegPythia8EvtGen_AZNLOCTEQ6L1_Zmumu.deriv.DAOD_PHYS.e3601_e5984_s3126_s3136_r10724_r10726_p4164
`ds_Ztautau`		mc16_13TeV.361108.PowhegPythia8EvtGen_AZNLOCTEQ6L1_Ztautau.deriv.DAOD_PHYS.e3601_e5984_s3126_s3136_r10724_r10726_p4164
`ds_bphys`		mc16.999031.P8BEG_23lo_ggX18p4_Upsilon1Smumu_4mu_3pt2.deriv.DAOD_BPHY4.e8304_a875_r10724_r10726_p3712_pUM999999
`ds_jz2_exot15`	15989500	mc16_13TeV.361022.Pythia8EvtGen_A14NNPDF23LO_jetjet_JZ2W.deriv.DAOD_EXOT15.e3668_s3126_r9364_r9315_p4696

You can find the datasets like ds_Zee defined in the config.py file. If running against a servicex backend, they are defined as:

from func_adl_servicex_xaodr21 import SXDSAtlasxAODR21

ds = SXDSAtlasxAODR21(rucio_dataset_name, backend="xaod")

And if running on a local file, then the following definition is made:

from func_adl_servicex import SXLocalxAOD
from func_adl_servicex_xaodr21.event_collection import Event

class xAODLocalTyped(SXLocalxAOD[Event]):
    def __init__(self, file_path: Path):
        super().__init__(file_path, item_type=Event)

ds = xAODLocalTyped(local_path_of_dataset_file)

Note the use of the class Event: this defines the types and metadata for the xAOD backend and allows you to access all the various methods and types without having to resort to heuristics (which work for simple cases, but not complex ones).

The ds_jz2_exot15 is a special case - that isn’t a DAOD_PHYS file. It is used to demonstrate some access patterns that are only possible in derivations. DAOD_PHYS has had some of the information removed to keep its size under control. Its declaration is as follows:

from func_adl_servicex_xaodr21 import calib_tools

ds_jz2_exot15 = calib_tools.query_update(
    base_jz2_exot15,
    jet_collection="AntiKt4EMTopoJets",
    jet_calib_truth_collection="AntiKt4TruthJets",
    perform_overlap_removal=False,
)

Here, base_jz2_exot15 is the dataset defined in the normal way above (as running on a local file or a servicex backend). for more details about using calib_tools see the calibration chapter.

The datasets were picked on advice from experts in ATLAS and then using the AMI dataset finder, which is a very useful way to scan for datasets you already know must exist.

Using the ATLAS xAOD ServiceX Backend

Data Samples

Contents

Data Samples¶

R21 (Run 2)¶