centralpage

name	centralpage
description	Use when searching for ATLAS Monte Carlo or data samples: querying the PMG central sample database for MC datasets by process, generator, or campaign, looking up cross-sections and filter efficiencies, or identifying which centrally produced samples are available for a given physics process.

Central Page

Overview

The ATLAS Central Page is a tool for finding and querying Monte Carlo and data samples in the PMG (Physics Modelling Group) central sample database. It helps physicists locate the correct dataset containers for their analyses by searching through centrally produced sample catalogs, cross-referencing with AMI for dataset metadata.

When to Use

• Searching for MC samples by physics process, generator, or campaign
• Looking up cross-sections, k-factors, and filter efficiencies for MC samples
• Identifying which MC campaigns (mc20, mc21, mc23) have a given process
• Finding the correct dataset container name for a specific signal or background

Key Concepts

Concept	Notes
PMG central database	Authoritative catalog of centrally produced MC samples
MC campaign	Production campaign (mc20, mc21, mc23) with fixed config
Cross-section database	PMG-maintained xsec, k-factor, filter efficiency values
DSID	Dataset ID — unique 6-digit number per MC process config
Generator	Powheg, Sherpa, MadGraph, Pythia, Herwig, etc.

Canonical Patterns

The Central Page organises samples under PMG hashtags arranged in levels (PMGL1–PMGL4; level 3 = "Baseline"). The centralpage command navigates those levels positionally — there is no search subcommand.

Setup

setupATLAS
lsetup centralpage
voms-proxy-init --voms atlas     # listing/querying requires a valid grid proxy
centralpage --help

List available hashtags

A scope is required (centralpage -l alone errors with "Scope required"):

centralpage -s <scope> -l           # list hashtags for the scope
centralpage -s <scope> -P 3 -l      # only the level-3 (Baseline) hashtags

Browse samples by hashtag level

Pass hashtags positionally as L1 L2 L3 L4 to drill down to the samples:

centralpage -s <scope> <L1> <L2> <L3> <L4>

Add detail to the output:

• -m / --metadata — also print dataset metadata (cross-section, filter eff.)
• -a / --aod — also print AODs (standard tags) for each sample
• -p / --phys — also print DAOD_PHYS containers
• -L / --physlite — also print DAOD_PHYSLITE containers
• -f <fmt> / --outformat — restrict to a single output format
• -t / --table — emit a LaTeX table

Reverse lookup — find the hashtags for a known dataset

centralpage -d <dataset_name>

Verify cross-sections

The metadata shown with -m comes from the PMG database, but always re-check cross-sections, k-factors, and filter efficiencies against the latest PMG recommendations before using them in an analysis.

Gotchas

• Hashtag levels, not free text: the CLI navigates PMG hashtags positionally (L1 L2 L3 L4) or by -d <dataset>; there is no free-text search command. Use -s <scope> -l to discover the valid hashtags first.
• Grid credentials required: Full sample listing requires valid ATLAS grid credentials (voms-proxy-init --voms atlas).
• Campaign coverage varies: Not all MC campaigns have the same processes available — mc23 may lack samples that exist in mc20.
• Cross-section verification: Always verify cross-sections against the PMG cross-section database; values may be updated between production rounds.
• Dataset naming conventions change: Container naming patterns differ across MC campaigns — do not hardcode dataset name patterns.
• All ATLAS energy/momentum values are in MeV: Cross-sections are in pb, but kinematic values follow ATLAS MeV conventions.

Interop

• ami-mcp: The AMI MCP server (ami-mcp) provides LLM-friendly access to the same PMG cross-section database and dataset metadata that centralpage queries. Prefer ami-mcp for programmatic lookups of cross-sections, k-factors, filter efficiencies, and dataset tags within a Claude session — it returns structured results without requiring lsetup. Requires ~/.globus credentials and a valid VOMS proxy (voms-proxy-init --voms atlas).
• Rucio: Datasets found via centralpage can be located and downloaded with Rucio — Rucio has a dedicated MCP server (rucio-mcp).
• atlasopenmagic: The ATLAS Open Magic MCP server can also search for datasets and provides complementary search capabilities.
• setupATLAS: lsetup centralpage requires a working ALRB environment — see the setupatlas skill.

Contact: See PMG infrastructure team

Docs

https://gitlab.cern.ch/atlas-physics/pmg/infrastructure/central-page