DSProd

DSProd is a small LAW-based tool for custom CMS Monte Carlo production inside the FLAF ecosystem. It takes a physics process from a gridpack all the way to analysis-ready NanoAOD, running the standard CMSSW generation chain for the relevant Run3 era, and stages the output where FLAF can pick it up directly.

It exists for samples that are not available in central CMS production — private signal points, extra mass scans, alternative generator settings — but which still have to match the central per-era conditions (global tags, pileup, CMSSW releases) so they can be analysed side by side with the central samples.

Scope

DSProd is deliberately much smaller than FLAF. It is one linear production chain (gridpack → NanoAOD) with a handful of tasks, generic over the physics process. Everything process-specific lives in a small process customization module; the tasks themselves stay generic.

The big picture

A production is described by a single production setup YAML (which process, which eras, which NanoAOD versions, which points). LAW then builds only the steps needed to produce the requested output:

flowchart TD
    IG[ImportGridpack<br/>gridpack from the store] --> RP[RunProd<br/>fused GEN→…→MiniAOD→NanoAOD per seed]
    GP[MakeGridpack<br/>generate what the store lacks] --> RP
    RP --> NM[NanoMergeTask<br/>hadd per-seed nanos into groups]
    NM --> FLAF[FLAF private-nano input]
    IC[InstallCMSSW<br/>per-era CMSSW releases] -.provides.-> RP

Each box is a LAW task. You normally run only the last task you care about — LAW pulls in everything upstream automatically. Every task can run on three backends: local, HTCondor (CERN batch), or CRAB (the WLCG grid, for large productions).

How the output feeds FLAF

NanoMergeTask writes merged NanoAOD following the private-nano (HLepRare) convention that FLAF already understands:

<fs_default>/<output>/nanoAOD_<version>/<era>/<process-point>/nano_<version>_<n>.root

So once a production finishes, its samples are consumed by FLAF as ordinary NanoAOD inputs — no extra bookkeeping step.

Where to go next

• Installation — clone the repo and source the environment.
• Your first production — run the bundled test setup end to end.
• Architecture — the production chain and how the pieces fit.
• Production setups — the YAML that describes a production.
• Contributing — code style and the formatting check.