The ATLASWatchMan D3PD Model

Introduction

The ATLASWatchMan package dynamically generates the Analysis Code. Running this generated code (locally or on the Grid) produces as output a D3PD file, the ATLAS name for a plain ROOT file, containing only ROOT TTree and/or histograms.

As you might have seen in previous pages, ATLASWatchMan is an automated Analysis Code Generator, that accepts as inputs a file defining as many channels/analyses as the user wants (the so called Steering File) and it returns a plain, clear analysis code, written in Python, and ready to be run.

Once the generated analsyis code is run, locally or on the Grid, the output is a D3PD file, containing one single ROOT TTree.

This TTree contains all the outputs from all the analyses the user defines in the Steering File.

Information is stored in a clever way, to store all the quantities needed, but saving as much space as possible. This in order to have a lightweight D3PD .

This intent is reached setting flags to events, particles and other quantities stored in the TTree; from those flags the user can know, and he/she can select, those events belonging to a specific channel/analysis, and those particles/quantities passing a specific object selection or overlap removal cut.

Content of the ATLASWatchMan D3PD

If you open the ATLASWatchMan D3PD with the ROOT TBrowser (see note (2) here)

Fig. 1 An example of a minimal ATLASWatchMan D3PD content: "meff" and "SphCSC" are user-defined quantities.

Fig. 2 How ATLASWatchMan give a flag to events having passed all the cuts of one analysis/channel

Fig. 3 How ATLASWatchMan give a flag to objects having passed all the object selection and overlap removal cuts that the user defined in a specific analysis/channel.

How to read your ATLASWatchMan D3PD, the output of your analsys!

...more info coming soon...

-- RiccardoMariaBianchi - 28 Jan 2009