Running a minimal DaVinci job locally

Objectives

Learn what DaVinci is and how to run it locally!

DaVinci is the offline analysis software for LHCb. In a nutshell, we use it to process the DST files that come out of the sprucing and produce the ROOT ntuples, ready for further analysis. When running over the full output of the sprucing you will use analysis productions to submit jobs to the grid. However, to set up these large productions you need to be able to run DaVinci interactively to develop your steering options for the productions, develop new tools, or implement and improve selections.

Reminder - released applications and `lb-run`

The LHCb applications are released and then installed on CVMFS. They are therefore available to use in the standard LHCb linux environment. To use an application you must first setup its environment using the lb-run command, which starts the environment and then executes the subsequent command. For example, running lb-run DaVinci/v65r0 python starts a python shell with all the modules of DaVinci readily available in the environment. This is useful if you want to explore Functor definitions (these will be explained more in detail in the next lesson):

import Functors as F
help(F.BPVLTIME)

gives you the docstring of the functor that obtains the particle decay time, with respect to the "best" PV.

Running DaVinci with `lbexec`

For Run 3 the software can be run with the lbexec command. This is a replacement for gaudirun (which is still available if desired). A typical DaVinci execution looks like:

lb-run DaVinci/v65r0 lbexec mytuplemaker:main myoptions.yaml

The first two parts start the DaVinci environment. mytuplemaker refers to a python file called mytuplemaker.py which includes a defined function called main (the naming of the file and the function in it which is executed is arbitrary). This is also explained in the Introduction to LHCb software lesson This python file is where the FunTuple is initiated and all the branches and their functors are listed. The options.yaml is a markdown file with the typical steering options - it includes the data to be run over, the number of events, conditions tags etc.

The benefit of this setup over the previous gaudiruny.py is the separation of the data steering from the tuple making - allowing for much easier application of the tupling to a variety of disparate data files.

A minimal DaVinci job

Let us create a minimal job to print the decay trees stored in a DST. In a file mymodule.py put the following:

from DaVinci import Options, make_config
from DaVinci.algorithms import create_lines_filter
from PyConf.Algorithms import PrintDecayTree
from PyConf.reading import get_particles

def printDecayTree(options: Options):
    # The name of the line we want to look at
    line = "Hlt2B2OC_BdToDsmPi_DsmToKpKmPim"
    input_data = get_particles(f"/Event/HLT2/{line}/Particles")

    # Add a filter: We are not really filtering over particles
    # If the event hasn't fired a HLT2 line then the corresponding TES location does
    # obviously not exist and therefore if any algorithm tries to look for this location,
    # it will of course fail.
    # Resolve this with a filter, where:
    # - 1st argument is a user defined name.
    # - 2nd argument is the line decision (inspect hlt2_starterkit.tck.json if needed))
    my_filter = create_lines_filter("HDRFilter_SeeNoEvil", lines=[f"{line}"])

    # Defining an algorithm. The algorithm here prints the decaytree
    pdt = PrintDecayTree(name="PrintBsToDsPi", Input=input_data)

    user_algorithms = [my_filter, pdt]

    return make_config(options, user_algorithms)

Now in a separate file we can put the steering options in a file called options.yaml:

input_files:
- root://eoslhcb.cern.ch//eos/lhcb/wg/dpa/wp7/Run3SK/exampleDST/00257716_00000011_1.hlt2.dst
input_type: ROOT
evt_max: 100
ntuple_file: davinci_ntuple.root
input_process: Hlt2
print_freq: 1
simulation: true
conddb_tag: sim10-2024.W37.39-v00.00-md100
dddb_tag: dddb-20240427

You can see that because we are running on simulation, then we have to set the appropriate conddb_tag and dddb_tag. We also need to set the appropriate process that produced the input file. This will normally be Spruce when running over spruced data. Now run the example with

lb-run DaVinci/v65r0 lbexec mymodule:printDecayTree options.yaml

The output of this example is the printing out of the decay trees stored at the TES location /Event/HLT2/Hlt2B2OC_BdToDsmPi_DsmToKpKmPim/Particles:

PrintBsToJpsiPhi.PrintDecayTreeTool    INFO <-------------------------------------------- Particle -------------------------------------------->
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO                 Name         E         M         P        Pt       phi        Vz    P(C/K)   PP(C/K)
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO                            MeV       MeV       MeV       MeV      mrad        mm
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO B~0                  77456.12   5350.44  77271.10  11006.56  -2894.20    -13.07      36/0       N/A
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO +-->D_s+             61360.38   1964.95  61328.91   9732.58  -3078.29     -6.67      37/0       N/A
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO |+-->K+              34549.66    493.68  34546.14   6104.16  -3091.80    -13.69     27/41      3/41
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO |+-->K-              16242.15    493.68  16234.65   2472.72  -3074.02    -15.05     27/40      3/40
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO |+-->pi+             10571.58    139.57  10570.65   1159.77  -3015.92    -17.99      5/16      3/16
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO +-->pi-              16100.04    139.57  16099.44   2289.99  -2002.85    -14.64      5/48      3/48
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO Used TES locations :-
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO      37 = '/Event/DoNotUseExplicitly/AnonymousDependencyOnly/SourceID_2/0xffffffe2'
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO       5 = '/Event/DoNotUseExplicitly/AnonymousDependencyOnly/SourceID_2/0xffffffe7'
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO      27 = '/Event/DoNotUseExplicitly/AnonymousDependencyOnly/SourceID_2/0xffffffe8'
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO      36 = '/Event/DoNotUseExplicitly/AnonymousDependencyOnly/SourceID_2/0xfffffffb'
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO       3 = '/Event/HLT2/Rec/ProtoP/Long'
PrintBsToJpsiPhi.PrintDecayTreeTool    INFO

Running a minimal DaVinci job locally

Reminder - released applications and lb-run

Running DaVinci with lbexec

A minimal DaVinci job

Reminder - released applications and `lb-run`

Running DaVinci with `lbexec`