fast-vertex-quality-inference 1.0.14

Inference code for fast generation of vertex quality variables

Fast Vertexing Variables at LHCb - Inference Library

Full documentation is avalibale here.

Description

This tool provides a quick approximation of the LHCb reconstruction process. Built on top of RapidSim, the run function automatically communicates with RapidSim to generate kinematic information which is then smeared before predictions of high-level vertexing variables are generated.

The software utilizes Variational Autoencoders (VAEs) to estimate these variables. These VAEs are trained on output from the LHCb simulation software.

Generated tuples can be integrated with other tools such as TriggerCalib and PIDCalib2, completing the full chain of estimation of reconstruction efficiencies and mass shapes for background studies at LHCb.

Disclaimer

This tool is not designed to replace the full simulation software. It is designed to quickly return reasonable estimates of mass shapes and efficiencies.

Environment Setup

RaidSim is required to use the full functionalty of this library. The environment variables $RAPIDSIM_ROOT and $EVTGEN_ROOT that are used by the code to access the install.

• RAPIDSIM_ROOT: The root directory for RapidSim.
• EVTGEN_ROOT (optional): The root directory for EVTGEN, if applicable.

Example Usage

run()

The run() function is the primary method to execute FastVertexing. It handles several key operations in the vertexing and event simulation process:

from fast_vertex_quality_inference import run

run(
    events=1000,
    decay="B+ -> { D0b -> K+ e- anti-nue } pi+",
    naming_scheme="B_plus -> { NA -> K_plus e_minus NA } e_plus",
    decay_models="PHSP -> { ISGW2 -> PHSP PHSP PHSP } PHSP",
    mass_hypotheses={"e_plus": "e+"},
    intermediate_particle={"Jpsi": ["e_minus", "e_plus"]},
)

run_from_tuple()

The run_from_tuple() function only executes the vertexing network on an existing tuple and can be used without a RapidSim installation.

from fast_vertex_quality_inference import run_from_tuple

run_from_tuple(
    file="decay_tree.root",
    mother_particle="MOTHER",
    daughter_particles=["DAUGHTER1", "DAUGHTER2", "DAUGHTER3"],
    fully_reco=False,
    nPositive_missing_particles=0,
    nNegative_missing_particles=0,
    mass_hypotheses={"DAUGHTER2": "e+"},
    intermediate_particle={"INTERMEDIATE": ["DAUGHTER2", "DAUGHTER3"]},
    branch_naming_structure={"true_momenta_component": "{particle}_TRUE_P{dim}"},
)