analyzer.modules.singlestop.nn_reco

Classes

NNMassReco	Reconstruct top quark and neutralino masses using a Neural Network.
NNMassPlots	Create histograms for reconstructed mass variables.

Module Contents

class analyzer.modules.singlestop.nn_reco.NNMassReco

Bases: analyzer.core.analysis_modules.AnalyzerModule

Reconstruct top quark and neutralino masses using a Neural Network.

This module uses a trained PyTorch model to resolve jet combinatorics and reconstruct the mass of the top quark ($m_{tilde{t}}$) and neutralino ($m_{chi}$).

Parameters

input_colColumn

Input column containing the jet collection (e.g. GoodJet).

m3_outputColumn

Output column name for the reconstructed neutralino mass ($m_{chi}$).

m4_outputColumn

Output column name for the reconstructed top quark mass ($m_{tilde{t}}$).

model_pathstr

Path to the trained PyTorch model file (.pt).

scaler_pathstr

Path to the scaler file (.pkl) used for input feature normalization.

input_col: analyzer.core.columns.Column

m3_output: analyzer.core.columns.Column

m4_output: analyzer.core.columns.Column

model_path: str

scaler_path: str

inputs(metadata)

outputs(metadata)

neededResources(metadata)

run(columns, params)

class analyzer.modules.singlestop.nn_reco.NNMassPlots

Bases: analyzer.core.analysis_modules.AnalyzerModule

Create histograms for reconstructed mass variables.

Generates 1D and 2D histograms for the reconstructed top squark and chargino masses, as well as their ratio.

Parameters

m3_inputColumn

Column containing the reconstructed chargino mass ($m_{chi}$).

m4_inputColumn

Column containing the reconstructed top squark mass ($m_{tilde{t}}$).

prefixstr

Prefix required for all generated histograms to ensure uniqueness.

m3_input: analyzer.core.columns.Column

m4_input: analyzer.core.columns.Column

prefix: str

m4_range: tuple[int, int] = (0, 3000)

m3_range: tuple[int, int] = (0, 3000)

ratio_range: tuple[float, float] = (0.0, 1.0)

ratio_only: bool = False

bins: int = 60

y_bins: int | None = None

run(columns, params)

outputs(metadata)

inputs(metadata)