Abstract:The particle-flow (PF) algorithm, which infers particles based on tracks and calorimeter clusters, is of central importance to event reconstruction in the CMS experiment at the CERN LHC, and has been a focus of development in light of planned Phase-2 running conditions with an increased pileup and detector granularity. In recent years, the machine learned particle-flow (MLPF) algorithm, a graph neural network that performs PF reconstruction, has been explored in CMS, with the possible advantages of directly optimizing for the physical quantities of interest, being highly reconfigurable to new conditions, and being a natural fit for deployment to heterogeneous accelerators. We discuss progress in CMS towards an improved implementation of the MLPF reconstruction, now optimized using generator/simulation-level particle information as the target for the first time. This paves the way to potentially improving the detector response in terms of physical quantities of interest. We describe the simulation-based training target, progress and studies on event-based loss terms, details on the model hyperparameter tuning, as well as physics validation with respect to the current PF algorithm in terms of high-level physical quantities such as the jet and missing transverse momentum resolutions. We find that the MLPF algorithm, trained on a generator/simulator level particle information for the first time, results in broadly compatible particle and jet reconstruction performance with the baseline PF, setting the stage for improving the physics performance by additional training statistics and model tuning.

Modular Implementation of Particle Flow Algorithm with Minimized Dependence on the Detector Geometry

Iowa Particle Flow Algorithm

Detector Optimization for SiD using PFA

A Handy Tool for History Keeping of Geant4 Tracks and its Application to Studies of Fundamental Limits on PFA Performance

Tests of a particle flow algorithm with CALICE test beam data

Development of Particle Flow Calorimetry

Particle Flow Calorimetry

The Pandora Particle Flow Algorithm

Particle Flow Calorimetry and the PandoraPFA Algorithm

Progress towards an improved particle flow algorithm at CMS with machine learning

CPID: A Comprehensive Particle Identification Framework for Future e$^+$e$^-$ Colliders

Improved particle-flow event reconstruction with scalable neural networks for current and future particle detectors

Time-based Reconstruction of Hyperons at PANDA at FAIR

Charged Particle Tracking in Real-Time Using a Full-Mesh Data Delivery Architecture and Associative Memory Techniques

Parnassus: An Automated Approach to Accurate, Precise, and Fast Detector Simulation and Reconstruction

PFA Oriented ECAL Optimization for the CEPC

HGPflow: Extending Hypergraph Particle Flow to Collider Event Reconstruction

Fast evaluation of heavy-quark contributions to DIS in APFEL++

Implementation of Particle Flow Algorithm and Muon Identification

A C++ program for estimating detector sensitivities to long-lived particles: displaced decay counter