MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks

Abstract

In general-purpose particle detectors, the particle flow algorithm may be used to reconstruct a coherent particle-level view of the event by combining information from the calorimeters and the trackers, significantly improving the detector resolution for jets and the missing transverse momentum. In view of the planned high-luminosity upgrade of the CERN Large Hadron Collider, it is necessary to revisit existing reconstruction algorithms and ensure that both the physics and computational performance are sufficient in a high-pileup environment. Recent developments in machine learning may offer a prospect for efficient event reconstruction based on parametric models. We introduce MLPF, an end-to-end trainable machine-learned particle flow algorithm for reconstructing particle flow candidates based on parallelizable, computationally efficient, scalable graph neural networks and a multi-task objective. We report the physics and computational performance of the MLPF algorithm on on a synthetic dataset of tt events in HL-LHC running conditions, including the simulation of multiple interaction effects, and discuss potential next steps and considerations towards ML-based reconstruction in a general-purpose particle detector.

Unleashing the full power of LHCb to probe stealth new physics

“…Novel jet clustering techniques can also be used to face the run 3 environment. For example clustering algorithms based on graph neural networks are considered promising [44,45]. These algorithms can be trained on specific channels (i.e.…”

Point Cloud Transformers applied to Collider Physics

“…Neural network architectures that treat collision events as point clouds have recently grown in number given their state-of-the-art performance when applied to different collider physics problems. A few examples of such applications are jet-tagging [7,8], secondary vertex finding [9], event reconstruction [10][11][12][13], and jet parton assignment [14]. A comprehensive review of the different methods is described in [15].…”

Learning Uncertainties the Frequentist Way: Calibration and Correlation in High Energy Physics

“…This inference task can be challenging when the reconstruction requires high-dimensional inputs. Machine learning (ML) is a natural tool for performing high-dimensional reconstruction, and there has been significant progress in utilizing ML method for estimating the energies of various objects, including photons [1], muons [2], single hadrons [3][4][5][6][7][8], and sprays of hadrons (jets) [9][10][11][12][13][14][15][16][17][18][19] at colliders; kinematic reconstruction in deep inelastic scattering [20,21]; and neutrino energies in a variety of experiments [22][23][24][25][26][27]. Further ideas can be found at Ref.…”

“…The Gaussian Ansatz enables an elegant strategy to extract Eqs. ( 7) and (8). Since the optimal T (x, z) is bounded from above, we can take D(x) to be everywhere zero without loss of expressivity.…”