Neuromorphic Readout for Hadron Calorimeters

TL;DR

This work investigates neuromorphic readout for hadron calorimeters by simulating a PbWO$_4$ calorimeter and processing its temporal light signals with a spiking neural network to estimate total deposited energy, shower centroid, and dispersions. It defines a spike-based encoding, a two-layer SNN, and rate-based decoding to perform single- and multi-target regression, achieving sub-cell centroid precision and reasonable energy/dispersion predictions while highlighting the role of architecture optimization. The study demonstrates that topological information about showers can be extracted without heavy calorimeter segmentation, offering a path toward real-time, low-power online reconstruction. A nanophotonic III-V nanowire hardware proposal is presented to enable ultra-fast, energy-efficient spike processing within the detector, suggesting a practical route to implement such neuromorphic readouts in future high-energy physics experiments.

Abstract

We simulate hadrons impinging on a homogeneous lead-tungstate (PbWO4) calorimeter to investigate how the resulting light yield and its temporal structure, as detected by an array of light-sensitive sensors, can be processed by a neuromorphic computing system. Our model encodes temporal photon distributions as spike trains and employs a fully connected spiking neural network to estimate the total deposited energy, as well as the position and spatial distribution of the light emissions within the sensitive material. The extracted primitives offer valuable topological information about the shower development in the material, achieved without requiring a segmentation of the active medium. A potential nanophotonic implementation using III-V semiconductor nanowires is discussed. It can be both fast and energy efficient.

Figures (11)

• Figure S1: Distribution of the $log\text{(}E/MeV)$ variable across all the dataset.
• Figure S2: Graphical representation of the encoding scheme.
• Figure S3: Data processing pipeline and network architecture.
• Figure S4: Output of the $log(E/MeV)$ network, and the respective results for the true value of the energy. The plots show the correlation between targets and predictions (above) and the residuals (below).
• Figure S5: Output of the energy centroid network, for the $x$, $y$ and $z$ coordinates. The plots show the correlation between targets and predictions, (above) and the residuals (below).