A Differentiable Neural Surrogate for Photon Propagation in Neutrino Telescopes

Felix J. Yu; Berthy T. Feng; Nicholas Kamp; Carlos A. Argüelles

astro-ph.HESep 4, 2026

A Differentiable Neural Surrogate for Photon Propagation in Neutrino Telescopes

Authors: Felix J. Yu, Berthy T. Feng, Nicholas Kamp, Carlos A. Argüelles

Organizations: Harvard University · NSF IAIFI · Massachusetts Institute of Technology

Abstract

Large-volume neutrino telescopes infer neutrino properties from Cherenkov light, but simulating the transport of billions of photons through highly scattering ice or water is computationally costly. We introduce candela, a differentiable SIREN neural field that learns the photon Green's function of the IceCube Neutrino Observatory, a cubic-kilometer detector embedded in Antarctic glacial ice. Given a point-like energy deposit and sensor, it predicts the expected photon yield and full arrival-time distribution at the sensor. Complete events are simulated by decomposing charged-particle energy deposits into point-like sources and superposing their predicted sensor responses. Trained on Monte-Carlo simulations, candela generates events $50$ -- $100\times$ faster than existing methods, with cost scaling only weakly with neutrino energy. It keeps median yields within $2\%$ of the MC expectation and timing distributions at the MC statistical floor across six photon-count decades. The model also provides end-to-end gradients with respect to event parameters and opens a path toward optimizing scattering-medium properties, which often dominate systematic uncertainties in neutrino telescopes.

Figures & tables

Figure 1: The candela event-generation pipeline. (a) Decompose a muon into point-like energy deposits. (b) Encode each source–receiver query by track-frame geometry

(R,\Phi,l)

, source direction and energy, and absolute positions. (c) (c) Predict per-GeV yield

\Lambda

and timing density

\pi(\tau)

for each query using one SIREN. (d) Scale each source response by its deposited energy and sum over sources to obtain the expected photon rate at every sensor, defining the complete event.

Figure 2: Event-level agreement for a $100\,\mathrm{TeV}$ $\nu_{\mu}$ charged-current event. Top: PPC (left) and candela (right) realizations of identical energy losses. The size and color encode photon count and first-hit time. Bottom: candela timing densities versus high-statistics PPC histograms at saturated, near-, and far-track sensors. Agreement spans both the prompt peak(s) and scattering tails.

Figure 3: Speed and fidelity on independent test events. (a) Event-generation wall time for matched energy losses. (b) candela photon-yield prediction relative to the high-statistics PPC expectation. (c) KS distance to high-statistics PPC timing. Lines and bands give bin medians and 16th–84th percentiles over events (a) or sensors (b,c) ; PPC–PPC is the MC floor.

Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 4: Performance of candela through a sweep of

\epsilon_{\mathrm{LC}}

values.

Figure 5: Demonstration of deposited energy reconstruction using the fully differentiable candela .

Explore similar work

End-to-end Differentiable Calibration and Reconstruction for Optical Particle Detectors
Feb 27, 2026Omar Alterkait, César Jesús-Valls, Ryo Matsumoto +2Differentiable Physics Differentiable Optimization
Department of Physics and Astronomy, Tufts University, 574 Boston Ave, Medford, MA 02155, USA · The NSF AI Institute for Fundamental Interactions, MIT, Cambridge, MA 02139, USA · European Organization for Nuclear Research (CERN), 1211 Geneva 23, Switzerland +3
Neural posterior estimation of the neutrino direction in IceCube using transformer-encoded normalizing flows on the sphere
Apr 21, 2026R. Abbasi, M. Ackermann, J. Adams +411Normalizing Flows Transformer
Suppression of $^{14}\mathrm{C}$ photon hits in large liquid scintillator detectors via spatiotemporal deep learning
Mar 29, 2026Junle Li, Zhaoxiang Wu, Guanda Gong +5High-Energy Physics
School of Aeronautics and Astronautics, Zhejiang University, Hangzhou 310027, China · Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China · School of Physical Sciences, University of Chinese Academy of Science, Beijing 100049, China +1

Feb 27, 2026J/K move · Enter open · S save

End-to-end Differentiable Calibration and Reconstruction for Optical Particle Detectors

Omar Alterkait, César Jesús-Valls, Ryo Matsumoto, Patrick de Perio+1

Department of Physics and Astronomy, Tufts University, 574 Boston Ave, Medford, MA 02155, USA · The NSF AI Institute for Fundamental Interactions, MIT, Cambridge, MA 02139, USA · European Organization for Nuclear Research (CERN), 1211 Geneva 23, Switzerland+3

Differentiable Physics Differentiable Optimization

Open paper AskPDF arXiv

Large-scale homogeneous detectors with optical readouts are widely used in particle detection, with Cherenkov and scintillator neutrino detectors as prominent examples. Analyses in experimental physics rely on high-fidelity simulators to translate sensor-level information into physical quantities of interest. This task critically depends on accurate calibration, which aligns simulation behavior with real detector data, and on tracking, which infers particle properties from optical signals. We present the first end-to-end differentiable optical particle detector simulator, enabling simultaneous calibration and reconstruction through gradient-based optimization. Our approach unifies simulation, calibration, and tracking, which are traditionally treated as separate problems, within a single differentiable framework. We demonstrate that it achieves smooth and physically meaningful gradients across all key stages of light generation, propagation, and detection while maintaining computational efficiency. We show that gradient-based calibration and reconstruction greatly simplify existing analysis pipelines while matching or surpassing the performance of conventional non-differentiable methods in both accuracy and speed. Moreover, the framework's modularity allows straightforward adaptation to diverse detector geometries and target materials, providing a flexible foundation for experiment design and optimization. The results demonstrate the readiness of this technique for adoption in current and future optical detector experiments, establishing a new paradigm for simulation and reconstruction in particle physics.

Abstract

Explore similar work

End-to-end Differentiable Calibration and Reconstruction for Optical Particle Detectors

Neural posterior estimation of the neutrino direction in IceCube using transformer-encoded normalizing flows on the sphere

Suppression of 14C^{14}\mathrm{C}14C photon hits in large liquid scintillator detectors via spatiotemporal deep learning

Suppression of $^{14}\mathrm{C}$ photon hits in large liquid scintillator detectors via spatiotemporal deep learning