arXiv subjects

Patrick Huber

Publications and source records attributed to Patrick Huber.

At least 19 recordsLinked to original sources

arXiv2026

Calorimetric approach to paleo-detection of dark matter

We present the first paleo-detector dark matter sensitivity analysis based on a calorimetric readout, in which the number of stable lattice vacancies produced by each nuclear recoil is used as a per-event observable complementary to the track length. Using full-cascade SRIM simulations in olivine, we compute the expected sensitivity for a 100 gGyr exposure. We find that a vacancy-only readout reaches a sensitivity envelope very similar to that of state-of-the-art track-only analyses. The combination of the two observables provides an event-by-event proxy for |dE/dx| and hence for the recoiling nuclear species. Since the neutron-nucleus cross section is approximately flat in nuclear mass while the dark-matter--nucleus cross section scales as $A^2$, this discrimination suppresses the dominant neutron background by more than an order of magnitude at moderate dark matter masses. The combined-analysis sensitivity reaches spin-independent dark-matter--nucleon cross sections of order $10^{-48}\,\mathrm{cm}^2$ at WIMP masses of a few tens of GeV, comparable to future direct detection experiments. A two-stage readout combining selective-plane illumination microscopy with scanning electron microscopy is identified as a path to making a 100 g-scale analysis plausible.

hep-ph↗

arXiv2026

Short Data, Long Context: Distilling Positional Knowledge in Transformers

Extending the context window of language models typically requires expensive long-context pre-training, posing significant challenges for both training efficiency and data collection. In this paper, we present evidence that long-context retrieval capabilities can be transferred to student models through logit-based knowledge distillation, even when training exclusively on packed short-context samples within a long-context window. We provide comprehensive insights through the lens of Rotary Position Embedding (RoPE) and establish three key findings. First, consistent with prior work, we show that phase-wise RoPE scaling, which maximizes rotational spectrum utilization at each training stage, also achieves the best long-context performance in knowledge distillation setups. Second, we demonstrate that logit-based knowledge distillation can directly enable positional information transfer. Using an experimental setup with packed repeated token sequences, we trace the propagation of positional perturbations from query and key vectors through successive transformer layers to output logits, revealing that positional information systematically influences the teacher's output distribution and, in turn, the distillation signal received by the student model. Third, our analysis uncovers structured update patterns in the query state during long-context extension, with distinct parameter spans exhibiting strong sensitivity to long-context training.

cs.CL↗

arXiv2026

MobileLLM-Flash: Latency-Guided On-Device LLM Design for Industry Scale Deployment

Real-time AI experiences call for on-device large language models (OD-LLMs) optimized for efficient deployment on resource-constrained hardware. The most useful OD-LLMs produce near-real-time responses and exhibit broad hardware compatibility, maximizing user reach. We present a methodology for designing such models using hardware-in-the-loop architecture search under mobile latency constraints. This system is amenable to industry-scale deployment: it generates models deployable without custom kernels and compatible with standard mobile runtimes like Executorch. Our methodology avoids specialized attention mechanisms and instead uses attention skipping for long-context acceleration. Our approach jointly optimizes model architecture (layers, dimensions) and attention pattern. To efficiently evaluate candidates, we treat each as a pruned version of a pretrained backbone with inherited weights, thereby achieving high accuracy with minimal continued pretraining. We leverage the low cost of latency evaluation in a staged process: learning an accurate latency model first, then searching for the Pareto-frontier across latency and quality. This yields MobileLLM-Flash, a family of foundation models (350M, 650M, 1.4B) for efficient on-device use with strong capabilities, supporting up to 8k context length. MobileLLM-Flash delivers up to 1.8x and 1.6x faster prefill and decode on mobile CPUs with comparable or superior quality. Our analysis of Pareto-frontier design choices offers actionable principles for OD-LLM design.

cs.LG↗

arXiv2025

MobileLLM-Pro Technical Report

Efficient on-device language models around 1 billion parameters are essential for powering low-latency AI applications on mobile and wearable devices. However, achieving strong performance in this model class, while supporting long context windows and practical deployment remains a significant challenge. We introduce MobileLLM-Pro, a 1-billion-parameter language model optimized for on-device deployment. MobileLLM-Pro achieves state-of-the-art results across 11 standard benchmarks, significantly outperforming both Gemma 3-1B and Llama 3.2-1B, while supporting context windows of up to 128,000 tokens and showing only minor performance regressions at 4-bit quantization. These improvements are enabled by four core innovations: (1) implicit positional distillation, a novel technique that effectively instills long-context capabilities through knowledge distillation; (2) a specialist model merging framework that fuses multiple domain experts into a compact model without parameter growth; (3) simulation-driven data mixing using utility estimation; and (4) 4-bit quantization-aware training with self-distillation. We release our model weights and code to support future research in efficient on-device language models.

cs.LG↗

arXiv2025

Colossal Effect of Nanopore Surface Ionic Charge on the Dynamics of Confined Water

Interfacial interactions significantly alter the fundamental properties of water confined in mesoporous structures, with crucial implications for geological, physicochemical, and biological processes. Herein, we focused on the effect of changing the surface ionic charge of nanopores with comparable pore size (3.5-3.8 nm) on the dynamics of confined liquid water. The control of the pore surface ionicity was achieved by using two periodic mesoporous organosilicas (PMOs) containing either neutral or charged forms of a chemically similar bridging unit. The effect on the dynamics of water at the nanoscale was investigated in the temperature range of 245 -300 K, encompassing the glass transition by incoherent quasielastic neutron scattering (QENS), For both types of PMOs, the water dynamics revealed two distinct types of molecular motions: rapid local movements and translational jump diffusion. While the neutral PMO induces a moderate confinement effect, we show that the charged PMO drastically slows down water dynamics, reducing translational diffusion by a factor of four and increasing residence time by an order of magnitude. Notably, by changing the pore filling values, we demonstrate that for charged pore this effect extends beyond the interfacial layer of surface-bound water molecules to encompass the entire pore volume. Thus, our observation indicates a dramatic change in the long-range character of the interaction of water confined in nanopores with surface ionic charge compared to a simple change in hydrophilicity. This is relevant for the understanding of a broad variety of applications in (nano)technological phenomena and processes, such as nanofiltration and membrane design.

physics.chem-ph↗

arXiv2025

Mineral Detection of Neutrinos and Dark Matter 2025 Proceedings

The third ``Mineral Detection of Neutrinos and Dark Matter'' (MD$\nu$DM'25) meeting was held May 20-23, 2025 in Yokohama, Japan, hosted by the Yokohama Institute for Earth Sciences, Japan Agency for Marine-Earth Science and Technology (JAMSTEC). These proceedings compile contributions from the workshop and update the progress of mineral detector research. MD$\nu$DM'25 was the third such meeting, following the first in October of 2022 held at the IFPU in Trieste, Italy and the second in January of 2024 hosted by the Center for Neutrino Physics at Virginia Tech in Arlington, USA. Mineral detectors record and retain damage induced by nuclear recoils in synthetic or natural mineral samples. The damage features can then be read out by a variety of nano- and micro-scale imaging techniques. Applications of mineral detectors on timescales relevant for laboratory experiments include reactor neutrino monitoring and dark matter detection, with the potential to measure the directions as well as the energies of the induced nuclear recoils. For natural mineral detectors which record nuclear recoils over geological timescales, reading out even small mineral samples could be sensitive to rare interactions induced by astrophysical neutrinos, cosmic rays, dark matter and heavy exotic particles. A series of mineral detectors of different ages could measure the time evolution of these fluxes, offering a unique window into the history of our solar system and the Milky Way. Mineral detector research is highly multidisciplinary, incorporating aspects of high energy physics, condensed matter physics, materials science, geoscience, and AI/ML for data analysis. Although realizing the scientific potential of mineral detectors poses many challenges, the MD$\nu$DM community looks forward to the continued development of mineral detector experiments and the possible discoveries that mineral detectors could reveal.

physics.ins-det↗

arXiv2025

Detectability of Covert Fissile Material Production in Nuclear Fusion Reactors via Antineutrino Emissions

Fusion power systems can in principle be used to make significant amounts of fissile material. To do so, an operator would have to introduce fertile material, such as uranium-238, in a suitable region of the reactor where it is exposed to an intense neutron flux. The possibility of using a fusion reactor for this purpose has raised the question of how these facilities can be monitored to ensure their peaceful use. This study examines whether covert production of fissile material in a declared fusion plant could be detected with an onsite antineutrino detector. We find that even a relatively small detector should be able to confirm production rates of a few kilograms of plutonium over 30 days, despite the cosmogenic background and the antineutrino emissions associated with neutron activation of reactor components.

physics.ins-det↗

arXiv2025

LIQUIDating the Gallium Anomaly

The gallium anomaly has a global significance of greater than $5\sigma$. Most viable BSM solutions quickly run into strong tensions with reactor and solar neutrino data. We propose to use indium (${}^{115}\text{In}$) as a target as it offers a low threshold and reasonably high cross section. The neutrino-indium charged current cross section can be calibrated using the well-constrained solar ${}^{7}\text{Be}$ neutrino flux that lies very close in energy to the ${}^{51}\text{Cr}$ neutrino lines. The triple coincidence provided by ${}^{115}\text{In}$ neutrino capture can be fully exploited by an opaque scintillation detector that also provides energy and position information. We show that a $100$ ton indium target combined with 2 source runs of a $3.4$ MCi ${}^{51}\text{Cr}$ source can probe the complete parameter space of the gallium anomaly, both in the context of a vanilla sterile neutrino as well as more involved BSM scenarios.

hep-ph↗

arXiv2025

AutoMixer: Checkpoint Artifacts as Automatic Data Mixers

In language model training, it is desirable to equip models with capabilities from various tasks. However, it is not clear how to directly obtain the right data mixtures for these capabilities as the relationship between data and tasks is difficult to be modeled. In this work, we observe that checkpoint models exhibit emerging capabilities at different points in the training trajectory. Often, the training process saves checkpoints as artifacts that are under-utilized as a source of in-training data signals. We identify these artifact models based on their respective capabilities on the benchmarks and leverage them as data mixers by using their aggregated first-order influence approximation over source data. We demonstrated on eight reasoning benchmarks that the proposed framework shows significant improvements in the pretraining setting, with performance improvements of up to 1.93%. Overall, this shows the potential of checkpoint models to enhance data quality and optimize data mixtures.

cs.CL↗

arXiv2025

Directional Dark Field for Nanoscale Full-Field Transmission X-Ray Microscopy

Dark-field X-ray imaging visualizes structural inhomogeneities through small-angle scattering, but existing directional methods are confined to the micrometer scale. While recent advances have extended dark-field capabilities to nanoscale transmission X-ray microscopy, directional scattering retrieval - critical for characterizing anisotropic nanostructures - has remained inaccessible for imaging resolutions in the sub-micrometer scale. Here, we demonstrate the first directional dark-field setup for nanoimaging, achieving orientation mapping of scattering features below the spatial resolution limit. Our method is experimentally simple to implement with existing transmission X-ray microscopy setups. We validate its performance by successfully resolving sub-resolution test structure orientations, cross-correlating orientational changes within hierarchical nanoporous materials, and mapping the directional arrangement of hydroxyapatite nanocrystals 30 - 70 nm within human tooth enamel. By utilizing shadow regions in the optical configuration, we further extend the detectable scattering vector range, demonstrating a pathway toward size-selective dark-field imaging. This advancement enables the quantitative structural characterization of anisotropic nanomaterials, which are critical to biomineralization, advanced materials, and nanotechnology applications.

physics.optics↗

arXiv2025

Heavy Neutral Lepton Decay Searches using Solar Neutrinos

We study the sensitivity to the decay of a heavy neutral lepton into $e^+e^-$-pairs using the solar boron-8 neutrino flux as source. We provide a fully differential cross section for this process including the interference of neutral and charged current amplitudes. We revisit a previous bound from Borexino and make predicitions for the expected sensitivity in future large liquid noble gas detectors, like XLZD, Argo and DUNE, as well as high-resolution scintillator detectors based on the LiquidO technology. We find that more than two orders of magnitude improvement in mixing angle reach is possible relative to existing bounds.

hep-ph↗

arXiv2025

Nuclear recoil detection with color centers in bulk lithium fluoride

We present initial results on nuclear recoil detection based on the fluorescence of color centers created by nuclear recoils in lithium fluoride. We use gamma rays, fast and thermal neutrons, and study the difference in responses they induce, showing that this type of detector is rather insensitive to gamma rays. We use light-sheet fluorescence microscopy to image nuclear recoil tracks from fast and thermal neutron interactions deep inside a cubic-centimeter sized crystal and demonstrate automated feature extraction in three dimensions using machine learning tools. The number, size, and topology of the events agree with expectations based on simulations with TRIM. These results constitute the first step towards 10-1000g scale detectors with single-event sensitivity for applications such as the detection of dark matter particles, reactor neutrinos, and neutrons.

nucl-ex↗

arXiv2025

CONFLUX: A Standardized Framework to Calculate Reactor Antineutrino Flux

Nuclear fission reactors are abundant sources of antineutrinos. The flux and spectrum of antineutrinos emitted by a reactor can indicate its activity and composition, suggesting potential applications of neutrino measurements beyond fundamental scientific studies that may be valuable to society. The utility of reactor antineutrinos for applications and fundamental science is dependent on the availability of precise predictions of these emissions. For example, in the last decade, disagreements between reactor antineutrino measurements and models have inspired revision of reactor antineutrino calculations and standard nuclear databases as well as searches for new fundamental particles not predicted by the Standard Model of particle physics. Past predictions and descriptions of the methods used to generate them are documented to varying degrees in the literature, with different modeling teams incorporating a range of methods, input data, and assumptions. The resulting difficulty in accessing or reproducing past models and reconciling results from differing approaches complicates the future study and application of reactor antineutrinos. The CONFLUX (Calculation Of Neutrino FLUX) software framework is a neutrino prediction tool built with the goal of simplifying, standardizing, and democratizing the process of reactor antineutrino flux calculations. CONFLUX include three primary methods for calculating the antineutrino emissions of nuclear reactors or individual beta decays that incorporate common nuclear data and beta decay theory. The software is prepackaged with the current nuclear database. It includes the capability to predict time-dependent neutrino model, adjust decay information entries, and propagate uncertainties. This paper describes the software structure, details the methods used for flux and spectrum calculations, and talks about potential use cases.

physics.ins-det↗

arXiv2025

CoSMoEs: Compact Sparse Mixture of Experts

Sparse Mixture of Expert (MoE) models are popular foundational architectures at large scale, however, under-explored at smaller sizes. Here, we show how to enable Compact Sparse Mixture of Experts (CoSMoEs) for on-device inference. Specifically, we tackle the three main on-device dimensions: Quality, Memory and Latency. Along the quality axis, we show that in a fair evaluation (removing confounding factors) MoE architectures outperform FLOP-aligned dense models at on-device scale. We introduce weight-decomposed experts, further improving the MoE model performance. Regarding model memory and latency, we significantly improve model offloading efficiency and, in turn, reduce model inference latency.

cs.LG↗

arXiv2025

New Physics versus Quenching Factors in Coherent Neutrino Scattering

Recent results on the Coherent Elastic Neutrino-Nucleus Scattering (CE$\nu$NS) on germanium present significant discrepancies among experiments. We perform a combined analysis of the Dresden-II, CONUS+ and COHERENT data, quantifying the impact of quenching factor uncertainties on their CE$\nu$NS cross section measurement. No choice of quenching factor can bring these three data sets into mutual agreement, whereas the combination of COHERENT with either Dresden-II or CONUS+ agrees well albeit for very different quenching factors. We further study the quenching factor dependence on the sensitivity of these experiments to a large neutrino magnetic moment, finding that the constraints can vary by up to an order of magnitude. Our work highlights the importance of reducing this uncertainty on quenching factors in order to probe new physics from neutrinos at the low-energy frontier.

hep-ph↗

arXiv2024

First-principles Spin and Optical Properties of Vacancy Clusters in Lithium Fluoride

Vacancy-cluster color centers in lithium fluoride have been studied in detail both theoretically and experimentally for over a century, giving rise to various applications in solid-state lasers, broadband photonic devices, and radiation dosimeters. These color centers are also attractive candidate platforms for applications in quantum information science, due to their spin properties and strong coupling to the crystal lattice, which allows their properties to be easily tuned. Here we present hybrid functional calculations of common vacancy defects in lithium fluoride, including their energetic, spin, and optical properties. We show that for a wide range of hybrid functional parameters tuned to match the experimental band gap, certain defects have little variation in their predicted optical properties. We further demonstrate that the parameters needed to satisfy the generalized Koopman's theorem and correctly position defect levels within the gap, can vary dramatically, even for different charge states of the same defect. Our work establishes the accuracy of the computationally lightweight hybrid-functional approach for predicting the optical and energetic properties of color centers in polar materials.

cond-mat.mtrl-sci↗

arXiv2024

Maximum entropy mediated liquid-to-solid nucleation and transition

Molecular Dynamics (MD) simulations are a powerful tool for studying matter at the atomic scale. However, to simulate solids, an initial atomic structure is crucial for the successful execution of MD simulations, but can be difficult to prepare due to insufficient atomistic information. At the same time Wide Angle X-ray Scattering (WAXS) measurements can determine the Radial Distribution Function (RDF) of atomic structures. However, the interpretation of RDFs is often challenging. Here we present an algorithm that can bias MD simulations with RDFs by combining the information of the MD atomic interaction potential and the RDF under the principle of maximum relative entropy. We show that this algorithm can be used to adjust the RDF of one liquid model, e.g., the TIP3P water model, to reproduce the RDF and improve the Angular Distribution Function (ADF) of another model, such as the TIP4P/2005 water model. In addition, we demonstrate that the algorithm can initiate crystallization in liquid systems, leading to both stable and metastable crystalline states defined by the RDF, e.g., crystallization of water to ice and liquid TiO2 to rutile or anatase. Finally, we discuss how this method can be useful for improving interaction models, studying crystallization processes, interpreting measured RDFs, or training machine learned potentials.

physics.comp-ph↗

arXiv2024

Clarity through the Neutrino Fog: Constraining New Forces in Dark Matter Detectors

The PANDAX-4T and XENONnT experiments present indications of Coherent Elastic Neutrino Nucleus Scattering (CE$\nu$NS) from ${}^{8}$B solar neutrinos at 2.6$\sigma$ and 2.7$\sigma$, respectively. This constitutes the first observation of the neutrino "floor" or "fog", an irreducible background that future dark matter searches in terrestrial detectors will have to contend with. Here, we first discuss the contributions from neutrino-electron scattering and from the Migdal effect in the region of interest of these experiments, and we argue that they are non-negligible. Second, we make use of the recent PANDAX-4T and XENONnT data to derive novel constraints on light scalar and vector mediators coupling to neutrinos and quarks. We demonstrate that these experiments already provide world-leading laboratory constraints on new light mediators in some regions of parameter space.

hep-ph↗