Search

The search for dark matter and physics beyond the Standard Model has grown to encompass a highly interdisciplinary approach. In this review, we survey recent searches for light, weakly-coupled particles - axions and dark photons - over the past decade, focusing on new experimental results and the incorporation of technologies and techniques from fields as diverse as quantum science, microwave engineering, precision magnetometry, and condensed matter physics. We also review theoretical progress which has been useful in identifying new experimental directions, and identify the areas of most rapid experimental progress and the technological advances required to continue exploring the parameter space for axions and dark photons., Comment: 31 pages, 5 figures. Invited review for Annual Reviews Nuclear Particle Science 2025

We associate an $L$-function $L^{\mathrm{near}}(M,s)$ to any geometric motive over a global field $K$ in the sense of Voevodsky. This is a Dirichlet series which converges in some half-plane and has an Euler product factorisation. When $M$ is the dual of $M(X)$ for $X$ a smooth projective variety, $L^{\mathrm{near}}(M,s)$ differs from the alternating product of the zeta functions defined by Serre in 1969 only at places of bad reduction; in exchange, it is multiplicative with respect to exact triangles. If $K$ is a function field over $\mathbf{F}_q$, $L{\mathrm{near}}(M,s)$ is a rational function in $q^{-s}$ and enjoys a functional equation. The techniques use the full force of Ayoub's six (and even seven) operations.

We show that a restricted version of a conjecture of M. Talagrand on the relation between "expectation thresholds" and "fractional expectation thresholds" follows easily from a strong version of a second conjecture of Talagrand, on "selector processes." The selector process conjecture was proved by Park and Pham, and the quantitative strengthening used here is due to Bednorz, Martynek, and Meller., Comment: 3 pages

Despite having triggered devastating pandemics in the past, our ability to quantitatively assess the emergence potential of individual strains of animal influenza viruses remains limited. This study introduces Emergenet, a tool to infer a digital twin of sequence evolution to chart how new variants might emerge in the wild. Our predictions based on Emergenets built only using 220,151 Hemagglutinnin (HA) sequences consistently outperform WHO seasonal vaccine recommendations for H1N1/H3N2 subtypes over two decades (average match-improvement: 3.73 AAs, 28.40\%), and are at par with state-of-the-art approaches that use more detailed phenotypic annotations. Finally, our generative models are used to scalably calculate the current odds of emergence of animal strains not yet in human circulation, which strongly correlates with CDC's expert-assessed Influenza Risk Assessment Tool (IRAT) scores (Pearson's $r = 0.721, p = 10^{-4}$). A minimum five orders of magnitude speedup over CDC's assessment (seconds vs months) then enabled us to analyze 6,354 animal strains collected post-2020 to identify 35 strains with high emergence scores ($> 7.7$). The Emergenet framework opens the door to preemptive pandemic mitigation through targeted inoculation of animal hosts before the first human infection., Comment: 35 pages, 15 figures

Optically-addressable spin qubits form the foundation of a new generation of emerging nanoscale sensors. The engineering of these sensors has mainly focused on solid-state systems such as the nitrogen-vacancy (NV) center in diamond. However, NVs are restricted in their ability to interface with biomolecules due to their bulky diamond host. Meanwhile, fluorescent proteins have become the gold standard in bioimaging, as they are genetically encodable and easily integrated with biomolecules. While fluorescent proteins have been suggested to possess a metastable triplet state, they have not been investigated as qubit sensors. Here, we realize an optically-addressable spin qubit in the Enhanced Yellow Fluorescent Protein (EYFP) enabled by a novel spin-readout technique. A near-infrared laser pulse allows for triggered readout of the triplet state with up to 44% spin contrast. Using coherent microwave control of the EYFP spin at liquid-nitrogen temperatures, we measure a spin-lattice relaxation time of $(141 \pm 5)$ {\mu}s, a $(16 \pm 2)$ {\mu}s coherence time under Carr-Purcell-Meiboom-Gill (CPMG) decoupling, and a predicted oscillating (AC) magnetic field sensitivity with an upper bound of $183 \, \mathrm{fT}\, \mathrm{mol}^{1/2}\, \mathrm{Hz}^{-1/2}$. We express the qubit in mammalian cells, maintaining contrast and coherent control despite the complex intracellular environment. Finally, we demonstrate optically-detected magnetic resonance at room temperature in aqueous solution with contrast up to 3%, and measure a static (DC) field sensitivity with an upper bound of $93 \, \mathrm{pT}\, \mathrm{mol}^{1/2}\, \mathrm{Hz}^{-1/2}$. Our results establish fluorescent proteins as a powerful new qubit sensor platform and pave the way for applications in the life sciences that are out of reach for solid-state technologies.

Dramatic increases in the capabilities of neural network models in recent years are driven by scaling model size, training data, and corresponding computational resources. To develop the exceedingly large networks required in modern applications, such as large language models (LLMs), model training is distributed across tens of thousands of hardware accelerators (e.g. GPUs), requiring orchestration of computation and communication across large computing clusters. In this work, we demonstrate that careful consideration of hardware configuration and parallelization strategy is critical for effective (i.e. compute- and cost-efficient) scaling of model size, training data, and total computation. We conduct an extensive empirical study of the performance of large-scale LLM training workloads across model size, hardware configurations, and distributed parallelization strategies. We demonstrate that: (1) beyond certain scales, overhead incurred from certain distributed communication strategies leads parallelization strategies previously thought to be sub-optimal in fact become preferable; and (2) scaling the total number of accelerators for large model training quickly yields diminishing returns even when hardware and parallelization strategies are properly optimized, implying poor marginal performance per additional unit of power or GPU-hour.

Social impact evaluations are emerging as a useful tool to understand, document, and evaluate the societal impacts of generative AI. In this provocation, we begin to think carefully about the types of experts and expertise that are needed to conduct robust social impact evaluations of generative AI. We suggest that doing so will require thoughtfully eliciting and integrating insights from a range of "domain experts" and "experiential experts," and close with five open questions.

This document is comprised of a collection of updated preliminary parameters for the key parts of the muon collider. The updated preliminary parameters follow on from the October 2023 Tentative Parameters Report. Particular attention has been given to regions of the facility that are believed to hold greater technical uncertainty in their design and that have a strong impact on the cost and power consumption of the facility. The data is collected from a collaborative spreadsheet and transferred to overleaf.

Prostate cancer is the second most common form of cancer, though most patients have a positive prognosis with many experiencing long-term survival with current treatment options. Yet, each treatment carries varying levels of intensity and side effects, therefore determining the severity of prostate cancer is an important criteria in selecting the most appropriate treatment. The Gleason score is the most common grading system used to judge the severity of prostate cancer, but much of the grading process can be affected by human error or subjectivity. Finding biomarkers for prostate cancer Gleason scores in a quantitative, machine-driven approach could enable pathologists to validate their assessment of a patient cancer sample by examining such biomarkers. In our study, we identified biomarkers from multi-omics data using machine learning, statistical tools, and deep learning to train models against the Gleason score and capture the most important features that could potentially serve as biomarkers for the Gleason score. Through this process, multiple genes, such as COL1A1 and SFRP4, and cell cycle pathways, such as G2M checkpoint, E2F targets, and the PLK1 pathways, were found to be important predictive features for particular Gleason scores. The combination of these analytical methods shows potential for more accurate grading of prostate cancer, and greater understanding of biological processes behind prostate cancer severity that could provide additional therapeutic targets.

An exposition of the mathematics underpinning the neural network architecture of a GPT-3-style LLM., Comment: 10 pages

We give a sheaf-theoretic version of the universal coefficient theorem.

This paper focuses on creating synthetic data to improve the quality of image captions. Existing works typically have two shortcomings. First, they caption images from scratch, ignoring existing alt-text metadata, and second, lack transparency if the captioners' training data (e.g. GPT) is unknown. In this paper, we study a principled approach Altogether based on the key idea to edit and re-align existing alt-texts associated with the images. To generate training data, we perform human annotation where annotators start with the existing alt-text and re-align it to the image content in multiple rounds, consequently constructing captions with rich visual concepts. This differs from prior work that carries out human annotation as a one-time description task solely based on images and annotator knowledge. We train a captioner on this data that generalizes the process of re-aligning alt-texts at scale. Our results show our Altogether approach leads to richer image captions that also improve text-to-image generation and zero-shot image classification tasks., Comment: accepted by EMNLP 2024; Meta CLIP 1.2 Data Engine

Rapid search and rescue is critical to maximizing survival rates following natural disasters. However, these efforts are challenged by the need to search large disaster zones, lack of reliability in the communications infrastructure, and a priori unknown numbers of objects of interest (OOIs), such as injured survivors. Aerial robots are increasingly being deployed for search and rescue due to their high mobility, but there remains a gap in deploying multi-robot autonomous aerial systems for methodical search of large environments. Prior works have relied on preprogrammed paths from human operators or are evaluated only in simulation. We bridge these gaps in the state of the art by developing and demonstrating a decentralized active search system, which biases its trajectories to take additional views of uncertain OOIs. The methodology leverages stochasticity for rapid coverage in communication denied scenarios. When communications are available, robots share poses, goals, and OOI information to accelerate the rate of search. Extensive simulations and hardware experiments in Bloomingdale, OH, are conducted to validate the approach. The results demonstrate the active search approach outperforms greedy coverage-based planning in communication-denied scenarios while maintaining comparable performance in communication-enabled scenarios.

The edge space $\mathcal{E}(G)$ of a graph $G$ is the vector space $\mathbb{F}_2^{E(G)}$ with members naturally identified with subgraphs of $G$, and the $H$-space is the subspace $\mathcal{C}_H(G)$ of $ \mathcal{E}(G)$ spanned by copies of the graph $H$. We are interested in when the random graph $G = G_{n,p}$ is likely to satisfy \[\mathcal{C}_H(G) = \mathcal{W}_H(G),\] where $\mathcal{W}_H(G)$ takes one of four natural values, depending on the value of $\mathcal{C}_H(K_n)$. We show that for strictly $2$-balanced $H$, w.h.p. the above equality holds whenever every edge of $G$ is in a copy of $H$., Comment: arXiv admin note: text overlap with arXiv:1610.01276

Generative artificial intelligence (AI) technology is revolutionizing the computing industry. Not only its applications have broadened to various sectors but also poses new system design and optimization opportunities. The technology is capable of understanding and responding in multiple modalities. However, the advanced capability currently comes with significant system resource demands. To sustainably scale generative AI capabilities to billions of users in the world, inference must be fast and efficient. This paper pinpoints key system design and optimization opportunities by characterizing a family of emerging multi-modal generation models on real systems. Auto-regressive token generation is a critical latency performance bottleneck, typically dominated by GPU idle time. In addition to memory-intensive attention across the generative AI models, linear operations constitute significant inference latency due to the feed forward networks in Transformer-based models. We demonstrate that state-of-the-art optimization levers, spanning from applications to system software and hardware, set a 3.88x better baseline., Comment: 13 pages including references. 8 Figures. Under review to HPCA 2025 Industry Track

Diagnostic imaging studies are an increasingly important component of the workup and management of acutely presenting patients. However, ordering appropriate imaging studies according to evidence-based medical guidelines is a challenging task with a high degree of variability between healthcare providers. To address this issue, recent work has investigated if generative AI and large language models can be leveraged to help clinicians order relevant imaging studies for patients. However, it is challenging to ensure that these tools are correctly aligned with medical guidelines, such as the American College of Radiology's Appropriateness Criteria (ACR AC). In this study, we introduce a framework to intelligently leverage language models by recommending imaging studies for patient cases that are aligned with evidence-based guidelines. We make available a novel dataset of patient "one-liner" scenarios to power our experiments, and optimize state-of-the-art language models to achieve an accuracy on par with clinicians in image ordering. Finally, we demonstrate that our language model-based pipeline can be used as intelligent assistants by clinicians to support image ordering workflows and improve the accuracy of imaging study ordering according to the ACR AC. Our work demonstrates and validates a strategy to leverage AI-based software to improve trustworthy clinical decision making in alignment with expert evidence-based guidelines., Comment: 15 pages main text, 4 figures, 1 table

Passively collected "big" data sources are increasingly used to inform critical development policy decisions in low- and middle-income countries. While prior work highlights how such approaches may reveal sensitive information, enable surveillance, and centralize power, less is known about the corresponding privacy concerns, hopes, and fears of the people directly impacted by these policies -- people sometimes referred to as experiential experts. To understand the perspectives of experiential experts, we conducted semi-structured interviews with people living in rural villages in Togo shortly after an entirely digital cash transfer program was launched that used machine learning and mobile phone metadata to determine program eligibility. This paper documents participants' privacy concerns surrounding the introduction of big data approaches in development policy. We find that the privacy concerns of our experiential experts differ from those raised by privacy and development domain experts. To facilitate a more robust and constructive account of privacy, we discuss implications for policies and designs that take seriously the privacy concerns raised by both experiential experts and domain experts.

We show that the continuous \'etale cohomology groups $H^n_{\mathrm{cont}}(X,\mathbf{Q}_l(n))$ of smooth varieties $X$ over a finite field $k$ are spanned as $\mathbf{Q}_l$-vector spaces by the $n$-th Milnor $K$-sheaf locally for the Zariski topology, for all $n\ge 0$. Here $l$ is a prime invertible in $k$. This is the first general unconditional result towards the conjectures of arXiv:math/9801017 (math.AG) which put together the Tate and the Beilinson conjectures relative to algebraic cycles on smooth projective $k$-varieties.

Typical multi-mode fibers exhibit strong intra-group mode coupling and weak inter-group mode coupling. Mode scramblers can be inserted at periodic intervals to enhance inter-group coupling. The deterministic mode coupling of the mode scramblers, in concert with the random mode coupling of the fiber spans, can effect strong random mode coupling between all modes. This reduces both modal dispersion and mode-dependent loss, thereby decreasing receiver complexity and increasing link capacity. In this paper, we analyze the effect of mode scramblers on end-to-end group-delay and mode-dependent loss standard deviations in long-haul multi-mode fiber links. We develop analytical tools in the generalized Jones and Stokes representations. We propose design criteria for mode scramblers that ensure strong end-to-end coupling: the mode-group-averaged power coupling matrix should be primitive and its non-dominant eigenvalues should be near zero. We argue that when the mode scramblers satisfy these criteria, the probability distribution of the system transfer matrix asymptotically approaches that of a system with strong random mode coupling between all modes. Consequently, group-delay and mode-dependent loss standard deviations become sufficient statistics of the eigenvalues of the group-delay operator and the modal gains operator, respectively. We also show that under certain conditions on the uncoupled group delays, it is possible to design self-compensating mode scramblers to reduce group delay accumulation below that of standard strong random coupling., Comment: * These authors contributed equally

Social scientists quickly adopted large language models due to their ability to annotate documents without supervised training, an ability known as zero-shot learning. However, due to their compute demands, cost, and often proprietary nature, these models are often at odds with replication and open science standards. This paper introduces the Political DEBATE (DeBERTa Algorithm for Textual Entailment) language models for zero-shot and few-shot classification of political documents. These models are not only as good, or better than, state-of-the art large language models at zero and few-shot classification, but are orders of magnitude more efficient and completely open source. By training the models on a simple random sample of 10-25 documents, they can outperform supervised classifiers trained on hundreds or thousands of documents and state-of-the-art generative models with complex, engineered prompts. Additionally, we release the PolNLI dataset used to train these models -- a corpus of over 200,000 political documents with highly accurate labels across over 800 classification tasks., Comment: 26 pages, 5 figures

Knowledge Distillation (KD) is a powerful approach for compressing a large model into a smaller, more efficient model, particularly beneficial for latency-sensitive applications like recommender systems. However, current KD research predominantly focuses on Computer Vision (CV) and NLP tasks, overlooking unique data characteristics and challenges inherent to recommender systems. This paper addresses these overlooked challenges, specifically: (1) mitigating data distribution shifts between teacher and student models, (2) efficiently identifying optimal teacher configurations within time and budgetary constraints, and (3) enabling computationally efficient and rapid sharing of teacher labels to support multiple students. We present a robust KD system developed and rigorously evaluated on multiple large-scale personalized video recommendation systems within Google. Our live experiment results demonstrate significant improvements in student model performance while ensuring consistent and reliable generation of high quality teacher labels from a continuous data stream of data.

We introduce Transfusion, a recipe for training a multi-modal model over discrete and continuous data. Transfusion combines the language modeling loss function (next token prediction) with diffusion to train a single transformer over mixed-modality sequences. We pretrain multiple Transfusion models up to 7B parameters from scratch on a mixture of text and image data, establishing scaling laws with respect to a variety of uni- and cross-modal benchmarks. Our experiments show that Transfusion scales significantly better than quantizing images and training a language model over discrete image tokens. By introducing modality-specific encoding and decoding layers, we can further improve the performance of Transfusion models, and even compress each image to just 16 patches. We further demonstrate that scaling our Transfusion recipe to 7B parameters and 2T multi-modal tokens produces a model that can generate images and text on a par with similar scale diffusion models and language models, reaping the benefits of both worlds., Comment: 23 pages

SPIDER is a balloon-borne instrument designed to map the cosmic microwave background at degree-angular scales in the presence of Galactic foregrounds. SPIDER has mapped a large sky area in the Southern Hemisphere using more than 2000 transition-edge sensors (TESs) during two NASA Long Duration Balloon flights above the Antarctic continent. During its first flight in January 2015, SPIDER observed in the 95 GHz and 150 GHz frequency bands, setting constraints on the B-mode signature of primordial gravitational waves. Its second flight in the 2022-23 season added new receivers at 280 GHz, each using an array of TESs coupled to the sky through feedhorns formed from stacks of silicon wafers. These receivers are optimized to produce deep maps of polarized Galactic dust emission over a large sky area, providing a unique data set with lasting value to the field. In this work, we describe the instrument's performance during SPIDER's second flight., Comment: Submitted to SPIE Astronomical Telescopes + Instrumentation 2024, JATIS

While silicon solar cells dominate global photovoltaic energy production, their continued improvement is hindered by the single junction limit. One potential solution is to use molecular singlet exciton fission to generate two electrons from each absorbed high-energy photon. We demonstrate that the long-standing challenge of coupling molecular excited states to silicon solar cells can be overcome using sequential charge transfer. Combining zinc phthalocyanine, aluminum oxide, and a shallow junction crystalline silicon microwire solar cell, the peak charge generation efficiency per photon absorbed in tetracene is (138 +- 6)%, comfortably surpassing the quantum efficiency limit for conventional silicon solar cells and establishing a new, scalable approach to low cost, high efficiency photovoltaics., Comment: 34 pages, 14 figures

Fixed-target proton-beam experiments produce a multitude of charged pions that rescatter in the beam dump. These charged pion scattering events can be an additional irreducible source of exotic particles which couple to photons or hadrons. We analyze the sensitivity of the DUNE Near Detector complex to millicharged particles (MCPs) and heavy axion-like particles (ALPs) with low-energy couplings to gluons. Using the framework of chiral perturbation theory, we demonstrate regimes of parameter space where the charged pion production channel dominates over previously-considered production mechanisms for both MCPs and ALPs, thereby improving the sensitivity of DUNE to these new particles compared to previous studies., Comment: 11+1 pages, 11 figures. v2: corrected implementation of momentum transfer cutoff, conclusions unchanged

The International Muon Collider Collaboration (IMCC) [1] was established in 2020 following the recommendations of the European Strategy for Particle Physics (ESPP) and the implementation of the European Strategy for Particle Physics-Accelerator R&D Roadmap by the Laboratory Directors Group [2], hereinafter referred to as the the European LDG roadmap. The Muon Collider Study (MuC) covers the accelerator complex, detectors and physics for a future muon collider. In 2023, European Commission support was obtained for a design study of a muon collider (MuCol) [3]. This project started on 1st March 2023, with work-packages aligned with the overall muon collider studies. In preparation of and during the 2021-22 U.S. Snowmass process, the muon collider project parameters, technical studies and physics performance studies were performed and presented in great detail. Recently, the P5 panel [4] in the U.S. recommended a muon collider R&D, proposed to join the IMCC and envisages that the U.S. should prepare to host a muon collider, calling this their "muon shot". In the past, the U.S. Muon Accelerator Programme (MAP) [5] has been instrumental in studies of concepts and technologies for a muon collider., Comment: This document summarises the International Muon Collider Collaboration (IMCC) progress and status of the Muon Collider R&D programme

We study the structure of geodesics in the fractal random metric constructed by Kendall from a self-similar Poisson process of roads (i.e, lines with speed limits) in $\mathbb{R}^2$. In particular, we prove a conjecture of Kendall stating that geodesics do not pause en route, i.e, use roads of arbitrary small speed except at their endpoints. It follows that the geodesic frame of $\left(\mathbb{R}^2,T\right)$ is the set of points on roads. We also consider geodesic stars and hubs, and give a complete description of the local structure of geodesics around points on roads. Notably, we prove that leaving a road by driving off-road is never geodesic., Comment: 57 pages, 14 figures. Comments are welcome!

It is shown that the following holds for each $\varepsilon >0$. For $G$ an $n$-vertex graph of maximum degree $D$, lists $S_v$ ($v\in V(G)$), and $L_v$ chosen uniformly from the ($(1+\varepsilon)\ln n$)-subsets of $S_v$ (independent of other choices), \[ \mbox{$G$ admits a proper coloring $\sigma$ with $\sigma_v\in L_v$ $\forall v$} \] with probability tending to 1 as $D\to \infty$. When each $S_v $ is $\{1\dots D+1\}$, this is an asymptotically optimal version of the ``palette sparsification'' theorem of Assadi, Chen and Khanna that was proved in an earlier paper by the present authors., Comment: 37 pages, 0 figures. arXiv admin note: text overlap with arXiv:2306.00171

The Vera C. Rubin Observatory is a unique facility for survey astronomy that will soon be commissioned and begin operations. Crucial to many of its scientific goals is the achievement of sustained high image quality, limited only by the seeing at the site. This will be maintained through an Active Optics System (AOS) that controls optical element misalignments and corrects mirror figure error to minimize aberrations caused by both thermal and gravitational distortions. However, the large number of adjustment degrees of freedom available on the Rubin Observatory introduces a range of degeneracies, including many that are \textit{noise-induced} due to imperfect measurement of the wavefront errors. We present a structured methodology for identifying these degeneracies through an analysis of image noise level. We also present a novel scaling strategy based on Truncated Singular Value Decomposition (TSVD) that mitigates the degeneracy, and optimally distributes the adjustment over the available degrees of freedom. Our approach ensures the attainment of optimal image quality, while avoiding excursions around the noise-induced subspace of degeneracies, marking a significant improvement over the previous techniques adopted for Rubin, which were based on an Optimal Integral Controller (OIC). This new approach is likely to also yield significant benefits for all telescopes that incorporate large numbers of degrees of freedom of adjustment., Comment: 19 pages, 8 figures, 4 tables. Submitted to ApJ

Semantic relations among entities are a widely accepted method for relation extraction. PromptORE (Prompt-based Open Relation Extraction) was designed to improve relation extraction with Large Language Models on generalistic documents. However, it is less effective when applied to historical documents, in languages other than English. In this study, we introduce an adaptation of PromptORE to extract relations from specialized documents, namely digital transcripts of trials from the Spanish Inquisition. Our approach involves fine-tuning transformer models with their pretraining objective on the data they will perform inference. We refer to this process as "biasing". Our Biased PromptORE addresses complex entity placements and genderism that occur in Spanish texts. We solve these issues by prompt engineering. We evaluate our method using Encoder-like models, corroborating our findings with experts' assessments. Additionally, we evaluate the performance using a binomial classification benchmark. Our results show a substantial improvement in accuracy -up to a 50% improvement with our Biased PromptORE models in comparison to the baseline models using standard PromptORE.

We show that, for a polarised smooth projective variety $B \hookrightarrow \mathbb{P}^n_k$ of dimension $\geq 2$ over an infinite field $k$ and an abelian variety $A$ over the function field of $B$, there exists a dense Zariski open set of smooth geometrically connected hyperplane sections $h$ of $B$ such that $A$ has good reduction at $h$ and the specialisation homomorphism of Lang-N\'eron groups at $h$ is injective (up to a finite $p$-group in positive characteristic $p$). This gives a positive answer to a conjecture of the first author, which is used to deduce a negative definiteness result on his refined height pairing. This also sheds a new light on N\'eron's specialisation theorem., Comment: Comments welcome