Search

We give upper and lower bounds for weighted Chebyshev and residual polynomials on subsets of the real line. As an application, we prove a Szeg\H{o}-type theorem in the setting of Parreau--Widom sets.

We present a thorough study of spin-to-charge current interconversion in bulk and thin films of (0001) {\alpha}-Fe2O3 /Pt heterostructures by means of all-optical polarization-controlled microwave excitation at sub-Terahertz frequencies. Our results demonstrate that coherent spin pumping is generated through excitations of both the acoustic and optical modes of antiferromagnetic resonance, provided that the corresponding selection rules are met for the relative orientation between the microwave magnetic field h_ac and the magnetic moment m_0 of the Hematite. In particular, our results unanimously show that while a microwave field with h_ac perpendicular to m_0 pumps a net spin angular momentum from the acoustic mode, spin pumping from the optical mode is only enabled when h_ac parallel to m_0, as expected from the selection rules imposed by the Neel vector dynamics. Our results support the current understanding of spin mixing conductance in antiferromagnetic/non-magnetic interfaces, contrary to recent reports where the absence of spin pumping from the optical mode in Hematite was interpreted as a cancellation effect between the diagonal and off-diagonal components of the spin mixing conductance. We also provide an explanation for the previously reported observations and show how the optical spin pumping actually vanishes for thin films, which we speculate being either due to an increased level of inhomogeneities or to insufficient film thickness for the optical mode to fully realize., Comment: 14 pages, 4 figures

Social media empower distributed content creation by algorithmically harnessing "the social fabric" (explicit and implicit signals of association) to serve this content. While this overcomes the bottlenecks and biases of traditional gatekeepers, many believe it has unsustainably eroded the very social fabric it depends on by maximizing engagement for advertising revenue. This paper participates in open and ongoing considerations to translate social and political values and conventions, specifically social cohesion, into platform design. We propose an alternative platform model that the social fabric an explicit output as well as input. Citizens are members of communities defined by explicit affiliation or clusters of shared attitudes. Both have internal divisions, as citizens are members of intersecting communities, which are themselves internally diverse. Each is understood to value content that bridge (viz. achieve consensus across) and balance (viz. represent fairly) this internal diversity, consistent with the principles of the Hutchins Commission (1947). Content is labeled with social provenance, indicating for which community or citizen it is bridging or balancing. Subscription payments allow citizens and communities to increase the algorithmic weight on the content they value in the content serving algorithm. Advertisers may, with consent of citizen or community counterparties, target them in exchange for payment or increase in that party's algorithmic weight. Underserved and emerging communities and citizens are optimally subsidized/supported to develop into paying participants. Content creators and communities that curate content are rewarded for their contributions with algorithmic weight and/or revenue. We discuss applications to productivity (e.g. LinkedIn), political (e.g. X), and cultural (e.g. TikTok) platforms., Comment: 60 pages

Multi-contrast MRI methods acquire multiple images with different contrast weightings, which are used for the differentiation of the tissue types or quantitative mapping. However, the scan time needed to acquire multiple contrasts is prohibitively long for 3D acquisition schemes, which can offer isotropic image resolution. While deep learning-based methods have been extensively used to accelerate 2D and 2D + time problems, the high memory demand, computation time, and need for large training data sets make them challenging for large-scale volumes. To address these challenges, we generalize the plug-and-play multi-scale energy-based model (MuSE) to a regularized subspace recovery setting, where we jointly regularize the 3D multi-contrast spatial factors in a subspace formulation. The explicit energy-based formulation allows us to use variable splitting optimization methods for computationally efficient recovery.

X-ray microtomography is a powerful non-destructive technique allowing 3D virtual histology of resected human tissue. The achievable imaging field-of-view, is however limited by the fixed number of detector elements, enforcing the requirement to sacrifice spatial resolution in order to image larger samples. In applications such as soft-tissue imaging, phase-contrast methods are often employed to enhance image contrast. Some of these methods, especially those suited to laboratory sources, rely on optical elements, the dimensions of which can impose a further limitation on the field-of-view. We describe an efficient method to double the maximum field-of-view of a cone-beam X-ray microtomography system, without sacrificing on spatial resolution, and including multi-contrast capabilities. We demonstrate an experimental realisation of the method, achieving exemplary reconstructions of a resected human lung sample, with a cubic voxel of 10.5 $\mu$m linear dimensions, across a horizontal field-of-view of 4.3 cm. The same concepts are applied to free-space propagation imaging of a 2.7 mm segment of the same sample, achieving a cubic voxel of 450 nm linear dimensions. We show that the methodology can be applied at a range of different length-scales and geometries, and that it is directly compatible with complementary implementations of X-ray phase-contrast imaging., Comment: 14 pages (inc. references), 5 figures

As digital services increasingly replace traditional analogue systems, ensuring that older adults are not left behind is critical to fostering inclusive access. This study explores how digital educators support older adults in developing essential digital skills, drawing insights from interviews with $34$ educators in Ireland. These educators, both professional and volunteer, offer instruction through a range of formats, including workshops, remote calls, and in-person sessions. Our findings highlight the importance of personalized, step-by-step guidance tailored to older adults' learning needs, as well as fostering confidence through hands-on engagement with technology. Key challenges identified include limited transportation options, poor internet connectivity, outdated devices, and a lack of familial support for learning. To address these barriers, we propose enhanced public funding, expanded access to resources, and sustainable strategies such as providing relevant and practical course materials. Additionally, innovative tools like simulated online platforms for practicing digital transactions can help reduce anxiety and enhance digital literacy among older adults. This study underscores the vital role that digital educators play in bridging the digital divide, creating a more inclusive, human-centered approach to digital learning for older adults.

Recent work has explored data thinning, a generalization of sample splitting that involves decomposing a (possibly matrix-valued) random variable into independent components. In the special case of a $n \times p$ random matrix with independent and identically distributed $N_p(\mu, \Sigma)$ rows, Dharamshi et al. (2024a) provides a comprehensive analysis of the settings in which thinning is or is not possible: briefly, if $\Sigma$ is unknown, then one can thin provided that $n>1$. However, in some situations a data analyst may not have direct access to the data itself. For example, to preserve individuals' privacy, a data bank may provide only summary statistics such as the sample mean and sample covariance matrix. While the sample mean follows a Gaussian distribution, the sample covariance follows (up to scaling) a Wishart distribution, for which no thinning strategies have yet been proposed. In this note, we fill this gap: we show that it is possible to generate two independent data matrices with independent $N_p(\mu, \Sigma)$ rows, based only on the sample mean and sample covariance matrix. These independent data matrices can either be used directly within a train-test paradigm, or can be used to derive independent summary statistics. Furthermore, they can be recombined to yield the original sample mean and sample covariance.

We introduce Granite Vision, a lightweight large language model with vision capabilities, specifically designed to excel in enterprise use cases, particularly in visual document understanding. Our model is trained on a comprehensive instruction-following dataset, including document-related tasks, such as content extraction from tables, charts, diagrams, sketches, and infographics, as well as general image tasks. The architecture of Granite Vision is centered around visual modality alignment with a decoder-only, 2 billion parameter Granite large language model. Additionally, we introduce a dedicated safety classification approach in test-time that leverages a sparse set of attention vectors to identify potential harmful inputs. Despite its lightweight architecture, Granite Vision achieves strong results in standard benchmarks related to visual document understanding, as well as on the LiveXiv benchmark, which is designed to avoid test set contamination by using a constantly updated corpus of recently published Arxiv papers. We are releasing the model under the Apache-2 license, allowing for both research and commercial use, while offering complete visibility into the training data and other relevant details. See https://huggingface.co/ibm-granite/ for model weights.

Interactive decision-making is essential in applications such as autonomous driving, where the agent must infer the behavior of nearby human drivers while planning in real-time. Traditional predict-then-act frameworks are often insufficient or inefficient because accurate inference of human behavior requires a continuous interaction rather than isolated prediction. To address this, we propose an active learning framework in which we rigorously derive predicted belief distributions. Additionally, we introduce a novel model-based diffusion solver tailored for online receding horizon control problems, demonstrated through a complex, non-convex highway merging scenario. Our approach extends previous high-fidelity dual control simulations to hardware experiments, which may be viewed at https://youtu.be/Q_JdZuopGL4, and verifies behavior inference in human-driven traffic scenarios, moving beyond idealized models. The results show improvements in adaptive planning under uncertainty, advancing the field of interactive decision-making for real-world applications.

A smoking gun for our current paradigm of the early universe would be direct evidence for the quantum mechanical origin of density perturbations which are conjectured to seed the large scale structure of our universe. A recently-proposed novel phenomenon is that of \textit{recoherence}, wherein a specific interaction between the adiabatic and the entropic sector leads to the adiabatic mode retaining a coherent state after a transient increase in linear entropy. In this paper, we choose the most general Gaussian action and analyze the evolution of linear entropy, complexity of purification (COP), and complexity of formation (COF) to capture the interplay between decoherence and recoherence in this model. In the presence of two types of couplings that drive these two opposing characteristics, we highlight how COF is an efficient tool for diagnosing dynamics for such an open quantum system., Comment: 20 pages, 7 figures

Machine learning (ML) models are increasingly used as decision-support tools in high-risk domains. Evaluating the causal impact of deploying such models can be done with a randomized controlled trial (RCT) that randomizes users to ML vs. control groups and assesses the effect on relevant outcomes. However, ML models are inevitably updated over time, and we often lack evidence for the causal impact of these updates. While the causal effect could be repeatedly validated with ongoing RCTs, such experiments are expensive and time-consuming to run. In this work, we present an alternative solution: using only data from a prior RCT, we give conditions under which the causal impact of a new ML model can be precisely bounded or estimated, even if it was not included in the RCT. Our assumptions incorporate two realistic constraints: ML predictions are often deterministic, and their impacts depend on user trust in the model. Based on our analysis, we give recommendations for trial designs that maximize our ability to assess future versions of an ML model. Our hope is that our trial design recommendations will save practitioners time and resources while allowing for quicker deployments of updates to ML models., Comment: 27 pages

The stress interval $S(u,v)$ between $u,v\in V(G)$ is the set of all vertices in a graph $G$ that lie on every shortest $u,v$-path. A set $U \subseteq V(G)$ is stress convex if $S(u,v) \subseteq U$ for any $u,v\in U$. A vertex $v \in V(G)$ is s-extreme if $V(G)-v$ is a stress convex set in $G$. The stress number $sn(G)$ of $G$ is the minimum cardinality of a set $U$ where $\bigcup_{u,v \in U}S(u,v)=V(G)$. The stress hull number $sh(G)$ of $G$ is the minimum cardinality of a set whose stress convex hull is $V(G)$. In this paper, we present many basic properties of stress intervals. We characterize s-extreme vertices of a graph $G$ and construct graphs $G$ with arbitrarily large difference between the number of s-extreme vertices, $sh(G)$ and $sn(G)$. Then we study these three invariants for some special graph families, such as graph products, split graphs, and block graphs. We show that in any split graph $G$, $sh(G)=sn(G)=|Ext_s(G)|$, where $Ext_s(G)$ is the set of s-extreme vertices of $G$. Finally, we show that for $k \in \mathbb{N}$, deciding whether $sn(G) \leq k$ is NP-complete problem, even when restricted to bipartite graphs.

This paper reviews pioneering works in microphone array processing and multichannel speech enhancement, highlighting historical achievements, technological evolution, commercialization aspects, and key challenges. It provides valuable insights into the progression and future direction of these areas. The paper examines foundational developments in microphone array design and optimization, showcasing innovations that improved sound acquisition and enhanced speech intelligibility in noisy and reverberant environments. It then introduces recent advancements and cutting-edge research in the field, particularly the integration of deep learning techniques such as all-neural beamformers. The paper also explores critical applications, discussing their evolution and current state-of-the-art technologies that significantly impact user experience. Finally, the paper outlines future research directions, identifying challenges and potential solutions that could drive further innovation in these fields. By providing a comprehensive overview and forward-looking perspective, this paper aims to inspire ongoing research and contribute to the sustained growth and development of microphone arrays and multichannel speech enhancement., Comment: accepted by ICASSP 2025

Large language models (LLMs) are increasingly applied in educational, clinical, and professional settings, but their tendency for sycophancy -- prioritizing user agreement over independent reasoning -- poses risks to reliability. This study introduces a framework to evaluate sycophantic behavior in ChatGPT-4o, Claude-Sonnet, and Gemini-1.5-Pro across AMPS (mathematics) and MedQuad (medical advice) datasets. Sycophantic behavior was observed in 58.19% of cases, with Gemini exhibiting the highest rate (62.47%) and ChatGPT the lowest (56.71%). Progressive sycophancy, leading to correct answers, occurred in 43.52% of cases, while regressive sycophancy, leading to incorrect answers, was observed in 14.66%. Preemptive rebuttals demonstrated significantly higher sycophancy rates than in-context rebuttals (61.75% vs. 56.52%, $Z=5.87$, $p<0.001$), particularly in computational tasks, where regressive sycophancy increased significantly (preemptive: 8.13%, in-context: 3.54%, $p<0.001$). Simple rebuttals maximized progressive sycophancy ($Z=6.59$, $p<0.001$), while citation-based rebuttals exhibited the highest regressive rates ($Z=6.59$, $p<0.001$). Sycophantic behavior showed high persistence (78.5%, 95% CI: [77.2%, 79.8%]) regardless of context or model. These findings emphasize the risks and opportunities of deploying LLMs in structured and dynamic domains, offering insights into prompt programming and model optimization for safer AI applications., Comment: 10 pages

A defining characteristic of the Kadomstev-Petviashvili (KP) model equations is that the well-posedness results are subject to the restriction that at all transverse positions, the mass $\int u \,dx = \text{constant independent of $y$}.$ In 2007, for a rather general class of equations of KP type, it was shown that the zero-mass (in $x$) constraint is satisfied at any non-zero time even if it is not satisfied at initial time zero. In an effort to remedy this "odd" behavior of the mass of a given solution to the KP Cauchy problem, a model modification is introduced which does not impose non-physical restrictions upon the initial data. In this article, we introduce a new modified KP system. After providing a variational derivation of the model, we analyze its Hamiltonian evolutionary structure. Furthermore, we prove linear estimates in the Bourgain spaces $X^{s,b}$ corresponding to the integral equation arising from the Duhamel formulation of system.

Here we present an efficient, visible-light, gigahertz-frequency acousto-optic modulator fabricated on a 200 mm wafer in a volume CMOS foundry. Our device combines a piezoelectric transducer and a photonic waveguide within a single microstructure that confines both a propagating optical mode and an electrically excitable breathing-mode mechanical resonance. By tuning the device's geometry to optimize the optomechanical interaction, we achieve modulation depths exceeding 2 rad with 15 mW applied microwave power at 2.31 GHz in a 2 mm long device. This corresponds to a modulation figure of merit of $V_{\pi}\cdot L$ = 0.26 Vcm in a visible-light, integrated acousto-optics platform that can be straightforwardly extended to a wide range of optical wavelengths and modulation frequencies. For the important class of gigahertz-frequency modulators that can handle hundreds of milliwatts of visible-light optical power, which are critical for scalable quantum control systems, this represents a 15x decrease in $V_{\pi}$ and a 100x decrease in required microwave power compared to the commercial state-of-the-art and existing work in the literature.

Reinforcement learning (RL) agents typically optimize their policies by performing expensive backward passes to update their network parameters. However, some agents can solve new tasks without updating any parameters by simply conditioning on additional context such as their action-observation histories. This paper surveys work on such behavior, known as in-context reinforcement learning.

Multimodal Large Language Models (MLLMs) represent the cutting edge of AI technology, with DeepSeek models emerging as a leading open-source alternative offering competitive performance to closed-source systems. While these models demonstrate remarkable capabilities, their vision-language integration mechanisms introduce specific vulnerabilities. We implement an adapted embedding manipulation attack on DeepSeek Janus that induces targeted visual hallucinations through systematic optimization of image embeddings. Through extensive experimentation across COCO, DALL-E 3, and SVIT datasets, we achieve hallucination rates of up to 98.0% while maintaining high visual fidelity (SSIM > 0.88) of the manipulated images on open-ended questions. Our analysis demonstrates that both 1B and 7B variants of DeepSeek Janus are susceptible to these attacks, with closed-form evaluation showing consistently higher hallucination rates compared to open-ended questioning. We introduce a novel multi-prompt hallucination detection framework using LLaMA-3.1 8B Instruct for robust evaluation. The implications of these findings are particularly concerning given DeepSeek's open-source nature and widespread deployment potential. This research emphasizes the critical need for embedding-level security measures in MLLM deployment pipelines and contributes to the broader discussion of responsible AI implementation., Comment: 19 pages, 4 figures

The effect of a negative sectoral shock on GDP depends on how important the shocked sector is as a direct and indirect supplier and how easily sectors can substitute inputs. Past estimates of the parameters that determine these qualities in the US have been restrictive: they have not been allowed to vary across industries or across time. This paper uses a novel empirical strategy to relax those restrictions, by exploiting variation in input expenditure share shifts within industries rather than across industries. The resulting estimates exhibit significant sectoral and temporal heterogeneity, and are dynamically correlated with weighted patents. In a calibrated GE model of multi-sector production, this heterogeneity (1) raises[lowers] the GDP effect of negative shocks to sectors whose customers are less[more] able to substitute inputs (e.g. the GDP effect of "Chemical products" shocks rises), (2) raises[lowers] the GDP effect of negative sectoral shocks in years where sectors are less[more] able to substitute inputs, and (3) raises[lowers] the GDP effect of negative shocks to sectors as they become more[less] central input suppliers (e.g. between 1997 and 2023 the GDP effect of "Paper products" shocks fell and the GDP effect of "Computer and electronic products" shocks rose due to changes in their importance as input suppliers)., Comment: 30 pages, 7 tables, 5 figures, 2 appendix tables, 7 appendix figures

Dissipative cat-qubits are a promising architecture for quantum processors due to their built-in quantum error correction. By leveraging two-photon stabilization, they achieve an exponentially suppressed bit-flip error rate as the distance in phase-space between their basis states increases, incurring only a linear increase in phase-flip rate. This property substantially reduces the number of qubits required for fault-tolerant quantum computation. Here, we implement a squeezing deformation of the cat qubit basis states, further extending the bit-flip time while minimally affecting the phase-flip rate. We demonstrate a steep reduction in the bit-flip error rate with increasing mean photon number, characterized by a scaling exponent $\gamma=4.3$, rising by a factor of 74 per added photon. Specifically, we measure bit-flip times of 22 seconds for a phase-flip time of 1.3 $\mu$s in a squeezed cat qubit with an average photon number $\bar{n}=4.1$, a 160-fold improvement in bit-flip time compared to a standard cat. Moreover, we demonstrate a two-fold reduction in $Z$-gate infidelity, with an estimated phase-flip probability of $\epsilon_X = 0.085$ and a bit-flip probability of $\epsilon_Z = 2.65 \cdot 10^{-9}$ which confirms the gate bias-preserving property. This simple yet effective technique enhances cat qubit performances without requiring design modification, moving multi-cat architectures closer to fault-tolerant quantum computation.

We present the optical discovery and multiwavelength follow-up observations of AT2024kmq, a likely tidal disruption event (TDE) associated with a supermassive ($M_{\rm BH}\sim 10^{8} M_\odot$) black hole in a massive galaxy at $z=0.192$. The optical light curve of AT2024kmq exhibits two distinct peaks: an early fast (timescale 1 d) and luminous ($M\approx-20$ mag) red peak, then a slower (timescale 1 month) blue peak with a higher optical luminosity ($M\approx-22$ mag) and featureless optical spectra. The second component is similar to the spectroscopic class of "featureless TDEs" in the literature, and during this second component we detect highly variable, luminous ($L_X\approx 10^{44}$ erg s$^{-1}$), and hard ($f_\nu \propto \nu^{-1.5}$) X-ray emission. Luminous ($10^{29} $erg s$^{-1}$ Hz$^{-1}$ at 10 GHz) but unchanging radio emission likely arises from an underlying active galactic nucleus. The luminosity, timescale, and color of the early red optical peak can be explained by synchrotron emission, or alternatively by thermal emission from material at a large radius ($R\approx\mathrm{few}\times10^{15}$ cm). Possible physical origins for this early red component include an off-axis relativistic jet, and shocks from self-intersecting debris leading to the formation of the accretion disk. Late-time radio observations will help distinguish between the two possibilities., Comment: 23 pages, 7 figures, 6 tables. Submitted to journal on 11 Feb 2025. Comments welcome

We analyze the dynamics of a first order confinement/deconfinement phase transition in an expanding medium using an effective boundary description fitted to the holographic Witten model. We observe and analyze hot plasma remnants, which do not cool down or nucleate bubbles despite the expansion of the system. The appearance of the hot remnants, the dynamics of their shrinking and subsequent dissolution and further heating up is very robust and persists in such diverse scenarios as boost-invariant expansion with a flat Minkowski metric and cosmological expansion in a Friedmann-Robertson-Walker spacetime., Comment: 9 pages, 9 figures

The Transiting Exoplanet Survey Satellite (TESS) mission identified a potential 0.88 REarth planet with a period of 7.577 days, orbiting the nearby M1V star GJ 341 (TOI 741.01). This system has already been observed by the James Webb Space Telescope (JWST) to search for presence of an atmosphere on this planet. Here, we present an in-depth analysis of the GJ 341 system using all available public data. We provide improved parameters for the host star, an updated value of the planet radius, and support the planetary nature of the object (now GJ 341 b). We use 57 HARPS radial velocities to model the magnetic cycle and activity of the host star, and constrain the mass of GJ 341 b to upper limits of 4.0 MEarth (3 sigma) and 2.9 MEarth (1 sigma). We also rule out the presence of additional companions with M sin i > 15.1 MEarth, and P < 1750 days, and the presence of contaminating background objects during the TESS and JWST observations. These results provide key information to aid the interpretation of the recent JWST atmospheric observations and other future observations of this planet., Comment: 18 pages, 17 figures

We present the isovector axial, induced pseudoscalar, and pseudoscalar form factors of the nucleon using three twisted-mass fermion ensembles with degenerate up- and down-, strange-, and charm-quarks with masses tuned to their physical values (physical point). The three ensembles have lattice spacing $a$=0.08, 0.068, and 0.057 fm and approximately equal physical volume allowing for the continuum limit to be taken at the physical point. Excited-state contributions to the matrix elements are evaluated using several sink-source separations from 0.5 fm to 1.5 fm and multistate fits. We check the partially conserved axial-vector current (PCAC) hypothesis and the pion pole dominance (PPD) and show that in the continuum limit both relations are satisfied. We provide results at the continuum limit for the isovector nucleon axial charge, axial radius, pion-nucleon coupling constant, and for the induced pseudoscalar form factor at the muon capture point., Comment: 10 pages, 4 figures. Contributed talk to the 41st International Symposium on Lattice Field theory (LATTICE2024), July 28th - August 3rd, 2024, Liverpool, UK

Recent discoveries of transiting giant exoplanets around M dwarfs (GEMS) present an opportunity to investigate their atmospheric compositions and explore how such massive planets can form around low-mass stars contrary to canonical formation models. Here, we present the first transmission spectra of TOI-5205b, a short-period ($P=1.63~\mathrm{days}$) Jupiter-like planet ($M_p=1.08~\mathrm{M_J}$ and $R_p=0.94~\mathrm{R_J}$) orbiting an M4 dwarf. We obtained three transits using the PRISM mode of the JWST Near Infrared Spectrograph (NIRSpec) spanning $0.6-5.3$ um. Our data reveal significant stellar contamination that is evident in the light curves as spot-crossing events and in the transmission spectra as a larger transit depth at bluer wavelengths. Atmospheric retrievals demonstrate that stellar contamination from unocculted star spots is the dominant component of the transmission spectrum at wavelengths $\lambda\lesssim3.0$ um, which reduced the sensitivity to the presence of clouds or hazes in our models. The degree of stellar contamination also prevented the definitive detection of any $\mathrm{H_2O}$, which has primary absorption features at these shorter wavelengths. The broad wavelength coverage of NIRSpec PRISM enabled a robust detection of $\mathrm{CH_4}$ and $\mathrm{H_2S}$, which have detectable molecular features between $3.0-5.0$ um. Our gridded and Bayesian retrievals consistently favored an atmosphere with both sub-solar metallicity ($\log\mathrm{[M/H]}\sim-2$ for a clear atmosphere) and super-solar C/O ratio ($\log\mathrm{[C/O]}\sim3$ for a clear or cloudy atmosphere). This contrasts with estimates from planetary interior models that predict a bulk metallicity of 10--20%, which is $\sim100\times$ the atmospheric metallicity, and suggests that the planetary interior for TOI-5205b is decoupled from its atmosphere and not well mixed., Comment: 27 pages + appendix, includes figure sets that will appear in journal, submitted to AAS. Comments welcome

Recent studies have revealed magnetically controllable thermoelectric effects in superconductor/ferromagnet (S/F) structures. A tunable cryogenic thermoelectric generator needs not only a high conversion factor between electricity and heat, but also a large change in the thermoelectric output when switching the magnetic state of the device. Here, we experimentally measure and numerically model thermoelectric effects in fully epitaxial F/S/F junctions based on commercially available, easily grown materials, as well as their dependence on the magnetic configuration of the F electrodes. We observe sizeable Seebeck coefficients for the parallel alignment of the ferromagnetic electrodes, reaching values of about $100$~$\mu$V/K. Importantly, we find a decrease of the thermoelectric signal of more than an order of magnitude when switching from a parallel to an antiparallel configuration, constituting a large thermoelectric spin-valve effect. Theoretical modeling based on a self-consistent non-equilibrium Keldysh-Usadel Green function theory, combined with micromagnetic simulations, qualitatively reproduce the experimental findings. These findings pave the way for the development of efficient and versatile cryogenic thermoelectric heat engines., Comment: Submitted for publication (Supplementary material included)

We present Cryoscope -- a new 50 sq. deg field-of-view, 1.2 m aperture, K-dark survey telescope to be located at Dome C, Antarctica. Cryoscope has an innovative optical-thermal design wherein the entire telescope is cryogenically cooled. Cryoscope also explores new detector technology to cost-effectively tile the full focal plane. Leveraging the dark Antarctic sky and minimizing telescope thermal emission, Cryoscope achieves unprecedented deep, wide, fast and red observations, matching and exceeding volumetric survey speeds from the Ultraviolet Explorer, Vera Rubin Observatory, and Nancy Grace Roman Space Telescope. By providing coverage beyond wavelengths of 2 $\mu$m, we aim to create the most comprehensive dynamic movie of the most obscured reaches of the Universe. Cryoscope will be a dedicated discovery engine for electromagnetic emission from coalescing compact binaries, Earth-like exoplanets orbiting cold stars, and multiple facets of time-domain, stellar and solar system science. In this paper, we describe the scientific drivers and technical innovations for this new discovery engine operating in the K-dark passband, why we choose to deploy it in Antarctica, and the status of a fifth-scale prototype designed as a Pathfinder to retire technological risks prior to full-scale implementation., Comment: 36 pages, 20 figures, 4 tables; submitted to PASP on 2025-02-09

Training large neural network models requires extensive computational resources, often distributed across several nodes and accelerators. Recent findings suggest that it may be sufficient to only exchange the fast moving components of the gradients, while accumulating momentum locally (Decoupled Momentum, or DeMo). However, when considering larger models that do not fit on a single accelerate, the exchange of gradient information and the integration of DeMo needs to be reconsidered. Here, we propose employing a hybrid strategy, FlexDeMo, whereby nodes fully synchronize locally between different GPUs and inter-node communication is improved through only using the fast-moving components. This effectively combines previous hybrid sharding strategies with the advantages of decoupled momentum. Our experimental results show that FlexDeMo is on par with AdamW in terms of validation loss, demonstrating its viability.

Large numbers of unstable periodic orbits (UPOs) have been found recently in doubly periodic, two-dimensional Kolmogorov flow at moderate Reynolds numbers $Re \in \{40, 100\}$ using gradient based-optimisation. While these solutions lead to robust statistical reconstructions at the $Re$-values where they were obtained, it is unclear how their dynamical importance evolves as $Re$ is increased. We investigate this by performing arclength continuation of these UPOs to beyond $Re \approx 1000$, and examine how the contribution of each UPO to the reconstruction of statistics evolves with $Re$. Statistical reconstruction away from the $Re$-value where the states were converged is relatively robust for the mean and root-mean-squared velocities, but challenging for complete PDFs, as a large number of solutions rapidly leave the attractor. The UPOs can be grouped into three classes based on whether the limiting dissipation rate is above, beneath or within the range of values typical for the turbulent flow, and we introduce a labelling procedure for identifying dominant vortical interactions in the turbulent UPOs by searching for point vortex periodic orbits that reproduce the dynamics of large scale coherent vortices. Convergence of the point vortex UPOs is achieved via gradient-based optimisation of a scalar loss function which both (1) matches the dynamics of the point vortices to the turbulent vortex cores at $Re=100$ and (2) insist that the point vortex evolution is itself time-periodic. The point vortex UPOs identify a wide range of simpler dynamical processes in the turbulent collection, including slowly-propagating crystals, tripolar structures and bound states. These features persist in the high-dissipation class of UPOs as $Re$ increases, while the role of the forcing diminishes, suggesting a connection to solutions of the Euler equation as $Re \to \infty$.

Understanding the progression trajectories of diseases is crucial for early diagnosis and effective treatment planning. This is especially vital for life-threatening conditions such as Idiopathic Pulmonary Fibrosis (IPF), a chronic, progressive lung disease with a prognosis comparable to many cancers. Computed tomography (CT) imaging has been established as a reliable diagnostic tool for IPF. Accurately predicting future CT scans of early-stage IPF patients can aid in developing better treatment strategies, thereby improving survival outcomes. In this paper, we propose 4D Vector Quantised Generative Adversarial Networks (4D-VQ-GAN), a model capable of generating realistic CT volumes of IPF patients at any time point. The model is trained using a two-stage approach. In the first stage, a 3D-VQ-GAN is trained to reconstruct CT volumes. In the second stage, a Neural Ordinary Differential Equation (ODE) based temporal model is trained to capture the temporal dynamics of the quantised embeddings generated by the encoder in the first stage. We evaluate different configurations of our model for generating longitudinal CT scans and compare the results against ground truth data, both quantitatively and qualitatively. For validation, we conduct survival analysis using imaging biomarkers derived from generated CT scans and achieve a C-index comparable to that of biomarkers derived from the real CT scans. The survival analysis results demonstrate the potential clinical utility inherent to generated longitudinal CT scans, showing that they can reliably predict survival outcomes., Comment: 4D image synthesis, VQ-GAN, neural ODEs, spatial temporal disease progression modelling, CT, IPF

We show that the standard discrete update rule of transformer layers can be naturally interpreted as a forward Euler discretization of a continuous dynamical system. Our Transformer Flow Approximation Theorem demonstrates that, under standard Lipschitz continuity assumptions, token representations converge uniformly to the unique solution of an ODE as the number of layers grows. Moreover, if the underlying mapping satisfies a one-sided Lipschitz condition with a negative constant, the resulting dynamics are contractive, causing perturbations to decay exponentially across layers. Beyond clarifying the empirical stability and expressivity of transformer models, these insights link transformer updates to a broader iterative reasoning framework, suggesting new avenues for accelerated convergence and architectural innovations inspired by dynamical systems theory.

Stable diffusion models represent the state-of-the-art in data synthesis across diverse domains and hold transformative potential for applications in science and engineering, e.g., by facilitating the discovery of novel solutions and simulating systems that are computationally intractable to model explicitly. However, their current utility in these fields is severely limited by an inability to enforce strict adherence to physical laws and domain-specific constraints. Without this grounding, the deployment of such models in critical applications, ranging from material science to safety-critical systems, remains impractical. This paper addresses this fundamental limitation by proposing a novel approach to integrate stable diffusion models with constrained optimization frameworks, enabling them to generate outputs that satisfy stringent physical and functional requirements. We demonstrate the effectiveness of this approach through material science experiments requiring adherence to precise morphometric properties, inverse design problems involving the generation of stress-strain responses using video generation with a simulator in the loop, and safety settings where outputs must avoid copyright infringement.

We prove limiting absorption resolvent bounds for the semiclassical Schr\"odinger operator with a repulsive potential in dimension $n\ge 3$, which may have a singularity at the origin. As an application, we obtain time decay for the weighted energy of the solution to the associated wave equation with a short range repulsive potential and compactly supported initial data., Comment: 21 pages

We describe the equivariant syzygies of the ideal of $2 \times 2$ permanents of a generic $2 \times n$ matrix under its natural symmetric and torus group actions. Our proof gives us a new method of finding the Betti numbers of this ideal, which were first described by Gesmundo, Huang, Schenck, and Weyman., Comment: 11 pages

We determine the nucleon axial, scalar and tensor charges at the continuum limit by analyzing three $N_f=2+1+1$ twisted mass fermion ensembles with all quark masses tuned to approximately their physical values. We include all contributions from valence and sea quarks. We use the Akaike Information Criterion to evaluate systematic errors due to excited states and the continuum extrapolation. For the nucleon isovector axial charge we find $g_A^{u-d}=1.250(24)$, in agreement with the experimental value. We compute the axial, tensor and scalar charges for each quark flavor. The axial charge provides crucial information on the intrinsic spin carried by quark in the nucleon and the the latter two provide input for experimental searches of physics beyond the standard model. Moreover, we extract the nucleon $\sigma$-terms and find $\sigma_{\pi N}=41.9(8.1)$ MeV, for the strange $\sigma_{s}=30(17)$ MeV and for the charm $\sigma_{c}=82(29)$ MeV. We also present preliminary results on the isovector quantities using a fourth ensemble at smaller lattice spacing., Comment: 11 pages, 4 figures, 41th International Symposium on Lattice Field Theory - Lattice2024

Analytics database workloads often contain queries that are executed repeatedly. Existing optimization techniques generally prioritize keeping optimization cost low, normally well below the time it takes to execute a single instance of a query. If a given query is going to be executed thousands of times, could it be worth investing significantly more optimization time? In contrast to traditional online query optimizers, we propose an offline query optimizer that searches a wide variety of plans and incorporates query execution as a primitive. Our offline query optimizer combines variational auto-encoders with Bayesian optimization to find optimized plans for a given query. We compare our technique to the optimal plans possible with PostgreSQL and recent RL-based systems over several datasets, and show that our technique finds faster query plans.

We quantify how squeezed light can reduce quantum measurement noise to levels below the standard quantum limit in impulse measurements with mechanical detectors. The broadband nature of the signal implies that frequency-dependent squeezing performs better than frequency-independent squeezing. We calculate the optimal scaling of the impulse sensitivity with the squeezing strength, and quantify degradations due to photodetection losses. Even for lossless measurement, we find there exists a fundamental limit to the benefit of squeezing that depends only on the system's mechanical properties., Comment: 12 + 6 pages, 8 figures. v2: Figures reformatted, typos corrected

We derive a novel deterministic equivalence for the two-point function of a random matrix resolvent. Using this result, we give a unified derivation of the performance of a wide variety of high-dimensional linear models trained with stochastic gradient descent. This includes high-dimensional linear regression, kernel regression, and random feature models. Our results include previously known asymptotics as well as novel ones.

We study the light meson - charmonium - glueball mixing using flavor-singlet meson operators built from optimal distillation profiles together with purely gluonic operators in different $J^{PC}$ channels at two different pion masses ($\approx$ $420$, $800$ MeV) in two $N_{\rm f} = 3 + 1$ ensembles at close to physical charm quark mass. We observe non-zero mixing correlations between the different types of operators and quantify the overlaps between states created by them and the energy eigenstates by means of a GEVP formulation. We are particularly interested in the scalar glueball and its possible decay into two pions so we also include two-pion operators in our calculation., Comment: 11 pages, 6 figures, 1 table. Proceedings of the 41st International Symposium on Lattice Field Theory (Lattice 2024), University of Liverpool, UK, 2024

Bent monolithic active pixel sensors are the basis for the planned fully cylindrical ultra low material budget tracking detector ITS3 of the ALICE experiment. This paper presents results from testbeam campaigns using high-energy particles to verify the performance of 50 um thick bent ALPIDE chips in terms of efficiency and spatial resolution. The sensors were bent to radii of 18, 24 and 30 mm, slightly smaller than the foreseen bending radii of the future ALICE ITS3 layers. An efficiency larger than $99.9\%$ and a spatial resolution of approximately 5 um, in line with the nominal operation of flat ALPIDE sensors, is obtained at nominal operating conditions. These values are found to be independent of the bending radius and thus constitute an additional milestone in the demonstration of the feasibility of the planned ITS3 detector. In addition, a special geometry in which the beam particles graze the chip and traverse it laterally over distances of up to 3 mm is investigated.

While transformer-based models dominate NLP and vision applications, their underlying mechanisms to map the input space to the label space semantically are not well understood. In this paper, we study the sources of known representation vulnerabilities of vision transformers (ViT), where perceptually identical images can have very different representations and semantically unrelated images can have the same representation. Our analysis indicates that imperceptible changes to the input can result in significant representation changes, particularly in later layers, suggesting potential instabilities in the performance of ViTs. Our comprehensive study reveals that adversarial effects, while subtle in early layers, propagate and amplify through the network, becoming most pronounced in middle to late layers. This insight motivates the development of NeuroShield-ViT, a novel defense mechanism that strategically neutralizes vulnerable neurons in earlier layers to prevent the cascade of adversarial effects. We demonstrate NeuroShield-ViT's effectiveness across various attacks, particularly excelling against strong iterative attacks, and showcase its remarkable zero-shot generalization capabilities. Without fine-tuning, our method achieves a competitive accuracy of 77.8% on adversarial examples, surpassing conventional robustness methods. Our results shed new light on how adversarial effects propagate through ViT layers, while providing a promising approach to enhance the robustness of vision transformers against adversarial attacks. Additionally, they provide a promising approach to enhance the robustness of vision transformers against adversarial attacks., Comment: 10 pages, 5 figures

Alignment of large language models (LLMs) with human values and preferences, often achieved through fine-tuning based on human feedback, is essential for ensuring safe and responsible AI behaviors. However, the process typically requires substantial data and computation resources. Recent studies have revealed that alignment might be attainable at lower costs through simpler methods, such as in-context learning. This leads to the question: Is alignment predominantly superficial? In this paper, we delve into this question and provide a quantitative analysis. We formalize the concept of superficial knowledge, defining it as knowledge that can be acquired through easily token restyling, without affecting the model's ability to capture underlying causal relationships between tokens. We propose a method to extract and isolate superficial knowledge from aligned models, focusing on the shallow modifications to the final token selection process. By comparing models augmented only with superficial knowledge to fully aligned models, we quantify the superficial portion of alignment. Our findings reveal that while superficial knowledge constitutes a significant portion of alignment, particularly in safety and detoxification tasks, it is not the whole story. Tasks requiring reasoning and contextual understanding still rely on deeper knowledge. Additionally, we demonstrate two practical advantages of isolated superficial knowledge: (1) it can be transferred between models, enabling efficient offsite alignment of larger models using extracted superficial knowledge from smaller models, and (2) it is recoverable, allowing for the restoration of alignment in compromised models without sacrificing performance.

Diffusion models have been used in cosmological applications as a generative model for fast simulations and to reconstruct underlying cosmological fields or astrophysical images from noisy data. These two tasks are often treated as separate: diffusion models trained for one purpose do not generalize to perform the other task. In this paper, we develop a single diffusion model that can be used for both tasks. By using the Diffusion Posterior Sampling (DPS) approach, we use a diffusion model trained to simulate weak lensing maps for the inverse problem of reconstructing mass maps from noisy weak lensing data. We find that the standard DPS method leads to biased inference but we correct this bias by down weighting the likelihood term at early sampling time steps of the diffusion. Our method give us a way to reconstruct accurate high-resolution (sub-arcminute) mass maps that have the correct power spectrum and a range of non-Gaussian summary statistics. We discuss several applications enabled by the computational efficiency and accuracy of our model. These include generation of simulation quality mass maps, aiding covariance estimation for higher order statistics, and for finding filaments, voids and clusters from noisy lensing shear data., Comment: 14 pages, 9 figures, Comments welcome

We present a validation dataset of newly-collected kitchen-based egocentric videos, manually annotated with highly detailed and interconnected ground-truth labels covering: recipe steps, fine-grained actions, ingredients with nutritional values, moving objects, and audio annotations. Importantly, all annotations are grounded in 3D through digital twinning of the scene, fixtures, object locations, and primed with gaze. Footage is collected from unscripted recordings in diverse home environments, making HDEPIC the first dataset collected in-the-wild but with detailed annotations matching those in controlled lab environments. We show the potential of our highly-detailed annotations through a challenging VQA benchmark of 26K questions assessing the capability to recognise recipes, ingredients, nutrition, fine-grained actions, 3D perception, object motion, and gaze direction. The powerful long-context Gemini Pro only achieves 38.5% on this benchmark, showcasing its difficulty and highlighting shortcomings in current VLMs. We additionally assess action recognition, sound recognition, and long-term video-object segmentation on HD-EPIC. HD-EPIC is 41 hours of video in 9 kitchens with digital twins of 413 kitchen fixtures, capturing 69 recipes, 59K fine-grained actions, 51K audio events, 20K object movements and 37K object masks lifted to 3D. On average, we have 263 annotations per minute of our unscripted videos., Comment: 29 pages. Project Webpage and Dataset: http://hd-epic.github.io

Model selection and order selection problems frequently arise in statistical practice. A popular approach to addressing these problems in the frequentist setting involves information criteria based on penalized maxima of log-likelihoods for competing models. In the Bayesian context, similar criteria are employed, replacing the maxima of log-likelihoods with their posterior expectations. Despite their popularity in applications, the large-sample behavior of these criteria -- such as the deviance information criterion (DIC), Bayesian predictive information criterion (BPIC), and widely-applicable Bayesian information criterion (WBIC) -- has received relatively little attention. In this work, we investigate the almost sure limits of these criteria and establish novel results on posterior and generalized posterior consistency, which are of independent interest. The utility of our theoretical findings is demonstrated via illustrative technical and numerical examples.

We introduce a unified framework for iterative reasoning that leverages non-Euclidean geometry via Bregman divergences, higher-order operator averaging, and adaptive feedback mechanisms. Our analysis establishes that, under mild smoothness and contractivity assumptions, a generalized update scheme not only unifies classical methods such as mirror descent and dynamic programming but also captures modern chain-of-thought reasoning processes in large language models. In particular, we prove that our accelerated iterative update achieves an $O(1/t^2)$ convergence rate in the absence of persistent perturbations, and we further demonstrate that feedback (iterative) architectures are necessary to approximate certain fixed-point functions efficiently. These theoretical insights bridge classical acceleration techniques with contemporary applications in neural computation and optimization.

We propose a multi-scale deep energy model that is strongly convex in the local neighbourhood around the data manifold to represent its probability density, with application in inverse problems. In particular, we represent the negative log-prior as a multi-scale energy model parameterized by a Convolutional Neural Network (CNN). We restrict the gradient of the CNN to be locally monotone, which constrains the model as a Locally Convex Multi-Scale Energy (LC-MuSE). We use the learned energy model in image-based inverse problems, where the formulation offers several desirable properties: i) uniqueness of the solution, ii) convergence guarantees to a minimum of the inverse problem, and iii) robustness to input perturbations. In the context of parallel Magnetic Resonance (MR) image reconstruction, we show that the proposed method performs better than the state-of-the-art convex regularizers, while the performance is comparable to plug-and-play regularizers and end-to-end trained methods.

The mutation process in evolution strategies has been interlinked with the normal distribution since its inception. Many lines of reasoning have been given for this strong dependency, ranging from maximum entropy arguments to the need for isotropy. However, some theoretical results suggest that other distributions might lead to similar local convergence properties. This paper empirically shows that a wide range of evolutionary strategies, from the (1+1)-ES to CMA-ES, show comparable optimization performance when using a mutation distribution other than the standard Gaussian. Replacing it with, e.g., uniformly distributed mutations, does not deteriorate the performance of ES, when using the default adaptation mechanism for the strategy parameters. We observe that these results hold not only for the sphere model but also for a wider range of benchmark problems.

We investigate various positivity properties of line bundles on general blow ups of Hirzebruch surfaces motivated by \cite{Han}, where the author has studied general blow ups of $\mathbb{P}^2$. For each of the properties: ampleness, global generation, very ampleness, and $k$-very ampleness, we provide several sufficient numerical conditions., Comment: Comments are welcome