Peer Revenue Returns subject to peer review

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Monday 31 August

Computational Finance

Single- and Multilevel Quadrature with Error Control for Fourier Pricing under the Rough Heston Model

Unlike the classical Heston model, Fourier pricing under the rough Heston model requires solving a fractional Riccati equation at every quadrature point. Since the required resolution varies with model parameters and quadrature point, a single uniform time discretization can be inefficient. We develop single- and multilevel Gauss-Laguerre quadrature methods that balance the time discretization and Fourier quadrature errors. Both methods scale the laguerre weight to the estimated Fourier integrand decay. The single-level method allocates a prescribed tolerance between the two errors. The multilevel method splits the integrand into a level-zero term and level differences, selecting quadrature points separately at each level. Suppose that the Fourier integrand discretization error is $O(\Delta t^p)$, that evaluating the characteristic function once costs $O(\Delta t^{-\beta})$, and that the algebraic Gauss-Laguerre quadrature error is $O(N^{-s_{SL}/2})$, where $s_{{SL}}$ is the smoothness index. Under this estimate and assumptions on the regularity and decay of level differences, we prove that the proposed single-level method requires $O(\epsilon^{-(\beta/p+2/s_{{SL}})})$ computational work to achieve accuracy $\epsilon$, whereas the proposed multilevel method requires $O(\epsilon^{-\beta/p})$ computational work. We also study root-exponential Gauss-Laguerre error models for practical multilevel quadrature allocation. Numerical experiments support the observed fractional Riccati and Fourier integrand convergence rates and root-exponential quadrature behavior, and show substantial reductions in quadrature cost from the proposed scaling. The multilevel method provides clear computational savings over the single-level method. We further benchmark the multilevel fractional Riccati method against the BL2 Markovian approximation and report lower total CPU time in the tested configurations.
Read the full abstract

All authors Chiheb Ben Hammouda, Abderrahmene Ben Romdhane, Michael Samet, Raul F. Tempone

Astrophysics of Galaxies

Formation of Heavy Seed Black Holes and Little Red Dots-like Compact Clusters in Metal-enriched Star-forming Regions

The origin of supermassive black holes (SMBHs) in the early Universe remains uncertain. While the classical direct-collapse scenario requires pristine gas exposed to intense Lyman-Werner (LW) radiation, recent JWST observations suggest that early growing BHs may reside in compact, metal-enriched environments. We investigate early BH formation using a cosmological radiation-hydrodynamic simulation that self-consistently follows the formation and evolution of rapidly accreting supermassive stars (SMSs). We find that intense LW radiation develops in clustered star-forming regions; by the time the first heavy seed forms, the mass-weighted internal intensity has reached $J_{21}\sim1000$. Most seeds form in weakly metal-enriched gas with [Z/H]$\sim-3$ to $-2$, because the strongest LW fields occur in regions already enriched by previous stellar feedback. All halos above $10^9M_\odot$ in our zoom-in region host a BH more massive than $10^5M_\odot$. We construct stellar spectra using MIST/PHOENIX libraries for normal stars and $6000~\mathrm{K}$ blackbody spectra for bloated SMSs. Five of eight massive clusters with stellar masses larger than $10^6~M_\odot$ exhibit intrinsically V-shaped spectra similar to those of Little Red Dots (LRDs). In the most prominent cases, luminous SMSs dominate the red optical emission, although Balmer breaks in normal stellar populations also contribute. We also identify off-center LRD-like systems in our simulation resembling those recently reported in strongly lensed observations. Our results suggest that heavy-seed formation, compact-cluster formation, and LRD-like continuum emission are linked outcomes of clustered, weakly metal-enriched, strongly irradiated environments.
Read the full abstract

All authors Sunmyon Chon, Shingo Hirano, Ke-Jung Chen, Volker Springel

Signal Processing

Physiological Information Reliability: Cross-Layer Adaptive Resource Allocation for Cardiovascular Sensing

Cardiovascular sensing systems must preserve clinically useful information despite signal degradation, wireless losses, energy constraints, and edge-computation latency. We introduce Physiological Information Reliability (PIR), a cross-layer framework that represents physiological information value jointly with wireless, energy, and computation states and uses a contextual bandit to adapt sensing and communication decisions. We integrate multimodal ECG/PPG signal-quality estimation with physiological information value and an adaptive network-coding layer under burst-erasure conditions. Across controlled multiseed experiments, PIR-LinUCB demonstrates a promising low-energy operating point while maintaining medical latency constraints and competitive physiological estimation performance relative to fixed and heuristic policies. We analyze the resulting accuracy-energy-latency trade-offs and identify limitations of proxy PIV estimation and simulated communication dynamics. These results provide an initial computational demonstration of physiological-information-aware resource allocation and motivate future clinical and real-channel validation.
Read the full abstract

Artificial Intelligence

SAGE: State-Grounded, Abstention-Aware Evaluation of Task-Oriented Dialogue Agents

Evaluating task-oriented dialogue agents requires judging not merely whether a reply reads well but whether each turn advances the underlying workflow state correctly–a distinction conventional holistic LLM judges can miss because they evaluate the available context as a single unit and require one or more full-model calls per turn. We propose SAGE (State-Grounded Abstention-Aware Evaluation), which compiles a workflow specification and per-turn state diff into atomic, schema-grounded criteria and routes each through a cascade of symbolic and encoder/NLI verifiers that abstain rather than guess, aggregating criterion verdicts into a turn-level decision with an evidence trace. Its recommended operating point, SAGE-Core, decides 81–91% of criteria with only the compiler, symbolic rules, and on-device encoders–at zero paid LLM cost–while SAGE-LLM adds an optional focused-LLM fallback for open-class criteria. Across four slices spanning MultiWOZ, Schema-Guided Dialogue, and ABCD, no evaluated LLM-as-a-judge baseline–including a state-aware GPT-4.1 judge and cheaper GPT-4.1-mini variants–significantly exceeds SAGE-Core on any slice, even though the GPT-4.1 G-Eval judge costs $4.7--8.0 per 1,000 turns to SAGE-Core's $0. A two-annotator human audit (n=200, $\kappa$=0.94) confirms strong label fidelity on the transcript-visible failure classes–where, excluding the weak-salience IUV class, SAGE-Core is statistically tied with the strongest LLM judge–and honestly scopes ignored-user-value as a state-consistency signal with weak broad-human salience. We analyze construct-validity limits from injected failures and partial symbolic circularity.
Read the full abstract

Strongly Correlated Electrons

Bound states, resonances, and their thermodynamic properties in pseudospin-1 systems with short-range impurities

Bound states and resonances induced by short-range impurities modeled by circular potential wells are analyzed in the vicinity of flat and dispersive bands in gapped and gapless pseudospin-1 systems. We find that the bound and resonant states derived from the flat band show unusual characteristics originating from the multicomponent structure of pseudospin-1 fermions, which are distinct from those for pseudospin-$\tfrac{1}{2}$ fermions. Contrary to gapped Dirac systems and unlike bound states in the vicinity of the upper dispersive band, the bound states derived from the flat band occur for any value of the total angular momentum. The energies of these bound states with higher angular momentum $j$ tend to decrease with $|j|$. In addition, it is found that their wave functions are localized at the potential well edge and the localization increases with $|j|$. The signatures of the impurity states in the local density of states are determined. Using the Anderson model for independent electrons in the disorder potential, the thermodynamic potential, entropy density, and heat capacity are obtained. In the regime dominated by bound states derived from the flat band, the entropy density monotonically increases with temperature and saturates, whereas the heat capacity exhibits a single maximum.
Read the full abstract

Optics

Strong suppression of the photonic density of states in three-dimensional disordered silicon networks

Photonic bandgaps can open not only in crystalline dielectric materials but also in amorphous structures. Self-uniform amorphous gyroid networks have been proposed as promising disordered photonic architectures. Motivated by numerical studies, we combine direct laser-writing lithography with advanced materials processing to fabricate these structures from high-refractive-index silicon. Spectroscopic measurements at mid-infrared wavelengths reveal pronounced transmission minima. To investigate the formation of photonic band gaps in amorphous structures, we develop polarization-resolved transmission spectroscopy that separates ballistic and multiply scattered contributions, enabling direct identification of the underlying transport regimes. We observe a strong suppression of diffuse transmission, providing direct experimental evidence for a breakdown of conventional diffuse transport associated with a strongly reduced photonic density of states. Together with large-scale numerical simulations, our results establish the presence of a deep photonic pseudogap in an amorphous three-dimensional dielectric material and open new opportunities for observing disorder-induced localization phenomena, including Anderson localization of light.
Read the full abstract

All authors Abraham Aguilar Uribe, Francisco Hernández Alejandre, Mattis Reisner, Geoffroy Aubry, Luis S. Froufe-Pérez, Marian Florescu, Frank Scheffold

Audio and Speech Processing

U-PAST: A Phase-Aware Audio Spectrogram Transformer-U-Net for Single-Channel Speech Enhancement

Convolutional neural networks (CNNs), used widely and successfully in audio enhancement, capture long-range time-frequency dependencies only indirectly, through successive convolution and pooling. Here, we present U-PAST, a hybrid transformer-U-Net architecture that addresses this limitation through self-attention dependency-modeling in the complex spectrogram domain. U-PAST tokenizes a complex STFT representation, similarly to the magnitude spectrogram tokenization of the Audio Spectrogram Transformer (AST), applies a multi-layer transformer encoder, and reconstructs the enhanced complex spectrogram with a U-Net-style decoder. We evaluate four architectural variants with between 1.17M and 2.40M parameters on the DNS Challenge, VoiceBank-DEMAND, and LibriMix corpora under matched, acoustic mismatch, and two-dataset mismatch conditions. U-PAST attains the best SI-SDR of any evaluated model under acoustic mismatch and closely trails substantially larger convolutional and time-domain baselines by 0.26 dB to 0.63 dB SI-SDR under the remaining three conditions while achieving the strongest perceptual (DNSMOS) quality under dataset mismatch. The largest evaluated configuration, U-PAST-H (2.40M parameters), is consistently the strongest variant of the family, offering an attractive performance-to-cost trade-off at a small parameter footprint.
Read the full abstract

Cryptography and Security

Don't Trust the Code, Check Its Effects: Runtime Refinement for Regenerated Systems Code Under an Adversarial Generator

Recent work uses large language models to generate systems code from specifications, treating the specification as the durable artifact and the implementation as disposable. Regenerating the implementation specializes it to each workload and device. However, that work lives in a forgiving setting: a component's externally visible effects, its writes and device commands, are recoverable, and the generator is honest, so trust is discharged by re-execution. We target the unforgiving setting: systems code whose effects are irreversible, produced by a generator that may be adversarial. There, re-execution cannot check an effect after the fact, and a proof fails silently when its assumptions do. We take the position that the only safe way to operate here is to deny the generated code the authority to act. The generated code only plans, while a fixed trusted mediator owns every effect and performs one only when the specification would have produced it. Because the guarantee lives in the mediator, not the code, it survives regeneration. We instantiate this as a reference monitor for regenerated device drivers, and characterize the mediability envelope, six conditions on the effect vocabulary: legibility, spec-input observability, correlatability, completeness, outcome enumerability, and explicit durability. They decide when such mediation is possible.
Read the full abstract

Atmospheric and Oceanic Physics

Reconciling the Lack of a Robust Anvil Cloud Amount Response to Warming

The stability-iris hypothesis proposes that tropical anvil cloud amount decreases with warming as anvils rise into environments of greater static stability. Here we show that there is no substantial theoretical or simulated link from increased static stability to reduced anvil amount. We demonstrate analytically that the sensitivities of static stability and pressure velocity to surface warming arise mainly from decreasing air density, which does not weaken the detrainment supplying anvils. Convectively driven motion stays nearly fixed along isotherms, so detrainment changes little where anvils rise. An ensemble of RCEMIP simulations confirms that stability increases robustly but is not accompanied by a robust reduction in the detrainment of air or water at anvil levels. Overall, these findings dispel the expectation of a systematic reduction in anvil amount with warming arising from basic thermodynamic arguments.
Read the full abstract

All authors Zachary McGraw, Blaž Gasparini, Wouter Mol, Aiko Voigt

Earth and Planetary Astrophysics

Accelerating Chemical Kinetics for Exoplanet Atmospheres using Neural Networks

Observations increasingly reveal the coupled radiative, chemical, and dynamical processes that shape exoplanet atmospheres. Interpreting these atmospheres requires models that can capture this complexity. However, multidimensional models remain fundamentally limited by computational cost, and answering key questions requires simulating the governing physical mechanisms at speeds classical methods cannot achieve. As a result, models often rely on simplifying approximations, such as equilibrium chemistry, even when those assumptions miss important effects. There is a pressing need for fast and accurate chemical kinetics solvers to model planetary atmospheres. Here we present a machine learning local-box chemical kinetics solver for exoplanet atmospheres using a residual flow-map architecture. We demonstrate that this surrogate model is several orders of magnitude faster than a classical solver, achieving microsecond-scale inference while retaining percent-level accuracy. The surrogate model covers a parameter space that spans $T=300$-$3000$ K, $P=10^{-6}$-$10^{4}$ bar, $\Delta t=10^{-3}$-$10^{8}$ s, and compositions ranging from $10^{-2}$ to $10^{3}$ times solar in both C/O ratio and metallicity. Our model outperforms several commonly used machine learning architectures and performs robustly under the extreme stiffness characteristic of atmospheric chemistry. The machine learning framework presented here is a flexible and efficient approach to emulating state-to-state flow-map problems that commonly arise in numerical simulations.
Read the full abstract

All authors Isaac Malsky, Xi Zhang, Tiffany Kataria, Matthew Graham, Ziyu Huang, Boris Bonev, Shang-Min Tsai, Elspeth K.H. Lee

Artificial Intelligence

SpecMind: Enabling Spectrum Intelligence via Multi-Agent Hybrid Retrieval-Augmented Generation

The exponential growth of wireless devices is driving unprecedented spectrum demand, pushing spectrum management toward more fine-grained decisions across space, time, and device constraints. As a result, spectrum policymakers and engineers must process large volumes of data that come from diverse sources and take many different forms, such as text and tables. These data sources are often disaggregated and require significant time and effort to integrate, search, and interpret. Furthermore, most of this information is formatted for human understanding and is not readily accessible to automated systems. To address this challenge, we propose SpecMind, a novel Multi-Agent Retrieval-Augmented Generation (RAG) system for spectrum intelligence that performs reasoning over heterogeneous data sources. This system enables autonomous agents to coordinate specialized sub-agents that retrieve and synthesize knowledge across policy proceedings, legal regulations, and license databases. We develop SpecBench, a question and answer (Q&A) dataset based on real-world license records and policy proceedings, addressing the lack of evaluation resources for RAG systems in the spectrum domain. Experimental results demonstrate that SpecMind outperforms traditional, general-purpose RAG systems across spectrum-related tasks, achieving over 80% win rate against strong baselines. The agent-based design enables more accurate retrieval, better contextual reasoning, and improved task completion across diverse query types.
Read the full abstract

All authors Songwei Dong, Bingyan Lu, Makayla Kienlen, J. Nicholas Laneman, Cong Shen

Solar and Stellar Astrophysics

The Formation of Very Close Binary Stars

The majority of very close binaries (orbital periods P < 5 days) are in triples. Previous studies surmised that the tertiary hardened the inner binary via Kozai-Lidov cycles coupled with tidal friction. Alternatively, the trend could be due to a common origin, e.g., more massive molecular cores and protostellar disks are more likely to fragment twice into a triple and harden the inner binary to very short periods via disk migration. Disk migration occurs during the embedded Class I protostellar phase while secular evolution with a tertiary occurs on much longer timescales. To determine the dominant migration mechanism, I accurately measure the very close binary fractions of both late-type and early-type primaries in star-forming environments. After combining five surveys, the late-type Class II/III T Tauri very close binary fraction is F_VCB = 2.1_{-0.5}^{+1.0}%, which is consistent with the measurement for late-type field main-sequence stars (1.7% +/- 0.1%). More than 85% of very close late-type binaries hardened via disk migration during the Class I protostellar phase. I next measure the OGLE-III eclipsing binary (EB) fractions of young massive stars in the LMC. The very close EB fractions of massive young stellar objects (YSOs; F_VCEB = 8.3_{-3.2}^{+4.9}%), OB stars near massive YSOs (4.7% +/- 0.8%), and OB stars in small, young H II regions (6.4% +/- 1.5%) are consistent with each other and the measurement for OB stars in slightly older clusters (4.7% +/- 0.3%). More than 90% of very close massive binaries hardened via disk migration during the embedded protostellar phase.
Read the full abstract

Functional Analysis

Individual and Uniform Eventual Positivity of Self-Adjoint Semigroups

We construct a real self-adjoint semigroup on an $\ell^2$-space which is individually eventually strongly positive with respect to a strictly positive vector, but whose operators fail to be positive at arbitrarily large times. Moreover, the generator of this semigroup has compact resolvent. Consequently, individual and uniform eventual positivity are not equivalent for self-adjoint semigroups on $L^2$-spaces, even under the additional assumption that the generator has compact resolvent. The construction consists of a positive self-adjoint rank-one perturbation of a diagonal operator and a sequence of small spectral shifts on antisymmetric two-point modes.
Read the full abstract

Computational Engineering

MUFASA: An Information Utility-Aware Preprocessing Framework for Reliable Model Reasoning in Computational Pathology

Reliable computational pathology depends on preprocessing methods that identify informative tissue regions while excluding artifacts and low-utility regions from whole-slide images (WSI). However, existing preprocessing pipelines often retain such regions or discard diagnostically relevant tissue, thereby limiting downstream model performance, reliability, and robustness across heterogeneous cohorts. Here, we systematically evaluate how these regions affect downstream AI model performance across multiple clinically relevant applications and introduce MUFASA, a generalizable, information utility-aware preprocessing framework for H&E-stained WSI that excludes artifacts and low-utility regions while preserving biologically meaningful tissue. MUFASA integrates slide-level artifact masking, stain-aware tile filtering, reconstruction-based utility stratification of tiles, and targeted recovery of tissue tiles that are over-filtered by earlier phases. Across tumor diagnosis, tumor subtyping, biomarker status prediction, and survival prognostication tasks in diverse cancer cohorts, MUFASA consistently improves downstream model performance relative to widely used preprocessing baselines. These gains are accompanied by reduced artifact-associated attribution in model heatmaps, indicating improved alignment between retained tissue and model attention. Our findings establish WSI preprocessing as a critical determinant of downstream model performance and validity, revealing that even accurate predictions can conceal important failure modes stemming from anatomically implausible reasoning driven by retained artifact-containing and low information-utility tiles.
Read the full abstract

All authors Rathinaraja Jeyaraj, Barathi Subramanian, Songmi Noh, Mitchell N. Peterson, Terry Guo, George A. Fisher, Nigam H. Shah, Curtis P. Langlotz, Thomas J. Montine, Jeanne Shen

General Relativity and Quantum Cosmology

Computing the energy of some timelike singularities

The gravitational energies of timelike singularities in some asymptotically AdS spaces and in the Schwarzschild black hole with negative mass are explicitly computed. In all cases considered, the energy is finite and positive, depends on the geometry around the singularity, and compensates negative energy contributions from the asymptotic boundary, rendering the total gravitational energy positive. The definition used for energy is the on-shell Hamiltonian of the physical classical action proposed by Hawking and Horowitz~\cite{Hawking-Horowitz}. To deal with the singularity, spacetime is cut off by a timelike regulating boundary surrounding the singularity that is compatible with a foliation of spacetime in constant-time slices. It is shown that the total energy of static AdS-Kasner solutions with timelike Kasner singularities is equal to the energy of the planar AdS black hole, and that the Schwarzschild black hole with negative mass has the same energy as Minkowski space.
Read the full abstract

Logic in Computer Science

Exponential Gaps Between Intuitionistic Linear Extended Frege Systems

In this paper, we establish exponential separations between Extended Frege systems for a range of intuitionistic substructural and linear logics. More precisely, for any logic $L$ below the intuitionistic logic obtained by extending $\mathbf{ILL}$ with structural rules, and any logic $M$ not contained in $L$, we construct a family of $\mathsf{FL_e}$-provable formulas that have short proofs in $M$-Frege but require proofs of exponential size in $L$-Extended Frege. The same result holds in the $!$-free settings, using $\mathbf{IMALL}$ and $\mathbf{FL_e}$ in place of $\mathbf{ILL}$. The key ingredient in proving these separations is a variant of the feasible disjunction property for $L$-Frege, which may be of independent interest.
Read the full abstract

Hardware Architecture

VARA: A Voltage-Aware ReRAM-Based Accelerator for Energy-Efficient Computing

ReRAM-based in-memory computing (IMC) architectures are widely regarded as a promising approach to alleviating the computational bottleneck of conventional architectures. Since ReRAM crossbars perform matrix-vector multiplication (MVM) in the analog domain, their computational energy consumption is highly dependent on weight and activation distributions. However, most existing ReRAM accelerators focus primarily on weight optimization while paying limited attention to the impact of activations on computational energy consumption, leaving the energy-saving potential of activation sparsity largely underexploited. In this paper, we propose a voltage-aware ReRAM-based accelerator (VARA), along with its accompanying design methodology. Specifically, we first introduce a voltage-aware training (VAT) algorithm that incorporates a preset threshold into the activation function to steer the activation distribution toward zero values, thereby enhancing activation sparsity. Building upon this, we further propose a co-zero activation reordering (CAR) scheme for crossbar-level computation skipping. CAR clusters activation dimensions based on their co-zero correlations and consistently reorders both the activation matrix and its corresponding weights. This process consolidates scattered zero activations into contiguous zero-valued regions to maximize the benefits of crossbar-level computation skipping. Extensive experimental results demonstrate that, with only marginal accuracy loss, VARA reduces the average total system energy consumption by 60.12% and improves the average system energy efficiency by 2.68$\times$ compared to the baseline, outperforming existing state-of-the-art accelerators for sparse-activation optimization.
Read the full abstract

Machine Learning

How Temporal Correlations Shape Memory in Linear Recurrent Neural Networks

The linear recurrent neural network (LRNN) is a simple model for studying how much memory a network builds up as it trains. For uncorrelated inputs, earlier work found that training itself settles the network between keeping the past and reacting only to the present. Real sequences are correlated, and we solve the learning dynamics exactly for correlated inputs. In the solution, keeping the past carries a cost. The whole effect of correlation lands on that cost. This cost reduces to the earlier one when inputs are uncorrelated and grows once they are positively correlated. Three findings follow. (1) Correlation reshapes the course of learning, not only its end. Memory builds, overshoots, and is partly removed, and the settled network keeps less of the past. (2) Memory switches off at a threshold set by one number, how much each input resembles the one just before it. Neither sequence length nor longer-range correlation moves this threshold. Memory is worth keeping only when the task needs the previous input more than the current input already supplies it through correlation with the past. (3) The best network changes too. Zero error demands a feedthrough, a path that passes the current input straight to the network's output and remembers nothing, and training builds it unprompted when given one spare hidden dimension. Our work turns one property of the input into a prediction of whether a network learns memory and explains why correlated data turns recurrent networks into change detectors.
Read the full abstract

Logic

Every Nonrecursive Many-One Degree Contains Either One or Infinitely Many Finite-One Degrees

This paper proves that every nonrecursive many-one degree contains either exactly one or infinitely many finite-one degrees. This gives a negative answer to Open Question 2 of Richter, Stephan, and Zhang~\cite[p.~17]{RSZ}. Earlier work established that, for almost every $A\subseteq\N$ with respect to the standard product measure on $2^\N$, the degree $[A]_{\m}$ contains an infinite antichain of finite-one degrees~\cite{Cintioli}. Thus the question had already received an almost-sure negative answer; the present theorem settles it for every nonrecursive many-one degree.
Read the full abstract

Group Theory

Relative property (T) for uniformly bounded representations

For pairs $(G,H)$, where $G$ is a locally compact group and $H$ is a closed subgroup of $G$, we introduce analogous versions of the full group $\mathrm{C}^*$-algebra and the Fourier–Stieltjes algebra for uniformly bounded representations. We use these objects to characterise relative property (T) in this more general setting in terms of Kazhdan projections and invariant means. We also show that, for semidirect products of the form $G=H\ltimes N$, where $N$ is a nilpotent group, relative property (T) for the pair $(G,N)$ is equivalent to its uniformly bounded version. We first prove this result when $N$ is abelian, and then proceed by studying the behaviour of relative property (T) under quotients by central subgroups, thereby extending a theorem of Serre to the uniformly bounded setting.
Read the full abstract

Computation and Language

Late Transformer Layers Recode Syntax Canonically: Evidence from Greek Scrambling and Cross-Layer Generalisation

Probing studies have established that syntactic information is decodable in early and middle transformer layers, but what happens to that information in later layers remains poorly understood. We apply a cross-layer generalisation analysis to three Greek-tuned large language models evaluated on tightly controlled minimal pairs: object-relative constructions in Modern Greek, where canonical (Subject-Verb-Object; SVO) and non-canonical (Verb-Subject-Object; VSO) orders differ only in within-clause word order, while preserving propositional meaning. When a probe trained on late layers (20-31) is tested on each early layer individually, it produces below-chance transfer (cluster-corrected, p<0.01), classifying 99.3% of non-canonical sentences as canonical. Probe coefficients reverse sign around layer 22, indicating a directional recoding toward the canonical form rather than simple information loss. These findings characterise a representational format change in late transformer layers that goes beyond the well-established decline in syntactic decodability, and they generate a directly testable prediction for human EEG and MEG decoding studies using the same stimuli. Code and stimuli are publicly available on OSF.
Read the full abstract

All authors Christos Nikolaos Zacharopoulos, Revekka Kyriakoglou, Chara Tsoukala, Théo Desbordes

Analysis of PDEs

Separation for Schrödinger operators. A counterexample to Simon's conjecture

We construct, for every $d\ge 5$, a nonnegative radial potential $V\in L^\infty_{\mathrm{loc}}(\mathbb{R}^d\setminus\{0\})\cap L^1(\mathbb{R}^d)$ for which $V(-\Delta\dotplus V)^{-1}$ is not bounded on $L^2(\mathbb{R}^d)$.

Computer Vision

Unmasking Face Embeddings: Reading, Rendering and Naming with Foundation Models

Modern face recognition (FR) owes much of its success to deep neural networks that learn to extract compact identity embeddings from face images. These models are typically trained for identity discrimination, producing embeddings that are highly effective for biometric matching but largely opaque to semantic interpretation. In contrast, foundation models, pretrained on broad visual or vision–language tasks, provide rich interfaces for describing, retrieving, generating, and organizing visual content. This contrast raises a natural question: what capabilities become available when face embeddings from domain-specific FR models are made interoperable with foundation models? Building on recent work on embedding compatibility across models, we use simple pre-computed linear transformations, estimated from paired embeddings alone, to connect existing FR models with off-the-shelf foundation models. Once aligned with a foundation model, a face embedding can be 'unmasked' in multiple ways, without training or modifying either model: it can be read in natural language, enabling free-form text queries over a gallery of FR embeddings; rendered into a face image that recovers a person's appearance, using an unmodified diffusion decoder; and converted to a name, enabling identification even in the absence of an enrolled face gallery. In effect, one linear transformation turns an identity embedding into a rich embedding for web-scale foundation models. This interoperability exposes face embeddings as semantically and visually rich biometric representations, with direct implications for interpretability, retrieval, reconstruction, and template security.
Read the full abstract

Nuclear Theory

What Are We Talking About When We Talk About Nuclear Reactions

This introductory Chapter will try to briefly address 1. the basic physical ingredients that determine the nuclear spectrum and the associated energy scales; 2. the connection between the observed experimental results, typically in the form of cross sections, and the underlying nuclear structure presented in the previous point. The nature and scope of the present Chapter is to provide a roadmap of the field of nuclear reactions theory, and to set the stage for the more detailed discussions in the companion Chapters. The aim is to provide a conceptual framework that will allow the reader to understand how the different reaction theory approaches fit together, and how they relate to the underlying physics of nuclear reactions. The reader will find in other chapters in this volume detailed discussions about the methods addressing two seemingly quite distinct phenomena: direct and compound nuclear reactions. We hope that this brief Chapter will help them realize that whenever we are talking about either of these, we are talking about the same thing: nuclear reactions.
Read the full abstract

Operator Algebras

Dominance and equivalence for states on $C^*$-algebras: Quasi-Invariant states

We study the noncommutative generalization of measure-theoretic dominance and equivalence of states on $C^*$-algebras to explore quasi-invariance under group actions. For a dominated state, we derive an unbounded "Radon-Nikodym" derivative affiliated with the commutant algebra of the dominating state's GNS representation. Interestingly, this dominance is generally non-transitive because the product of the corresponding closed operators can be non-closable. When looking at group actions by $*$-automorphisms, the orbit of a fixed quasi-invariant state consists entirely of mutually equivalent states. However, the orbit closure may contain singular states, meaning the set of quasi-invariant states is closed under convex combinations but not topologically closed. The paper also provides a unitary implementation of the group action on the GNS Hilbert space-generalizing covariant representations and compares this approach with the Pedersen-Takesaki construction, where the Radon-Nikodym derivative sits in the centraliser instead of the commutant.
Read the full abstract

All authors Ameur Dhahri, Francesco Fidaleo, Chul Ki Ko, Hyun Jae Yoo

Statistical Computation

Non-Uniform Random Scans in Gibbs Sampling and CAVI

Gibbs sampling and coordinate ascent variational inference (CAVI) are two basic coordinate-wise methods for statistical computation. Recent analyses under strong log-concavity establish convergence rates for versions of these algorithms that update one uniformly selected block at each step. We extend both results to arbitrary fixed, strictly positive selection probabilities. The rates are governed by a selection-adapted convexity constant $\lambda^{\star}_{\theta}$, defined using the block-smoothness constants and the selection probabilities $\theta$. The same constant yields a contraction of relative entropy for the Gibbs sampler and a contraction of the mean-field objective gap for random-scan CAVI. The new bounds recover the uniform-scan results, and are never weaker than the naive comparison based on the smallest selection probability. They provide a principled way to adapt the scan to heterogeneous block geometry using curvature information.
Read the full abstract

Hardware Architecture

DynaNDE: Dynamic Near-Data Expert Scheduling for Batched MoE Inference

Mixture-of-Experts (MoE) models enable efficient scaling of large language model (LLM) inference but suffer from substantial data-movement overhead when deployed on neural processing unit (NPU)-based systems. Near-Data Processing (NDP) provides a promising way to mitigate this bottleneck via cooperative NPU-NDP execution. However, existing NPU-NDP MoE systems do not fully account for hardware heterogeneity, dynamic expert-level concurrency, and temporal expert reuse during batched inference. This paper presents DynaNDE, a dynamic near-data expert scheduling framework that exploits NPU-NDP collaboration to accelerate batched MoE inference. DynaNDE introduces an analytical performance model that captures hardware heterogeneity, data-movement costs, and communication-computation overlap in cooperative NPU-NDP execution. Guided by this model, DynaNDE determines per-layer expert scheduling across the NPU and NDP while accounting for expert-level concurrency. DynaNDE also incorporates a reuse-aware runtime that avoids redundant parameter movement when experts reside in NPU memory. Experimental results show that DynaNDE achieves substantial throughput improvements over the state-of-the-art NPU-NDP MoE serving framework, with average speedups of 2.6$\times$ and 2.2$\times$ for the prefill and decoding stages, respectively.
Read the full abstract

Dynamical Systems

Multiple ergodic averages for commuting multiplicative actions

We study the convergence of multiple ergodic averages involving several commuting multiplicative actions. We prove that for finitely generated systems, the averages $$\frac{1}{N}\sum_{n=1}^N S_{1,n}F_1\cdot\ldots\cdot S_{\ell,n}F_\ell$$ converge in norm, settling a conjecture of Frantzikinakis. Our proofs feature tools from ergodic theory and probabilistic number theory. We exploit some stability properties of additive functions to establish a novel variant of \Katai's orthogonality criterion, which leads to box seminorm control for our averages. Using the theory of magic extensions introduced by Host and a delicate complexity reduction argument, we eventually reduce our problem to known convergence results.
Read the full abstract

Machine Learning

A Multi-Branch Feature Fusion Approach for Health Misinformation Detection and Propagation

This paper presents a multi-branch fusion framework for detecting and characterising the propagation of health misinformation in online social networks (OSNs). Grounded in the Elaboration Likelihood Model (ELM) and the Theory of Planned Behaviour (TPB), the model fuses transformer-based semantics with rhetorical cues, stance representations, and psychologically motivated proxies in a unified multi-task architecture. In addition to binary classification, we introduce the Cognitive Propagation Score (CPS), an interpretable post-hoc auxiliary score computed from psychologically motivated, text-derived cues capturing argument complexity, emotional intensity, and content-derived virality potential, to support diffusion-risk reasoning when engagement ground truth is incomplete or unavailable. Experiments on three benchmark datasets, Constraint, COVID–19\_FNIR, and Monkeypox, show strong classification performance, achieving ROC–AUC up to 0.9999 on COVID–19\_FNIR, while propagation-oriented ranking achieves near-perfect agreement when engagement-derived supervision is available (Monkeypox, Spearman's $\rho = 0.9952$) and similarly high ranking alignment under proxy-based supervision on COVID–19\_FNIR ($\rho = 0.9954$). Compared with representative literature baselines, the fusion model improves detection on Constraint and COVID–19\_FNIR, while Monkeypox remains more challenging, reflecting domain- and signal-specific differences. Ablation analysis further indicates that psychological and rhetorical branches provide complementary gains beyond semantic embeddings. Overall, the framework bridges cognitive theory and neural modelling to improve transparency and to support scalable misinformation monitoring, with future work required to validate CPS against human-centred diffusion judgements.
Read the full abstract