Peer Revenue Returns subject to peer review

Papers

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

All of arXiv

Tuesday 1 September

Algebraic Geometry

On the solutions of relative regular holonomic D-modules

Let $S$ be a complex curve and let $X$ be a complex manifold. Given a relative regular holonomic $\DXS$-module $\shm$ with respect to the projection $X\times S\to S$, we construct an isomorphism between the solution functors respectively on the (relative subanalytic) sheaf of relative tempered holomorphic functions and the sheaf holomorphic solutions on $X\times S$.
Read the full abstract

Artificial Intelligence

SSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval

This article presents SSAKG 2.0, an open-source software package for constructing and operating Structural Sequential Associative Knowledge Graphs (SSAKGs). An SSAKG represents objects as graph vertices and ordered sequences as structural patterns of graph connections. The resulting sparse graph is used as an associative memory in which complete sequences can be reconstructed from a partial, unordered context. Version 2.0 introduces new algorithms that exploit individual bits of computer memory to efficiently search graph connections. The package is implemented in Python, while performance-critical graph operations are implemented in C and exposed through a Python interface. This hybrid implementation provides a flexible high-level programming environment while reducing the memory and computational overhead associated with large sparse graphs. The algorithms were evaluated using randomly generated numerical sequences, sequences derived from sentences in the NLTK corpus, and mRNA sequences. The experiments demonstrate the ability of the package to store and reconstruct sequences from partial contexts and provide a basis for evaluating the effects of graph density, sequence length, and memory size on retrieval performance. SSAKG 2.0 is distributed under the Apache 2.0 open-source license. The package includes documentation and reproducible examples and is publicly available through GitHub and the Python Package Index (PyPI).
Read the full abstract

Quantum Physics

Error accumulation in dissipative quantum circuits

Present-day quantum processors are open systems in which dissipation and decoherence degrade the information carried by a quantum state as it propagates through a circuit. We study this degradation in dissipative random quantum circuits, modeling each two-qubit gate as a diluted unitary that interpolates between the intended unitary operation and a random dissipative quantum channel. Using the fidelity between the ideal and noisy output states as a diagnostic, we show that sufficiently random unitary gates or dissipative Kraus operators lead to a universal decay of the average fidelity. This decay is determined only by the dissipation strength, system size, and circuit depth, while microscopic details of the gates and noise, including the Kraus rank, appear only in subleading corrections to higher moments. We derive a closed-form expression for the average fidelity in terms of three physically meaningful error parameters, dissipation strength, coherent two-qubit gate-error strength, and connectivity-error probability. Finally, we identify the cases when dissipative effects can be reproduced by an unitary-noise model and when dissipation-induced decoherence remains distinguishable from coherent noise at the level of the average fidelity.
Read the full abstract

All authors Nadir Samos-Sáenz de Buruaga, Rodrigo M. C. Pereira, Karol Życzkowski, Sergey Denisov, Pedro Ribeiro

Combinatorics

Polynomial configurations and pointwise averages along Piatetski-Shapiro sequences

In this paper, we prove that for every integer $k\geq2$ and every $c>1$ sufficiently close to $1$, there is $\kappa>0$ such that every sufficiently large subset of $\{1,\ldots,N\}$ of density at least $(\log\log N)^{-\kappa}$ contains \[ x,\quad x+\lfloor n^c\rfloor,\quad x+\lfloor n^c\rfloor^2, \quad\ldots,\quad x+\lfloor n^c\rfloor^k. \] We also prove pointwise almost-everywhere convergence of the associated multiple ergodic averages.
Read the full abstract

Computation and Language

Cite or Decline: A Strict Course-Grounded Chatbot for STEM Lecture Videos

Recorded lecture videos, often enhanced with search and summarization features, are a standard study resource. However, students cannot easily ask course specific questions or verify answers against an instructor's lecture. We report a semester-long deployment of VideoPoints platform with a retrieval-augmented chatbot that answers from course lecture materials and returns timestamped citations. The chatbot retrieves only from the active course, uses chapter summaries to guide transcript ranking, and returns clickable timestamped citations. Students used it for quick lookups and exam review. Across 833 messages, 70.5% included citations, none crossed a course boundary, and when no lecture evidence matched, the chatbot usually declined rather than answering. Among the users, citations were the most consistently useful feature, while practice-question generation was the strongest unmet request. We also evaluated the design on the real-world test split of EduVidQA, a public multimodal benchmark for lecture-video question answering. Our design improved correct-lecture retrieval by 6.3 percentage points over dense-only retrieval. Together, the results show that effective deployment depends on course isolation, supported citations, and alignment with students' study practices.
Read the full abstract

Computational Physics

Learning a general class of admissible multi-species collision operators from molecular dynamics

We develop a structure-preserving, data-driven collision operator for spatially homogeneous multi-species kinetic systems from molecular dynamics (MD). The operator consists of diagonal self-collision blocks and ordered off-diagonal cross-species blocks to describe intra- and inter-species momentum and energy exchange. Within a local and point-wise identifiable kernel class, we develop the necessary and sufficient condition for the admissible kernel class satisfying the conservation laws, the H-theorem, and the frame indifference. Unlike the classical Landau operator, the off-diagonal kernels are not restricted to be symmetric under permutation of the two velocity variables. This unique structural freedom captures the distinct responses of different species to unresolved correlations and many-body effects arising from micro-scale particle interactions. The equivalent parameterizable kernel formalization enables us to learn a generalized data-driven collision operator directly from MD, where the low-rank tensor representations and random sampling are used to achieve efficient kernel training and numerical simulation. Numerical experiments show that the learned operator accurately predicts transport coefficients and the non-equilibrium relaxation, while retaining discrete conservation and entropy production. In particular, it captures plasma kinetics in the moderately coupled regime, where the predictions of both the Landau and the data-driven model restricted to velocity-permutation symmetry show significant discrepancies.
Read the full abstract

Materials Science

Symmetry-selective nonrelativistic spin splitting in antiferromagnets driven by coherent phonons

Nonrelativistic spin splitting (NRSS) in antiferromagnets (AFMs) enables magnetization-free spin polarization for ultrafast spintronics. Here, we demonstrate that coherent phonons can dynamically induce and control NRSS in collinear AFMs. Excitation of $\Gamma$-point infrared-active phonons lifts the spin degeneracy of the ground state, while the residual sublattice-connecting symmetries determine the momentum-space form of the induced splitting. Because the relevant infrared modes couple to orthogonal in-plane light polarizations, distinct spin-split phases can be selectively activated by the polarization of the driving field. Using first-principles calculations for MnPS$_3$, we show that a mode that breaks all sublattice-connecting symmetries induces an $s$-wave spin-split state with $\Delta(\Gamma)\neq0$, whereas a symmetry-distinct mode that preserves a sublattice-connecting mirror symmetry generates a $d$-wave altermagnetic state with $\Delta(\Gamma)=0$. In both cases, the spin splitting grows linearly with the phonon amplitude and reverses the induced spin polarization when the displacement is reversed. Our results establish coherent lattice driving as a direct, polarization- and mode-selective route to dynamically induce distinct NRSS phases in AFMs.
Read the full abstract

All authors Sangeeta Rajpurohit, Mohsen Yarmohammadi, Sheikh Rubaiat Ul Haque, Tony F. Heinz, Aaron M. Lindenberg, Tadashi Ogitsu

K-Theory and Homology

Algebraic K-theory of quotient ring spectra

We study the algebraic K-theory of quotients of even ring spectra. Building on Efimov's work, we prove that the limit of the K-theories of these quotients is the continuous K-theory of nuclear modules. For quotients by elements in the height $n$ chromatic ideal, we show that this agrees with the K-theory of the ring spectrum, after chromatic localization at heights $\geq n{+}1$.
Read the full abstract

Statistical Methodology

Conditional Distribution Estimation Given Functional Covariates Using Deep Operator Networks

Functional linear models are commonly used for analyzing functional data with scalar responses, particularly for modeling the conditional mean of the response given functional covariates. In this work, we extend the traditional scalar-on-function regression in two major aspects: 1. estimating the conditional distribution function instead of a particular characteristic such as the conditional mean; 2. considering an arbitrary operator without imposing functional linearity or any model assumption. We use a likelihood approach for the conditional hazard function and apply convolutional neural networks to approximate the effects of functional inputs and estimate the conditional distribution using deep operator networks. Through simulations and a real world data example, we show the desirable robustness of the proposed method in comparison with the mean regression neural networks, demonstrating that our approach achieves better conditional distribution estimation and interval prediction in complex data settings.
Read the full abstract

Instrumentation and Detectors

Characterization of a low-power 3D photon-to-digital converter readout improved for system integration in meter-scale applications

Digital silicon photomultipliers (dSiPM) are arrays of single photon avalanche diodes (SPADs) where each SPAD has its own electronic readout. To maximize the photodetection area, we developed a readout integrated circuit (ROIC) 3D-integrated to a custom-designed SPAD layer fabricated at Teledyne Dalsa (Bromont, Canada) to form a photon-to-digital converter (PDC). This paper presents an improved version of a previously demonstrated ROIC, fabricated in TSMC 180 nm technology to manage a 64 x 64 pixel architecture. This ROIC has different outputs, such as a flag output that produces a pulse whenever one of the pixels triggers and a digital sum that samples the amount of triggered SPADs. Improvements lead to a reduction of the timing jitter on the flag output from 72.0 ps RMS to 22.6 ps RMS through optimized H-tree design. A tunable hold-off circuit provides adjustable dead time from 32 ns to 18 {\mu}s, addressing both afterpulsing mitigation and SPAD-to-SPAD variations due to process defects. Power consumption is lowered by 30% at trigger rates exceeding 10 kHz through digital logic optimization and clock gating. These measurements validated that the ROIC is ready for 3D integration into a PDC for systems in medical imaging, particle physics, and quantum sciences.
Read the full abstract

All authors Olivier Lepage, Tommy Rossignol, Nicolas Roy, Gabriel Lessard, Frédéric Vachon, Lorenzo Fabris, Serge A. Charlebois, Jean-François Pratte

Computer Vision

Cross-Model Distillation of a Human-Pose Foundation Model from Unannotated Infant Video for Markerless 3D Pose Estimation

Spontaneous movement is one of the earliest windows onto an infant's neuromotor health, and structured clinical instruments that score it are validated early predictors of cerebral-palsy risk. However, they require specially trained raters, are time-consuming, and carry inter-rater variability. This motivates automated, video-based markerless assessment, especially as marker-based motion capture is impractical in infants. Yet the foundation models that make markerless capture possible are trained almost entirely on adults: our recent multi-view infant study found that no single model is jointly best, with strong 2D keypoint accuracy and direct 3D body recovery split across different models. While that study identifies this trade-off, it does not resolve it. Here, we perform cross-model distillation from the Sapiens 2 pose model into the SAM 3D Body model, using unannotated infant video alone. A frozen teacher supplies dense pseudo-labels, and a differentiable renderer aligns the predicted mesh to them in the training loop. On eleven held-out infants (18 sessions, 173 recordings) under our prior study's multi-view protocol, fine-tuning improves same-view 2D keypoint agreement with the Sapiens reference (median body percentage of correct keypoints @ 10px 0.22 -> 0.42, face 0.22 -> 0.42) and Procrustes-aligned mean per joint 3D position error (25.5 -> 22.2 mm). This demonstrates how cross-model distillation improves SAM 3D Body model performance on infants.
Read the full abstract

Machine Learning

Import What You Need: Learning When and How to Augment EHR Graphs with External Knowledge

Longitudinal prediction from electronic health records (EHRs) is limited by the sparsity and irregularity in patient trajectories, and knowledge augmentation with external knowledge graphs (KGs) offers a promising way to alleviate these issues. However, KG expansion often follows predefined rules, with limited control over whether and how to augment each visit as a patient's state evolves. We propose ReTA, a Reinforcement learning-based dynamic Topology Augmentation framework that casts KG import as a per-visit, budget-aware policy. ReTA first constructs an offline refined pool of KG-grounded templates, then learns a policy to select one augment action per visit from three options: Soft Import, which enriches node features without modifying graph topology, Hard Import, which grafts a compact KG subgraph onto the visit graph to create message-passing shortcuts, and Skip, which leaves the visit unaugmented when the base encoder is already confident. To stabilize learning, ReTA employs a decoupled encoder that processes semantic and structural signals in separate channels and fuses them via adaptive gating. Experiments on MIMIC-III and MIMIC-IV across diagnosis prediction, mortality, and readmission show that ReTA consistently outperforms strong baselines while remaining efficient, with larger gains under sparse supervision. ReTA transfers across datasets and knowledge graphs and yields interpretable augmentation patterns.
Read the full abstract

All authors Chen Chen, Mohsen Nayebi Kerdabadi, Dongjie Wang, Mei Liu, Zijun Yao

Multiagent Systems

Differential Games for Compositional Handling of Competing Control Tasks

We introduce a novel Divide and Conquer control design methodology leveraging differential games in single-agent, multi-objective dynamical systems. The proposed framework associates each control objective with a virtual input and establishes a non-cooperative, finite or infinite horizon differential game among representative players. Each player optimizes a distinct virtual cost function tailored to its specific goal, the full system state, and the other virtual inputs, while accounting for the remaining players' optimal policies. By establishing a Nash Equilibrium for this game, we synthesize a composite controller that achieves a stable balance across competing objectives, providing control engineers with an intuitive and modular framework for parameter re-tuning throughout the design cycle. We provide formal mathematical derivations for both continuous-time and discrete-time dynamical systems, targeting large-scale single-agent applications where complex, dynamically conflicting control objectives make global weighting intractable. To demonstrate the methodology, we developed an open-source Python package implementing a novel numerical algorithm for solving Coupled Algebraic Riccati Equations arising in infinite-horizon differential games. We evaluate the approach on two benchmark case studies: an inverted pendulum on a cart and a non-linear hierarchically controlled quadrotor. The resulting closed-loop performance is compared against the classical Linear Quadratic Regulator (LQR) across various transient and steady-state control metrics, demonstrating superior trajectory tracking and robust multi-objective regulation.
Read the full abstract

Group Theory

Acylindrical Hyperbolicity and Pure Conjugating Automorphisms in Graph Products of Groups

We study graph products of groups over defining graphs of arbitrary cardinality from two closely related viewpoints. First, we characterize acylindrical hyperbolicity. If the defining graph is irreducible, has at least two vertices, and has a finite star base, then every parabolically full subgroup is either virtually cyclic or acylindrically hyperbolic. For vertex-full subgroups the finite star base condition is also necessary. In particular, the graph product itself is acylindrically hyperbolic exactly when the graph has a finite star base and the group is not virtually cyclic, or equivalently is not the infinite dihedral group. We also give the corresponding classification for reducible defining graphs. Second, motivated by our earlier work on countable right-angled Coxeter groups, we study pure conjugating automorphisms in the topology of pointwise convergence. We prove that they are topologically generated by finite-support factorwise automorphisms and partial conjugations, and establish closedness, density, exact-equality, and discreteness results. The same finite star base condition that appears in the acylindrical hyperbolicity criterion governs closedness and discreteness in the star-connected case; for arbitrary graphs, its natural refinement is the existence of a finite component witness set. Finally, for graph products of abelian groups we obtain a canonical topological semidirect decomposition of the pure conjugating automorphism group.
Read the full abstract

Cryptography and Security

Agent Memory Is a Surface for Endogenous Authorization Laundering

Long-running LLM agents rely on persistent memory to preserve not only task context but also the permissions and restrictions governing their actions. When memory misrepresents this evolving authorization state, the agent's own records can make unauthorized actions appear permitted, resulting in misaligned behavior without any external attacks. We term this failure endogenous authorization laundering, where memory creates or retains false authority through semantic misinterpretation or failed updates, and executors act on that authority. We then introduce EAL-Bench to measure how accurately persistent memory preserves evolving authorization state and whether memory errors propagate to unauthorized actions. We evaluate seven LLMs as memory writers and two as executors across procurement, cybersecurity, and finance tasks. Under incremental typed-memory updates, writers create false authority for up to 45.4% of unauthorized requests, and once false authority is present, executors act on it in 99.0% of trials. Memory-level safeguards reduce laundering but can reject legitimate actions, exposing a safety-utility tradeoff. Persistent memory is therefore part of an LLM agent's effective authorization policy, not merely a performance component.
Read the full abstract

Functional Analysis

On Exposed (Symmetric) Tensors

We provide an example of a projective tensor product whose unit ball contains an exposed tensor which is not elementary. We then transfer the construction to the symmetric projective tensor product and obtain an analogous result in that setting. Finally, we extend both constructions to projective tensor products with an arbitrary number of factors and to symmetric projective tensor products of every degree.
Read the full abstract

Artificial Intelligence

Architecting Conversational Data Systems for Stateless LLM APIs: The Hydration Proxy Pattern

As enterprise platforms transition to conversational reasoning interfaces, the stateless nature of LLM APIs creates an architectural gap. While statelessness enables horizontal scalability for AI providers, it forces client applications to manage the entire burden of conversational state and semantic memory. The work identifies the Hydration Proxy Pattern, an architecture that decouples session persistence from the reasoning engine. The framework ensures platform sovereignty over conversational data while enabling secure, multi-stage semantic grounding. We further propose the Context Stabilization Mandate to resolve the tradeoff between sovereign state management and KV caching.
Read the full abstract

Computation and Language

Candidate Generation and Definition-Guided Verification for Sentence-Level Depression Symptom Recognition

Sentence-level recognition of depression symptoms is challenging because similar expressions can differ in symptom relevance, and language-model inference is insufficiently grounded in diagnostic definitions. This study proposes a two-stage framework separating symptom-candidate generation from definition-grounded verification. A contrastively fine-tuned sentence encoder generates a symptom candidate per sentence, and a fine-tuned language model verifies whether the candidate is present or absent using the sentence, its context, and a candidate-specific diagnostic definition, checking its judgment against that definition before answering. Evaluated against encoder, inference-based, medical, and general LLM baselines and a matched single-stage supervised classifier, the proposed pipeline attains the best accuracy and F1 scores of all methods, with rationales matching expert-authored annotations. A preliminary clinical audit indicates moderate alignment with diagnostic definitions, with explanation quality strongly dependent on prediction correctness. The results support decomposing symptom recognition into candidate generation and definition-grounded verification, though performance remains limited for rare categories.
Read the full abstract

All authors Weiming Li, Catarina Barata, Miguel Constante, Joao Sanches

Computation and Language

Interpretable Symptom Vectors for Depression in a Large Language Model

Patients with depression present with diverse symptom profiles, yet clinical practice routinely reduces this variation to a single severity score. Large language models (LLMs) can potentially capture various symptoms and their severity from patient speech. However, how depressive symptoms are represented inside LLMs remains poorly understood, limiting clinical trust. To examine whether internal model activations match clinician judgment, we analyzed the residual stream of Gemma-3-27B-PT using mechanistic interpretability techniques. Recording activations across symptom descriptions drawn from validated clinical instruments, we found that symptom groups geometrically separated the most at layer 21 across multiple distance metrics. Using Semantic Projection, we then projected held-out naturalistic text onto Symptom Vectors constructed from these instruments. The resulting per-symptom coefficients preserved clinician-annotated rank ordering across mood, somatic, and suicidality axes. Furthermore, a single depression vector in Layer 21 separates held-out depressive from non-depressive text (AUC = 0.789), which can be used as an emotional valence gate that restricts symptom projection to depressive speech. These results reveal a decorrelated, clinician-aligned symptom signal readable directly from internal activations, offering a mechanistic foundation for interpretable depression-assessment tools.
Read the full abstract

All authors Fangyi Zhu, Ajay Subramanian, Allison Constant, Camille Wang, Ravish Gupta, Corey J. Keller

Quantum Gases

Kelvin-Helmholtz instability in annular superfluids across the UFG-BCS crossover

We investigate the Kelvin-Helmholtz instability of counter-rotating annular Fermi superfluids using time-dependent superfluid density functional theory. When the two initially separated superflows merge, a necklace of quantized vortices forms at their interface and subsequently becomes unstable. We extract mode-resolved instability growth rates and study their dependence on relative flow velocity, interaction strength, and temperature across the strongly interacting unitary regime and the weakly attractive BCS regime. In the unitary gas and moderate BCS regimes, the growth rates remain qualitatively consistent with the point-vortex model, though systematically at or above its predictions, with no significant dependence on interaction strength or temperature up to approximately $0.3 T_c$. Deeper in the BCS regime, the dynamics change qualitatively: boundary-generated antivortices proliferate faster than the Kelvin-Helmholtz instability itself, destroying the initial vortex necklace before the instability can fully develop. These results provide a microscopic benchmark for Kelvin-Helmholtz dynamics in fermionic superfluids and help constrain possible explanations of the existing experiment-theory discrepancy.
Read the full abstract

Geometric Topology

The bounded area class of negatively curved surfaces

Let $S$ be an oriented surface, possibly of infinite type, endowed with a complete Riemannian metric with pinched negative curvature. We prove that the area form defines a non-trivial class in the second bounded cohomology group of $S$, unless $S$ is diffeomorphic to the disc or the cylinder. This is in sharp contrast with the $n$-dimensional case, $n>2$, where the non-triviality of the volume form in bounded cohomology is related to the Cheeger constant of the manifold. We also discuss how the bounded area class depends on the metric: we prove that, for compact surfaces, it recognizes constant curvature metrics among pinched negatively curved ones, while, even in the case of surfaces of infinite type, it does not distinguish non-isometric hyperbolic structures. More precisely, when $S$ is compact we show that, among the negatively curved structures of fixed area, the ones with constant curvature provide the unique minimizers for the norm of the bounded area class.
Read the full abstract

Mathematical Physics

Extended Body Propagators

Three seemingly distinct extended-body propagators devised by Ansoldi et al., Erbin et al., and Stanford et al.\ respectively arise in completely different settings, though each can be considered a realization of the Polyakov path integral. This paper proves that the Ansoldi propagator corresponds to the characteristic function of the random variable with the probability density function $f(A)=C\,A\,(\alpha^{2}+A^{2})^{-D/2}\mathbf{1}_{A>0}$ in ambient dimension $D$, of scale $\alpha=\sqrt{b/2}$. The exponent is fixed by the dimension and by nothing else in the Ansoldi/geometric comparison; the on-shell two-point amplitude of Erbin et al. and JT gravity agree with its $D=4$ member in the senses made precise below. After an extensive literature review, the author believes that this observation and constructed area measure is novel. The area measure will, however, emerge as the natural candidate for the extension of the earlier non-extended body path space measures to two dimensional simplicial complexes. We further show that the area measure of this paper is exactly the law of the area of a triangle, two of whose sides are drawn from the measure obtained in the companion manuscripts for the one-dimensional Polyakov integral. We close with an argument, rigorous in part, which reduces the measure of tubular worldsheets between two arbitrary boundary curves to that of this triangle.
Read the full abstract

Computation and Language

AVERT: Audio-Verified Adjudication for Spoken Dialogue State Tracking

Spoken dialogue state tracking recovers slot-value pairs from speech, where ASR errors concentrate in entity values and persist across turns, making it both a generation and an editing problem. A strong per-turn text editor corrects much of this but, operating on the transcript alone, leaves three recoverable errors: a value predicted inconsistently across turns, an omitted slot, and a value the audio does not support. We present AVERT, which scores each candidate value by combining cross-turn agreement with a trained audio-conditioned verifier and resolves the three error types with three operators, vote, add, and swap, each restricted to the slots where its error is common. On SpokenWOZ, a base speech-LLM reaches 33.04 JGA, a text editor 38.34, and AVERT 40.13, without retraining either. This is in the range of a 1B end-to-end system that consumes the full spoken history (39.32), though AVERT uses two 1B decoders rather than one. The audio verifier contributes a statistically significant gain, and restricting each operator to a selected slot subset matters: removing it lets unrestricted voting overwrite correct categorical values and fall below the editor.
Read the full abstract

Computer Vision

SliceBridge: Context-Consistent Repair of Corrupted Slice Intervals in T1-Weighted MRI

Structural magnetic resonance imaging (MRI) images are sometimes corrupted over a contiguous set of slices, where acquisition, motion, hardware, or reconstruction effects leave a single slice or short interval inconsistent with its neighbors while the rest of the image remains usable. Such localized corruption can bias downstream morphometric analysis, yet discarding or reacquiring an otherwise usable image is costly. We formulate this as an image restoration problem: given the location of the affected interval, reconstruct those slices from the surrounding anatomical and imaging context. We propose SliceBridge, a framework for restoring corrupted slice intervals in T1-weighted MRI using rectified flow matching conditioned on the surrounding intact slices and their relative slice positions. Through-plane consistency is encouraged by coupling the slices within the interval through interval-correlated initial noise, a shared flow time, and synchronized sampling. The restored interval is then inserted back, leaving all other slices unchanged. We trained and validated the model on 9,877 T1-weighted brain MRI volumes from four datasets and evaluated it on 581 external subjects using clean interval withholding and controlled corruptions. Compared with a matched model that reconstructed target slices independently, SliceBridge reduced error in slice-to-slice changes within repaired intervals by 32.9%-41.3% across interval lengths and achieved higher SSIM at every interval length. In controlled-corruption cases, SliceBridge reduced the median error in regional brain volume estimates produced by a downstream segmentation model from 1.95% in corrupted volumes to 1.05%.
Read the full abstract

All authors Jiheng Li, Michael E. Kim, Trent M. Schwartz, Gaurav Rudravaram, Derek B. Archer, Timothy J. Hohman, for the Alzheimer's Disease Neuroimaging Initiative, Lianrui Zuo, Bennett A. Landman

Analysis of PDEs

Fully fractional Schrödinger equations with memory: well-posedness via extension

We study the nonlinear Schrödinger equation with memory $\Le^s u=\mu|u|^{p-1}u$ for $t>0$, where $\Le^s=(\partial_t-i\Delta_x)^s$, $0<s<1$, and the solution is prescribed for all $t\le0$. We identify a Hilbert space of histories and prove local well-posedness for $1<p\le p_*(s)=1+4s/(d+2-2s)$, with global solutions for sufficiently small histories at the critical exponent. We also establish the passage from this memory problem to the classical nonlinear Schrödinger equation as $s\uparrow1$. For every fixed Schwartz history and $1<p\le1+4/d$, including the classical mass critical exponent, the memory solutions converge in $C([0,T];L^2\cap L^{p+1})$ to the classical solution with initial value $u_0(\cdot,0)$, for every finite $T$ below its maximal existence time. Thus the limiting evolution depends on the prescribed past only through its value at time zero. Our construction uses an oscillatory Poisson extension and an exact identity expressing the weighted $L^2$ norm of the lifted initial datum in terms of the history. This identity allows us to apply the boundary estimates of our earlier work to the memory problem. We recover $\Le^s$ as a weighted Dirichlet-to-Neumann map and establish the corresponding extension bounds on $L^2$ for $0<s\le\frac12$, and from $\operatorname{Dom}(\Le^s)$ to $L^2$ for $\frac12<s<1$.
Read the full abstract

Number Theory

On everywhere locally soluble varieties in families over toric bases

We determine the density of locally soluble fibres in certain families of varieties over a toric base. For this we develop a new form of the hyperbola method for toric bases which allows for counting functions with a growth behaviour of the form $B^a(\log B)^{-\Delta}$.
Read the full abstract

All authors Vladi Mitankin, Leo Schäfer, Damaris Schindler, Florian Wilsch

Optics

Optical air waveguides in strong turbulence

Filament-generated air waveguides offer a route to low loss long-distance optical transport, but their viability in atmospheric turbulence has remained uncertain. Here, we demonstrate with experiments and simulations that guided beams in air waveguides can overcome prior estimated turbulence limits by orders of magnitude in the refractive index structure parameter C_n^2; turbulent index fluctuations much larger than the waveguide index contrast average out along the propagation path. Instead, turbulence imposes two distinct limits: scattering of the guided field beyond the waveguide acceptance aperture and, more importantly at atmospheric strengths, scintillation of the waveguide-forming beam. Scaling of experimentally validated models indicates that filament-generated air waveguides remain viable over kilometer-scale paths under strong near-ground turbulence. Our results establish turbulence limits for long-range optical guiding in the atmosphere.
Read the full abstract

Computer Vision

Kirin: Animal Motion Generation from In-the-Wild Video

Understanding animal motion is fundamental to modeling animal behavior and biomechanics, yet progress in this area lags far behind human motion research due to the scarcity of high-quality motion data. While human motion can be captured in controlled environments, it is impractical for most animal species, resulting in small, domain-limited datasets that restrict downstream applications such as animation. To address this challenge, we introduce Kirin, a framework that reconstructs motion from video, learns motion priors at scale, and generates realistic motion that can be directly applied to animated assets. Using large collections of in-the-wild animal videos, we reconstruct 3D motion sequences and pair them with captions to create AiM3D, the first large-scale dataset offering aligned video-text-motion tuples for quadruped animals. Building on this dataset, we develop a visual-guided motion generation model that conditions on both text and image to guide the generation of realistic motion across diverse animal species. Finally, by leveraging an off-the-shelf image-to-3D model, we automatically rig and animate 3D meshes using generated motion, producing ready-to-render animated animals. Together, our dataset and framework establish a new foundation for large-scale, text and image conditioned animal motion generation and animation. Project page: https://kirin-ani.github.io/.
Read the full abstract

All authors Brian Nlong Zhao, Zhuoyang Pan, James M. Rehg, Jiajun Wu, Shangzhe Wu

Solar and Stellar Astrophysics

Mass-dependent multiplicity fraction in low mass stars revealed by Gaia astrometry

Stellar multiplicity at different orbital periods is probed by different techniques: radial velocities at the shortest periods, direct imaging at the longest, and astrometry in between. Gaia DR3 provides unprecedented astrometric information to constrain binary populations with periods of tens to thousands of days. Beyond the small fraction of direct orbit solutions, Gaia publishes coarser astrometric diagnostics for all objects, such as on-sky accelerations, jerks, or the single-star goodness-of-fit measure, RUWE. We show that these diagnostics together are sensitive to a far wider range of orbital periods than orbit solutions alone. Built on the Gaiamock emulator, we develop a forward-modelling framework to constrain rates of multiplicity from Gaia data, using all astrometric solution types. We apply this framework to $366~027$ primaries of $0.2$–$0.75\,M_\odot$ between $50$ and $200$ pc, including vastly more M-dwarfs than previous multiplicity studies. For fiducial models of the period and mass-ratio distributions, the inferred multiplicity fraction is approximately constant at ${\sim}45\%$ from the solar-type regime down to $0.4\,M_\odot$, only then dropping to ${\sim}15\%$ by $0.2\,M_\odot$. This mass-multiplicity relationship revises the conventional picture of a monotonically rising mass–multiplicity relation, and aligns the upper M-dwarf regime with solar-type stars rather than placing it on a continuous decline.
Read the full abstract

All authors Ben Pennell, Hans-Walter Rix, Kareem El-Badry, Sahar Shahaf, Johanna Müller-Horn, Jiadong Li

Distributed Computing

Scaling Inference Prefill with High-Radix Photonic Interconnects

With the rise of inference as today's dominant AI workload, the industry is transitioning to high-bandwidth photonic interconnects to meet the large scale-up requirements of increasingly complex Mixture-of-Experts (MoE) models. This paper quantifies the benefits of 3D-integrated photonic interconnects for inference prefill by analyzing tradeoffs between high-concurrency throughput for Large Language Model (LLM) chat and the large context windows typically required for reasoning and agentic AI. We simulate three MoE models: short context (1K–8K tokens), medium context (128K tokens), and long context (1M tokens). We evaluate this workload across existing copper-based GPU systems and one with high bandwidth integrated photonics. We show 2.1–3.2x latency improvements in the stressed high-batch regimes and 2.8–5.8x improvements over baselines in communication-limited configurations. 3D photonics enable the 1152-GPU footprint required to lower time-to-first-token, yielding 2.2–4.5x speedups across production-grade platforms when electrical systems cross their inherent scale-up-pod limits.
Read the full abstract

All authors Arulselvan Madhavan, Peter Carson, Taylor Groves, Thomas Graham