Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Programming Languages

Design Docs Are All You Need: An AI-native Machine-Learning Performance Tool

Machine-learning performance modeling is a uniquely hostile terrain for long-lived software: the assumptions baked into today's abstractions are invalidated by tomorrow's models and systems, forcing perpetual refactoring of performance-modeling frameworks. Meanwhile, AI coding agents have become fast and capable enough that regenerating an entire library is cheaper than paying down the tech debt of incrementally patching it. We describe SMART, a rigorous symbolic performance-modeling library for ML systems whose main branch contains almost no code: the repository is a DAG of self-contained natural-language design docs, coding sub-agents regenerate the implementation from only the docs on new version updates, and every human change is a natural-language edit to a doc–self-documenting by construction. Two ingredients make regeneration reliable: (i) a design-doc style built around step-by-step worked examples that act as in-context demonstrations for the generating agents, and (ii) a minimal, recursively defined operator IR with symbolic (SymPy) cost expressions, a fast analytical roll-up mode for large sweeps, and a slow modulo-scheduling mode for fine-grained schedule studies. Regenerated implementations reproduce hand-audited reference models–including DeepSeek-V3 serving on a TPU pod slice–to round-off precision, suggesting that design docs–not code–can be the durable artifact for ML-systems co-design tools.
Read the full abstract

All authors Samuel Kushnir, Kimia Noorbakhsh, Kavya Sreedhar, Liqun Cheng, Ming Liu, Parthasarathy Ranganathan, Mohammad Alizadeh, Fred Kjolstad, Suvinay Subramanian

Machine Learning

Distill Globally, Adapt Locally: Reasoning Distillation and Product-Type Test-Time Training for Scalable Trade-Up Recommendation

Trade-up recommendation identifies higher-quality alternatives that preserve a customer's purchase intent while offering upgraded benefits. Large language models (LLMs) can reason about such distinctions, but applying them directly to hundreds of millions of product pairs is operationally impractical. We introduce a two-level framework that distills LLM reasoning into an efficient non-generative student and adapts its decision boundary to product-type-specific trade-up criteria. At Level 1, a retrieval-augmented few-shot LLM teacher generates structured relation labels and natural-language rationales. These rationales supervise a compact embedding-pair classifier through alignment and contrastive objectives; at inference, the student uses only two precomputed 768-dimensional product embeddings, with no LLM calls or text generation. On a fixed human-annotated benchmark of 8,352 pairs, a 15.5M-parameter four-class reasoning-distilled student achieves AUC 0.924 (95% CI [0.918, 0.929]), compared with 0.912 for the four-class label-only student. At Level 2, product-type test-time training (PT-TTT) uses few-shot demonstrations to optimize lightweight category-specific adapters over the frozen student. PT-TTT improves AUC from 0.924 to 0.941 and average precision from 0.920 to 0.940. On a 100K-pair proxy catalog, the distilled student on a single eight-GPU machine is approximately 5,000x faster and 10,000x lower in estimated cost than direct LLM inference.
Read the full abstract

All authors Siliang Liu, Mohammad Ghasemi, Sapan Patel, Amin Banitalebi-Dehkordi

Functional Analysis

Twisting exponential spectra

Klaja and Ransford exhibited a complex unital Banach algebra in which the exponential spectra of \(ab\) and \(ba\) differ away from zero, and asked whether this can occur in an algebra of bounded operators. We answer their question affirmatively. For a Bourgain–Delbaen space \(X\) obtained from Motakis' construction with Calkin algebra isomorphic as a Banach algebra to \(C(S^4)\), we construct \(S,T\in\B(X\oplus X)\) such that \[ \frac12\in\varepsilon_{\B(X\oplus X)}(ST) \quad\text{and}\quad \frac12\notin\varepsilon_{\B(X\oplus X)}(TS). \] Moreover, \(\ltsr\B(X\oplus X)=\rtsr\B(X\oplus X)=2\), which is the least possible stable rank for such an example. More generally, if \(K\) is compact metrisable and \(X\) is any space arising from Motakis' construction for \(C(K)\), then, for every \(n\geqslant1\), \[ \ltsr\B(X^n)=\rtsr\B(X^n)= \begin{cases} \left\lceil \lfloor\dim K/2\rfloor/n\right\rceil+1,&\dim K<\infty, \infty,&\dim K=\infty. \end{cases} \] This also gives counterexamples of every finite stable rank at least two and of infinite stable rank. Finally, we relate every failure of exponential-spectral commutativity to the kernel of the first matrix stabilisation map on the index group.
Read the full abstract

Robotics

Development of a Humanoid Robot Prototype for Multimodal Human-Robot Interaction

Human-robot interaction (HRI) enables intuitive and intelligent collaboration between humans and robots in real-world environments. This paper introduces a humanoid robot prototype designed as a flexible testbed for developing and integrating artificial intelligence (AI) modules in HRI tasks. The system features a 12 degree-of-freedom (DOFs) dual-arm mechanism and a 2 DOFs head with an expressive LCD screen to express facial emotions. All hardware components are controlled by a custom-designed controller board with real-time AI processing supported by an onboard Jetson module. The system incorporates three AI modules: (1) gesture recognition using MediaPipe Pose and an LSTM classifier, (2) object detection with YOLO and 3D localization, and (3) voice-command processing through speech recognition and large language model(LLM)-based semantic parsing. The platform is validated through experiments on positioning accuracy, with results showing average manipulation errors of approximately 1.83 cm. To demonstrate its versatility, experimental results show over 90% task accuracy, with gesture recognition reaching 96%, speech recognition reaching 92%. The results confirm the effectiveness of the proposed system as a reproducible and accessible humanoid platform for research and prototyping in HRI.
Read the full abstract

All authors Thang Tran Viet, Thanh Nguyen Canh, Huy Uong Gia, Phuc Dinh Van, Son Tran Duc, Ngoc Minh Do, Xiem HoangVan

Statistics Theory

Finite-sample nonparametric mean tests: Leave-one-out duality and asymptotic optimality

We study finite-sample valid tests of the one-sided mean hypothesis $H_0:\mu\leq 1$ against $H_1:\mu>1$ for nonnegative random variables. To do so, we develop a leave-one-out dual certificate framework, where certain pointwise inequalities imply p-value validity under the conditional mean null $\mathbb{E}[X_i\mid\mathbf{X}_{-i}]\leq 1$, and which also gives conditions that allow combining dual certificates for p-values to show that their pointwise minimum is also a valid p-value. The framework proves finite-sample validity of Wang and Zhao's nonparametric likelihood-ratio statistic $T_{\mathrm{nplr}}$, yields a new p-value $T_{\mathrm{bin}+}$ extending the Clopper–Pearson binomial test to general nonnegative random variables, and shows that the pointwise minimum $\min\{T_{\mathrm{nplr}},T_{\mathrm{bin}+}\}$ is itself a valid and more powerful p-value. We establish sharp optimality results for such testing problems in two regimes: both $T_{\mathrm{nplr}}$ and $T_{\mathrm{bin}+}$ attain a universal detectability boundary for the null $H_0$ without moment or tail assumptions, and $T_{\mathrm{bin}+}$ attains a nonparametric power lower bound under $n^{-1/2}$-local alternatives to $H_0$. Efficient algorithms and numerical experiments demonstrate substantial finite-sample power gains over existing valid methods.
Read the full abstract

Atmospheric and Oceanic Physics

Disentangling Internal and Forced Climate Variability with Convolutional Neural Networks using Multivariate Fields

Long-term climate data exhibit variations composed of internal and forced components. Internal variability arises from natural processes that could occur within a stable climate. Forced variability, on the other hand, reflects climate changes induced, for example, by anthropogenic greenhouse gas and aerosol emissions. Accurately distinguishing between these types of variability is crucial for attributing climate fluctuations and understanding internal variability processes and impacts. In this study, we apply a U-Net convolutional network, a model commonly used in computer vision, to separate forced and internal climate variability from 1950 to 2022 using a multi-model dataset. The dataset includes multiple fields from five single-model initial-condition large ensembles. Cross-validation is conducted by training the U-Net using the data from the ensembles of four models, leaving out the data from one model to assess performance. Validation results yield errors ranging from 0.1{\deg}C to 0.4{\deg}C for the forced variability of local-scale monthly surface air temperature, which is no more than half the magnitude of external forcing. The U-Net achieves better performance than a simple approach based on a fourth-order polynomial trend for estimating forced variability. The error is mainly due to insufficient sampling and poor agreement among the models, as the U-Net underestimates the warming for the model with the highest transient climate sensitivity during validation. Performance is lower for variables like sea-level pressure and precipitation because of their low ratio of forced to internal variability. This framework might be enhanced by incorporating a larger multi-model dataset in the training and validation.
Read the full abstract

All authors Guillaume Gastineau, Elena Provenzano, Constantin Bône, Nicolas Lebas, Robert C. J. Wills

Combinatorics

Purely Periodic Three-move Subtraction Games

We determine the full Sprague-Grundy sequence and its least period for a class of three-move subtraction games whose sequences are periodic from the start. Write the move set as $S=\{a,b,c\}$, with $0<a<b<c$ and $\gcd(a,b,c)=1$. The case $a=1$ is known and is summarized separately. For $a\ge2$ and $c\ne a+b$, we give three explicit sufficient tests for pure periodicity, associated with the candidate periods $a+b$, $c+a$, and $c+b$. For fixed $a,b$, the tests depend only on $c\bmod(a+b)$ and are computed from the losing positions of the two-move game $\{a,b\}$. Whenever a test succeeds, we give the losing positions in closed form, reconstruct the remaining values, and prove that the least period is the smallest candidate whose test succeeds. The construction shows how the added move $c$ modifies the two-move pattern to form a repeating block. We also give a uniform formulation of the known additive case $c=a+b$. For $a\ge2$, we conjecture that the tests cover every purely periodic non-additive game. In this range, when $c\ge2(a+b)$, we prove necessity for purely periodic games with least period at most $a+b$.
Read the full abstract

Probability

One-Cut Risk Profiles under Quadratic Loss: Discrete Convexity, Continuous Limits, and Higher Dimensions

In this note we study a two-regime representation of a loss random variable under quadratic error. For a finite law we compute exactly the change of the optimal risk when one atom crosses the cut. This turns the problem into a convexity question in cumulative-mass coordinates. On an equally spaced support, log-concavity gives this convexity, while weak symmetry locates the optimal cut, with an additional correction when the mean lies between two atoms. We also discuss the continuous analogue and extend the main identities to finitely supported random vectors, where a global optimal partition may be chosen as a halfspace.
Read the full abstract

Systems and Control

Data-Driven Generator Transient Prediction for Digital Twin Decision Support

This paper develops a calibrated transient forecasting surrogate model for generator digital twin (DT) decision support that evaluates planned active- and reactive power load commands before they are applied. The proposed event-conditioned Hankel Dynamic Mode Decomposition with Control (Hankel-DMDc) model combines delay-coordinate lifting, command-event memory features, and an event-weighted Hankel basis so that sparse load-transition dynamics influence the reduced representation and fitted dynamics. This design targets intervals where voltage/frequency deviations and recovery behavior determine whether a candidate load command keeps the system within acceptable limits. To provide operator-facing confidence information, a split-conformal calibration layer is applied to the frozen surrogate model to form event-conditioned joint prediction bands for voltage and frequency. The experimental results show event-window root-mean-square errors of 1.058 V and 0.155 Hz. For a nominal 90% target, the bands attain 90.17% pointwise joint voltage-frequency coverage, with mean band widths of 3.55 V and 0.566 Hz. A 50-s open-loop rollout is computed in 181 ms on a single CPU core, approximately 280 times faster than real time, with forecast accuracy evaluated over horizons up to 5 s. These results demonstrate a computationally efficient advisory framework for generator DTs that combines transient prediction with calibrated uncertainty.
Read the full abstract

All authors Emad Sadeghi, Emerson Miller, Kerry Sado, Adel Nasiri

Quantum Physics

Fundamental Limits of Quantum Metrology Beyond Fixed Causal Order

Quantum metrology with indefinite causal order (ICO) has attracted intense interest due to its potential to surpass the limitations of conventional fixed-order strategies. A key open question is whether ICO can fundamentally enhance asymptotic precision scaling. In this work, we bridge this gap for the estimation of a single parameter encoded in $N$ identical uses of a finite-dimensional quantum channel. We first establish a universal Heisenberg-scaling upper bound for the full general ICO process-matrix class and show that for unitary channels its optimal quantum Fisher information (QFI) coincides exactly with that of parallel strategies. For noisy channels, a structurally refined bound shows that channels restricted to the standard quantum limit (SQL) under parallel strategies remain SQL-limited under general ICO strategies. Most significantly, an asymptotically tight (AT) bound is derived to close the remaining possibility of an asymptotic ICO advantage by showing that general ICO and optimal parallel strategies have exactly the same leading QFI coefficient in both the SQL and Heisenberg regimes. For the operationally motivated class of quantum circuits with quantum control of causal order, we further obtain an iterative constraint on finite-query precision whose asymptotic limit agrees with that of the AT bound. Our results clarify the ultimate role of indefinite causality as a metrological resource for quantum channel estimation.
Read the full abstract

Analysis of PDEs

Existence of thermodynamically consistent solutions for data-driven porous media problems

Data-Driven Computational Mechanics (DDCM) replaces traditional phenomenological constitutive models by directly reformulating boundary-value problems in terms of local material state data obtained from experiments or fine-scale simulations. Standard DDCM formulations, only enforcing equilibrium and compatibility, do not inherently guarantee compliance with the second law of thermodynamics. This breakdown occurs particularly when input material data sets are subject to noise or local physical non-admissibility. In this work, we present a variational DDCM framework specifically tailored to diffusion–reaction problems. Taking advantage of the simplicity of the thermodynamic constraint in gradient-flux systems, we propose an augmented formulation that explicitly enforces the second law of thermodynamics as a hard constraint within the energy-minimization problem. Although the set of thermodynamically admissible states is non-convex and fails to be weakly closed in the ambient phase space, we establish existence of minimizers by proving that the intersection of the admissible set with the subspace of fields that are compatible and in equilibrium is weakly sequentially closed via a compensated compactness argument. To enable practical computations, we analyze both a Lagrange multiplier formulation and a penalization scheme. We prove the $\Gamma$-convergence of the penalized functionals to the exact constrained problem and establish a fully discrete convergence framework incorporating spatial finite-element discretization and empirical data-set approximations. Numerical experiments confirm that the proposed penalty scheme effectively restores thermodynamic consistency even in the presence of severely corrupted material data.
Read the full abstract

Probability

Higher Order Unbounded Rough Drivers

We construct a framework for higher order unbounded rough drivers defined by rough paths over vector fields. This allows us to consider a purely Eulerian perspective for rough transport equations where the driving rough path vector field is allowed to have temporal $\mathfrak{p}$-variation for any $\mathfrak{p} \in (1,\infty)$. We use this framework to show the well-posedness of an equation with two transport terms; one driven by a rough path vector field with aforementioned $\mathfrak{p}$-variation in time, and the other driven by a DiPerna–Lions-type drift vector field.
Read the full abstract

High Energy Physics (Phenomenology)

Displaced Signals from Long-lived Particles in Neutrinoless Double Beta Decay

Much of the literature on neutrinoless double beta ($0\nu\beta\beta$) decay with light new-physics mediators focuses on invisible missing energy. We investigate the scenario in which a massive Majoron-like particle $\phi$ is produced on-shell during $0\nu\beta\beta$ decay and subsequently decays into visible final states after travelling a macroscopic distance. Specifically, we analyze the displaced energy deposition from $\phi$ decays into a photon pair ($\gamma\gamma$), a photon and a dark photon ($\gamma\gamma_D$) and an electron-positron pair ($e^+e^-$). We demonstrate that the displaced decays modify the expected visible energy spectra and provide novel, distinct experimental signatures at current and upcoming $0\nu\beta\beta$ experiments, with the promise of improving the sensitivity of the standard invisible Majoron searches in $0\nu\beta\beta$ decay by more than two orders of magnitude. The relevant effective couplings can naturally arise in well-motivated ultraviolet-complete scenarios that conventional $0\nu\beta\beta$ decay searches cannot probe.
Read the full abstract

All authors Patrick D. Bolton, Noor-Inès Boudjema, Frank F. Deppisch, Chandan Hati, Chayan Majumdar

Computer Vision

MEOX: Modality-Conditioned Routing Before Fusion in a Compact Earth Observation Encoder

Multimodal Earth Observation models often fuse radar and optical inputs early, limiting sensor-conditioned processing, or maintain separate sensor streams through the encoder, increasing computational cost. We present MEOX (Multimodal Earth Observation with eXperts), which delays fusion until after a shared mixture-of-experts block that processes Sentinel-2 and ascending and descending Sentinel-1 streams independently. Co-located patches are then fused with validity-aware weights into a single sequence processed by fourteen shared blocks, reducing the sequence length from 3N + 5 to N + 5 tokens for N patches. We pretrain the 2.939M-parameter encoder on 1.228M MMEarth64 samples using masked radar-optical reconstruction. Under the reported frozen-transfer protocols, MEOX reaches 90.56% average class accuracy on EuroSAT and 64.42% mean intersection over union (mIoU) on cashew segmentation, 2.26 and 5.02 points above the strongest reported cross-sensor mixture-of-experts (CSMoE) variants. Full-encoder adaptation increases BigEarthNet micro-average precision from 57.22% to 72.95%. In a controlled same-corpus test using synthetic random patch removal, masking 75% of Sentinel-2 patches reduces WorldCover mIoU from 30.45% to 26.53% when both Sentinel-1 streams remain available; with Sentinel-2 alone, mIoU falls to 13.81%. For identical zeroed optical patches, declaring them missing through the validity mechanism improves mIoU from 19.42% to 26.53%. These results characterize a compact route-then-fuse encoder that retains sensor-conditioned processing and uses complementary radar observations when optical inputs are incomplete.
Read the full abstract

High Energy Astrophysical Phenomena

Electromagnetic alignment and jet precession around supermassive black holes: Quasi-periodic oscillations in tidal disruption events

We evaluate quasi-periodic oscillations and jet formation in tidal disruption events using the covariant formulation of electromagnetic angular-momentum transfer. General-relativistic frame-dragging tears apart misaligned transient accretion flows, forming an isolated inner mini-disk. The accumulation of magnetic flux on the event horizon powers a relativistic jet via the Blandford-Znajek mechanism. Because the magnetic field anchors to the precessing mini-disk, the jet axis rotates, generating geometric modulations in the observed X-ray and radio fluxes. To ensure physical consistency with the force-free magnetosphere required to launch a Blandford-Znajek jet, we model the electromagnetic back-reaction using a split-monopole magnetic field topology. By performing a small-spin expansion of the Noether current density over the event horizon, we derive a closed-form analytical reaction torque exerted by the electromagnetic field on the accretion plasma. We evaluate the resulting kinematics to show that the electromagnetic back-reaction induces a retrograde precession of the mini-disk, coupling with the prograde Lense-Thirring precession to dictate the global oscillation frequency. We formulate explicit predictions for observable transient signals and predict a monotonic attenuation of the peak-to-trough flux ratio as the mini-disk aligns, as well as a specific frequency drift signature characterized by an initial lengthening followed by an asymptotic shortening of the time interval between consecutive flares. We establish an analytical mechanism where magnetic flux depletion stalls alignment, predicting a constant residual modulation amplitude at late times. We formulate a methodology to extract the black hole spin and the magnetic flux density directly from the temporal derivatives of this predicted frequency drift, operating independently of spectral continuum fitting.
Read the full abstract

All authors Pau Amaro-Seoane, Leif Lui, Alejandro Torres-Orjuela, Xian Chen

Metric Geometry

Arithmetic Polyhedra

The Koebe-Andreev-Thurston theorem assigns a 3-dimensional hyperbolic reflection group to each combinatorial polyhedron. A natural question is: which of them are arithmetic? In 2016, Kontorovich-Nakamura conjectured that all arithmetic reflection groups obtained in this way are commensurable to those obtained from the tetrahedron, square pyramid, or cuboctahedron. In this paper, we prove the conjecture. It is a consequence of the following result of independent interest: all arithmetic ideal, right-angled hyperbolic polyhedra are obtained by gluing together copies of one of three “seed” polyhedra.
Read the full abstract

All authors Daniel Allcock, Pat Devlin, Anna Felikson, Alex Kontorovich, Ian Whitehead

Complex Variables

Sharp exponential integrability of conjugate functions

We prove that if a real-valued function $f\in L^1$ on the complex unit circle has a gap of width at least $\pi$ in its essential range, then $\exp(\tilde f)$ is not integrable, where $\tilde f$ is the conjugate function. More generally, the exponential function can be replaced by any nonnegative convex function $Y$ satisfying $$ \int_0^\infty Y(x) e^{-x}\, dx = \infty. $$ We also establish a local version on arcs of the unit circle: Under a natural condition on the inverse images of the two sides of the gap, $Y(\tilde f)$ fails to be integrable on the arc; in particular, it suffices that one of these inverse images is not an interval modulo null sets. Furthermore, we obtain corresponding results for complex-valued functions and a $Y$-integrability version of Zygmund's result.
Read the full abstract

Human-Computer Interaction

TherMosaic: Accelerating Perceived Thermal Transitions Through Spatiotemporal Thermal Feedback

Thermal feedback can enrich immersive interaction, but thermoelectric devices often change temperature too slowly to match interactive timing. We present TherMosaic, a spatiotemporal thermal feedback approach that accelerates perceived temperature transitions by leveraging two perceptual mechanisms: spatial summation and thermal adaptation. Focusing on the fingertip, we first investigate this approach using a custom 2*2 array of independently controlled Peltier modules. Across three controlled perceptual studies, we show that distributed thermal stimulation can preserve stable hot and cold percepts despite local deviations, that adaptation helps maintain these percepts during changing stimulation, and that combining these effects reduces perceived transition time by about 30%-40% for transitions originating from hot or cold states. We then translate the same design principles into a standalone wearable implementation of TherMosaic and evaluate it in virtual reality. Our results show that this approach reduces perceived thermal lag and improves temporal alignment between thermal and visual events in interactive use.
Read the full abstract

All authors Zining Zhang, Jiasheng Li, Myungin Lee, Zeyu Yan, Jin Ryong Kim, Huaishu Peng

Artificial Intelligence

Who Should Grade My Work? Student Perspectives on Transparent AI-Assisted Writing Assessment in Higher Education

The integration of GenAI tools into higher education assessment raises important questions about how students understand, interpret, and respond to AI-mediated evaluation. As instructors increasingly explore AI tools for providing feedback, prior research has examined whether GenAI-generated feedback improves writing performance and how students perceive its usefulness; comparatively little is known, however, about how students interpret such evaluation when they are explicitly informed that an AI system, rather than a human instructor, produced the feedback and the score. This study reports findings from a qualitative pedagogical inquiry conducted in an undergraduate technical communication course for computing students at a Saudi public university. Thirteen male undergraduate computing students completed an in-class handwritten writing task; the scanned submissions were evaluated by ChatGPT using a rubric-based prompt aligned with the task objectives. Students were then explicitly informed that ChatGPT had generated the score and feedback and were invited to reflect on the evaluation in writing. Inductive thematic analysis of these reflections identified four themes: perceived usefulness of feedback; awareness of AI's contextual and pedagogical limitations; conditional trust, distinguishing feedback utility from evaluative authority; and reflection on the institutional and pedagogical role of the human instructor. Participants accepted GenAI feedback as useful for surface-level revision but consistently positioned the human instructor as the appropriate authority over grading decisions. The study identifies this as a distinction between feedback utility and evaluative authority, two judgments that students treat as analytically separate rather than as opposite ends of a single approval scale...
Read the full abstract

Computers and Society

Moral Advice as Interactional Negotiation: Framing, User Pressure, and Social Position in Large Language Model Responses

As conversational AI becomes a source of everyday guidance, LLMs increasingly participate in the interpretation and legitimation of morally contested choices. We examine LLM moral advice as an interactional negotiation shaped by framing, sustained user pressure, and the moral subject's social position. Using GPT-4o-mini as an illustrative case, we conducted a factorial vignette experiment with a pre-specified three-round protocol. The model received eldercare dilemmas that varied in framing and persona, followed by two user challenges. We analyzed 1,620 configuration-framing cells, each repeated three times, yielding 4,860 conversational runs. Caregiving affirmation produced near-uniform endorsement, whereas non-caregiving framing produced more variable baseline stances. When users challenged caregiving endorsement, 90.1% of configurations shifted after one round. Non-caregiving framing produced more resistant and unstable trajectories. Never (27.9%) and Late (25.6%) accommodations were more common than Early accommodations (16.5%), and only 14.32% of configurations achieved perfect trajectory consistency, compared with 62.72% under caregiving framing. Advice also varied with social position. Female personas received more support for non-caregiving decisions, while the presence of sisters increased accommodation. The GPT-4o-mini case shows that LLM moral advice can develop through a partially stable negotiation between normative response tendencies and user pressure rather than express a fixed ethical framework. The framework and design support comparative research across models and moral domains. Such instability raises social, ethical, and technical concerns, as users may treat advice that is difficult to scrutinize as objective.
Read the full abstract

Quantum Gases

How dipolar interactions structure molecular droplets

We investigate how dipolar interactions between microwave-shielded polar molecules structure the self-bound droplets formed under variation of the interaction strength. We identify the transition from droplets to crystals as a finite-size first order transition. With droplet-ring states and transitional supersolid states we predict additional structure in the crystal and droplet phases, respectively. To describe this strongly correlated regime, and in particular the reconfiguration of quantum ground states, we design a variational Monte Carlo framework based on neural quantum states. It is especially suitable to describe ground states and almost degenerate states with very different configurations. Moreover, one can easily determine the superfluid fraction. Our results reveal the sequence of finite-size structures through which dipolar interactions reorganize molecular droplets into crystals.
Read the full abstract

Probability

On the phase transition for the number of collisions on comb graphs

We consider collisions of simple random walks on comb graphs $\mathrm{Comb}(\mathbb{Z},H)$, which are obtained by attaching vertical segments of the form $[0,H_x] \cap \mathbb{Z}$ to any point $x$ of the integer axis. For $\mathrm{Comb}(\mathbb{Z},H)$ with profile $H_x(x) = |x| \log^\gamma(|x| \vee 1)$, we show that two independent simple random walks starting from the same site collide infinitely often almost surely if $\gamma \leq 2$. If the tooth profile is taken as a typical realization of i.i.d. heavy-tailed random variables with $\textbf{P}(H_x > z) \sim Cz^{-\gamma}$ (with some $C > 0$) as $z$ tends to infinity, we show that infinitely many collisions occur almost surely for two independent random walks if $\gamma > 1/3$, whereas finitely many collisions occur almost surely if $\gamma \in (0,1/3)$, and for any $\gamma \in (0,1]$, three independent random walks only collide finitely many times, almost surely.
Read the full abstract

All authors Umberto De Ambroggio, Jenson Ng, Maximilian Nitzschner, Carlo Scali

Physics and Society

Mitigating Disease Spread by Design in Refugee and IDP Camps

Disease spread represents an increasing challenge in refugee and internally displaced person (IDP) settlements. The movement and interaction of people within camps is influenced by their layout, which therefore has the potential to significantly affect disease spread. This work aims at creating a methodology to explore the potential effects of different camp layouts as mitigating factors in the spread of diseases within settlements. We showcase proof-of-concept experiments by leveraging the JUNE agent-based epidemic model, discuss the kind of operational insights this methodology can facilitate, and provide a framework for future investigations.
Read the full abstract

All authors Giulia Zarpellon, Joseph Aylett-Bullock, Frank Krauss, Miguel Luengo-Oroz

Combinatorics

Bounded ratios for Lorentzian polynomials

We study multiplicative inequalities among the coefficients of Lorentzian polynomials through the notion of bounded ratios. Our main structural result completely characterizes the cone of bounded ratios for Lorentzian polynomials of degree $n$ in $k$ variables. We show that the dual of the cone of bounded ratios is generated by equivalence classes of M-convex functions modulo affine functions. For ternary Lorentzian forms of arbitrary degree $n\ge3$, we show that the cone of bounded ratios is generated by triangular ratios and determine the optimal bounding constant of every bounded ratio. Furthermore, we characterize the pairs $(n,k)$ for which the cone of bounded ratios can be computed by tropicalizing products of $n$ linear forms in $k$ variables with nonnegative coefficients.
Read the full abstract

Multiagent Systems

Trust-Aware Adaptive Disclosure for Inference Privacy Preservation in Multi-Agent Networks

Agent based systems are increasingly deployed in information critical systems including healthcare management systems, and smart grids. In this paper, we consider a multi-agent system where each agent has a latent goal that needs to be kept hidden from observing adversaries. More specifically, this paper studies privacy-preserving consensus in networked multi-agent systems under goal inference attacks. We propose a Trust-Aware Privacy Control framework that adapts message disclosure based on the dynamic trust relationships between agents. The proposed method controls information release using a trust-dependent stochastic policy. This enables a tradeoff between consensus performance and privacy preservation. Experiments demonstrate that the proposed method reduces adversarial goal inference accuracy compared to representative baselines, while maintaining competitive consensus utility, thereby highlighting the effectiveness of trust-aware mechanisms in privacy preservation of the agents in multi-agent systems.
Read the full abstract

Artificial Intelligence

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability

Model upgrades are routine; memory migrations are not. An agent can keep the same memory store and still forget: a new model may interpret old notes differently, mixed embedding versions may break retrieval, and repair may fail without the original evidence. We compare memory as the same history is preserved verbatim for long-context reading (LC-RAW), divided into chunks for retrieval-augmented generation (RAG), compressed by a model into natural-language notes (NOTES), or normalized into a fixed-schema knowledge graph (KG-fixed). The study uses 48 synthetic histories with randomized answer codes, exact scoring, and two open-weight models with sub 10 billion parameters. Our measurements show that fixed-schema structures transfer reliably, with KG-fixed accuracy changing by only $+0.0004 \pm 0.0020$ following a writer swap. Conversely, compressed NOTES exhibit high model coupling, with accuracy shifting asymmetrically by $+9.91$ or $-13.28$ percentage points depending on the specific migration direction. In RAG systems, partial embedding migrations using a 50/50 mixed index capture only a 4.96-point accuracy improvement, forfeiting the majority of the 11.90-point gain achieved through full re-embedding. Diagnostic decomposition attributes 80% ($0.467 \pm 0.014$) of the NOTES accuracy deficit to information lost during initial construction, whereas retrieval failures drive 81% ($0.364 \pm 0.012$) of the RAG deficit. Finally, store-only repair of NOTES fails to reach a 90% performance recovery target in all 48 test cases, whereas retaining the raw source history enables successful recovery in 34 of 48 cases for one tested direction. These findings highlight the necessity of direction-specific migration testing, strict embedding space isolation, and the retention of source histories for memory repair.
Read the full abstract

Classical Analysis and ODEs

A Log-Free $n^{1/5}$ Bound for Chowla's Cosine Problem

For a finite set $S$ of positive integers, put $K(S):=-\min_{x\in\mathbb T}\sum_{s\in S}\cos(2\pi sx)$. Bedert recently proved the uniform lower bound $K(S)\geq |S|^{1/5-o(1)}$. We remove the subpolynomial loss and prove that $K(S)\geq c|S|^{1/5}$ for an absolute constant $c>0$. The proof combines two estimates from Bedert's argument with an exact averaging identity for the asymmetric boundaries of additive intersections. This identity replaces the multiplicative-amplification step responsible for the logarithmic loss.
Read the full abstract

Machine Learning

Variational Continuation for Double Pendulum Periodic Orbits

We present a Hessian-based approach to numerically continue periodic orbits in dynamical systems. A loop (periodic orbit candidate) is parametrized as a Fourier series; a loss function is defined based on the deviation of the loop from the physical differential equations. Unlike previous work relying on hand-derived Jacobians, our method automates the process by leveraging automatic differentiation, a common machine learning technique. The continuation direction can be determined by the flat directions of the loss landscapes (directions with zero eigenvalues), making the search of periodic orbits efficient and guided. Our method is integrator-free, precisely initializes oscillations around unstable fixed points, and efficiently detects orbit family intersections and subharmonic bifurcations. As a demonstration, we present full continuations of periodic double pendulum oscillations from fixed points, showing bifurcations along orbit families and categorizing branches of periodic orbits. In particular, we find periodic orbits where both pendulum masses are never simultaneously at rest, which to our knowledge has been missing in the literature.
Read the full abstract

High Energy Physics (Theory)

The Junction That Remembers: Infrared Response of Two Coupled SYK Majorana Systems

Two strongly interacting Majorana systems of the Sachdev–Ye–Kitaev type, each with its own independent disorder, are joined by a link that is switched on in time. We ask how the infrared structure of the link is imprinted on the response and whether the system remembers how it was opened. At large $N$ the leading cross-correlation between the two systems is an exact product of two equilibrium propagators and the switching profile; because the disorder is independent, no ladder dresses it. The link operator has scaling dimension one half, which gives its susceptibility a universal logarithm whose coefficient and phase we fix and verify numerically, while the gap of the coupled equilibrium problem grows more slowly than the naive crossover estimate. Exact dynamics for up to twelve Majoranas per side show strong parity transfer and entanglement several times the free-fermion ceiling. The work done by the switch has a first-order part that does not depend on the switching time and a second-order part that grows in magnitude with it. Once the work is held fixed, the protocol dependence of the entanglement, the link expectation value and a scrambling diagnostic nearly vanishes, and the small residuals do not depend on whether the protocols differ at low or at high frequency. These small systems have no infrared window, so they neither confirm nor exclude memory at large $N$.
Read the full abstract

Cryptography and Security

The History Is the Detector: Executing CVE Patch History, End-to-End

Public vulnerability databases collect rich information about known software flaws, including their weakness types, affected components, and related patches. Fixing commits provide the exact code changes that removed these flaws. While these records capture why the original code was unsafe, they are documented mainly for human inspection rather than automated reuse. Consequently, the same unsafe conditions may still exist elsewhere in code without a known advisory, leaving much of this detection knowledge unused. We present BUGSTONE-E2E, a framework that transforms vulnerability history into executable detection rules and validates their findings. First, BUGSTONE-E2E mines reusable rules from verified fixing commits, capturing scan anchors, fix semantics, and CVE provenance and organizing them by CWE and language. Second, detection follows a funnel-shaped pipeline: early stages process a large pool of candidates using lightweight analysis, while later stages apply increasingly capable and expensive models to a shrinking set of targets. Specifically, BUGSTONE-E2E first enumerates call sites matching rule anchors using Tree-sitter, then removes benign sites using lightweight heuristics without LLM calls. Next, LLM-based agents inspect the remaining candidates guided by the rule. Following this inspection, the system re-triages surviving candidates and builds runtime verifications, then generates scope-checked patches validated via two-sided differential tests. Using 19,325 high-severity CVEs from 2022 to 2026, BUGSTONE-E2E identifies 2,710 fixing commits and constructs 1,033 detection rules across 56 CWE families, packaged into 172 skills. When applied across 14 programs, it produced runtime evidence for 644 findings. These results demonstrate that CVE history can be turned into an executable workflow, transforming past vulnerabilities into reproducible detection and repair.
Read the full abstract

All authors Qiushi Wu, Kevin Eykholt, Youngja Park, Xiaokui Shu, Dhilung Kirat, Douglas Lee Schales, Ian Molloy