Peer Revenue Publish or profit

Papers

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

All of arXiv

Wednesday 14 October · still arriving

Artificial Intelligence

HEM-Med: A Hierarchical Agentic Memory Framework for Safe Multi-Center Medication Recommendation

Min Xiao, Jin Li, Shoujin Wang et al. (+3 more)

Medication recommendation (MR) is critical for clinical decision support. While existing MR methods are mostly designed for single-center learning, where only one independent hospital is involved, multi-center MR with hospitals of different scales has become increasingly demanding in practice. In multi center settings, abundant data from large medical centers can improve recommendation performance for low-resource centers. Despite a few preliminary attempts on this task, two critical gaps remain. Gap 1: Existing multi-center methods fail to disentangle transferable cross-hospital clinical experience from hospital-specific prescribing patterns, limiting clinical feasibility and physician acceptance. Gap 2: They commonly overlook multi label safety constraints of MR, including drug-drug interaction risks and local formulary feasibility, which are essential for safe real-world deployment. To address these gaps, we propose HEM-Med, a novel hierarchical agentic memory framework for multi-center MR. To address Gap 1, we design a disentangled hierarchical memory module that builds patient-level medication context attribution records and disentangles them into cross hospital experience memory and hospital-specific residual memory. To address Gap 2, we introduce a safety-aware medication set generation module, which integrates dual-level experience retrieval with safety-aware set decoding to produce clinically relevant, safer, and locally adaptive medication recommendations. Experiments on a real-world multi-center eICU dataset show that HEM-Med outperforms representative transfer learning, LLM-based, and MR baselines in terms of both recommendation accuracy and safety.
Read the full abstract

All authors Min Xiao, Jin Li, Shoujin Wang, Yeguang Zhen, Yishuo Li, Wenpeng Lu

Artificial Intelligence

Value-Guided Repulsion for Offline Reinforcement Learning

Yixue Hao, Jiahao Zhang, Jiesong Lian et al. (+5 more)

Drifting models have emerged as a highly efficient and stable paradigm for policy design in offline reinforcement learning (RL). However, standard drift policies determine repulsion solely from geometric relationships among generated actions and therefore cannot distinguish them by expected value. To address this limitation, we introduce Value-Guided Repulsion Q-Learning (VRQL) to integrate action value estimates into the particle interaction field. For each state, VRQL combines geometric kernel distance with a normalized Q value scaling. This mechanism assigns greater repulsive influence to particles with lower values while strictly retaining attraction toward behavior action. Consequently, this value-guided interaction field aligns behavior regularization with policy improvement without adding an auxiliary behavior model stage. On 50 OGBench task configurations and 6 D4RL AntMaze configurations, VRQL demonstrates highly competitive performance against established baselines, suggesting that value-guided particle interaction is an effective method for offline RL.
Read the full abstract

All authors Yixue Hao, Jiahao Zhang, Jiesong Lian, Xufei Guo, Xumeng Guo, Wenhui Zhang, Long Hu, Rui Wang

Optics

From Single-Particle Thermal Asymmetry to Collective Control: Janus Plasmonic Colloids for Photothermal Applications

Cédric Rousseau, Javier González-Colsa, Victor Joris et al. (+3 more)

Janus nanoheaters provide a route to break the spatial symmetry of thermoplasmonic heating and generate directionally biased thermal gradients at the nanoscale. However, in realistic photothermal scenarios, nanoheaters rarely act as isolated objects, but rather operate within colloidal assemblies where collective thermal and electromagnetic effects reshape the local temperature landscape. Here, we investigate how the intrinsic thermal asymmetry of a single Janus plasmonic nanoheater is transformed when moving from isolated particles to planar random arrays under continuous-wave illumination. Using coupled electromagnetic and steady-state heat-transfer simulations, we first quantify the finite spatial range over which the single-particle Janus signature remains detectable. We show that neighboring nanoheaters introduce two competing effects: a nearly symmetric collective thermal background that screens and homogenizes the Janus factor, and short-range neighbor-induced asymmetries that partially compensate this screening when interparticle distances become comparable to the spatial extent of the single-particle thermal anisotropy. We derive a closed-form analytical model for the array-averaged Janus factor as a function of nanoparticle number and surface density. The model predicts the screening-dominated regime from single-particle quantities and identifies deviations associated with near-field neighbor-induced contributions at higher densities. These results demonstrate a central trade-off for biomedical photothermal applications in continuous-wave operation: collective heating is required to reach relevant temperature increments, but it progressively erases the directional Janus character by thermal homogenization. Our framework provides a practical route to predict and rationally design the spatial temperature distribution in plasmonic assemblies of Janus particles.
Read the full abstract

All authors Cédric Rousseau, Javier González-Colsa, Victor Joris, Pablo Albella, Bjorn Maes, Gilles Rosolen

Computational Complexity

Asymmetry and Complexity of Nondeterministic Computations

Halley Goldberg, Jinqiao Hu, Zhenjian Lu et al. (+2 more)

Symmetry of Information (SoI) [ZL70] is a fundamental theorem of classical Kolmogorov complexity whose status for time-bounded computation is closely related to central open problems in complexity theory and cryptography. In a recent development, [HMO26] showed that SoI fails for $rKt$ complexity, namely in the setting of randomized computation. By contrast, many basic properties of Kolmogorov complexity are known to remain valid in the presence of nondeterminism [HLO26], suggesting that SoI may provide a fruitful structural principle for exploring the relative power of randomness and nondeterminism in computation. Our results reveal that the validity of SoI for nondeterministic computation, namely for $nKt$ complexity, is a subtle question connected to several frontiers. We show that symmetry of information holds on average for randomized extensions of $nKt$, while its worst case validity would imply breakthrough complexity separations. Although the standard cryptographic argument for asymmetry does not extend to the nondeterministic setting, we provide evidence that information remains asymmetric in nondeterministic computations. Among other results, we show that SoI would imply a collapse of exponential time classes to low levels of the polynomial hierarchy. Then, building on the work of [Hir22b, GK22, KK26, HMO26], we develop a general framework connecting symmetry of information, meta-complexity, and explicit constructions in the nondeterministic setting. More precisely, we prove that suitable formulations of these three notions are quantitatively equivalent. We apply different aspects of this correspondence to make progress on complexity lower bounds and on the failure of SoI for nondeterministic computation. These consequences are unconditional and provide some of the strongest known lower bounds in meta-complexity.
Read the full abstract

All authors Halley Goldberg, Jinqiao Hu, Zhenjian Lu, Jingyi Lyu, Igor C. Oliveira

Machine Learning

Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning

Yanyi Li, Wendao Wu, Haihan Zhang et al. (+6 more)

Tracking changing comparators is a central problem in online learning. For exp-concave losses, previous work established the $T^{1/3}P_T^{2/3}$ dependence on the horizon $T$ and Euclidean path length $P_T$ for fixed dimension, while the optimal cost of dimension remained unresolved. We resolve this question by establishing a sharp $d^{1/3}$ curved dimension factor, improving the $d^{2/3}$ factor of recent switching and interval reductions [1, 2]. To achieve this improvement, we develop a proper, full-information learner on compact convex subsets of $\mathbb{R}^d$ that adapts to the unknown path length and, under fixed dimension-independent diameter, gradient, and curvature parameters, guarantees $$\operatorname{Reg}_T(u_{1:T}) \le \widetilde{O}\left(d+\min\left\{d^{1/3}T^{1/3}P_T^{2/3},\sqrt{TP_T},T\right\}\right).$$ A matching lower bound on a diameter-one Euclidean ball establishes the minimax frontier up to logarithmic factors when $T\gtrsim d^2$, including the transition from the curved rate to larger path budgets. The lower bound already holds for losses that depend on only one coordinate per round, isolating an unavoidable cost of learning across directions. Our upper bound follows from square-root Gaussian KL control of comparator movement and a cancellation argument that pays for history restarts. The curved learner maintains $O(\log^2 T)$ Gaussian states, uses one gradient per round, and runs in time near-linear in $T$ under a Mahalanobis-projection oracle.
Read the full abstract

All authors Yanyi Li, Wendao Wu, Haihan Zhang, Qinshi Wang, Chenheng Zhang, Chunyuan Zheng, Cong Fang, Haoxuan Li, Zhouchen Lin

Computation and Language

When Occupation-derived Skills Don't Help: On the Limits of Skill-Augmented Career Path Prediction

Ege Süalp, Elena Senger, Toine Bogers et al. (+1 more)

Skills are widely assumed to play a central role in career transitions, motivating their use in career path prediction (CPP) systems. In this work, we critically examine whether explicit skill representations provide complementary signal beyond text-based job representations. We evaluate multiple strategies for incorporating skills into CPP models, including score fusion and neural architectures, under two conditions: predicted skills and ground-truth ESCO annotations. Surprisingly, across two benchmark datasets, none of these settings yield substantial improvements over strong text-only baselines. Even when using ground-truth skills, gains remain marginal and inconsistent. Through detailed analysis, we identify key factors underlying this result: high redundancy between textual and skill representations, limited discriminative power of ESCO's taxonomy, and representational collapse in skill embeddings. Furthermore, we find that skill prediction difficulty correlates with overall sample difficulty, limiting the potential for complementary signals. Our findings challenge the common assumption that explicit skill modeling inherently benefits career prediction and suggest that dense text representations already capture most skill-relevant information.
Read the full abstract

All authors Ege Süalp, Elena Senger, Toine Bogers, Barbara Plank

Signal Processing

Radar-Only Long-Term Beam Prediction in mmWave Vehicular Networks: An Empirical Study of Doppler Input and Cross-Modal Distillation

Jeongwan Kang, Miguel Bordallo López, Nhan Thanh Nguyen et al. (+2 more)

Sensing-aided beam prediction could cut the beam-training overhead of millimeter-wave vehicular links, and two design choices could strengthen a radar-only predictor: a Doppler-resolved input and knowledge distilled from a camera during training. We measure what each is worth in received power rather than in top-1 accuracy, whose differences of a few points amount to only tenths of a decibel. We use real DeepSense 6G measurements, mainly at one roadside site, and one convolutional-recurrent predictor family evaluated open-loop up to 0.67 s ahead. Adding a range-Doppler map to the range-angle map lowers the mean beam-selection loss by 0.28 dB (95% bootstrap interval [0.14, 0.41]). The gain comes from the Doppler content, not from the extra input, and replicates, as pre-specified, at a second base station, although holding the last measured beam beats both predictors there. Cross-modal knowledge distillation from a motion-mask vision-radar teacher, by contrast, changes received power by -0.01 dB ([-0.07, +0.06]). The tested teacher, student, fusion and distillation variants show the same null in the metric each was scored on. Where distillation helps, in an exploratory run with few training passes, a radar-only teacher trained on all training passes reproduces most of the gain. For this predictor family and the full training set, the measurable gain at the link comes from Doppler-resolved radar content; no tested route for transferring training-only visual information adds a resolved gain.
Read the full abstract

All authors Jeongwan Kang, Miguel Bordallo López, Nhan Thanh Nguyen, Junhyeong Kim, Sunwoo Kim

Cosmology and Nongalactic Astrophysics

Galaxy-to-galaxy stochasticity has a limited impact on reionization and the 21-cm signal at fixed mean emissivity

Shihai Zhu and Yuxiang Qin

Galaxy-to-galaxy stochasticity may be less important for reionization, IGM heating, and the 21-cm signal than commonly assumed. We find that most of its apparent impact arises because introducing or removing scatter changes the mean emissivity of the galaxy population, rather than from redistributing radiation at fixed mean emissivity. We use the Meraxes semi-analytic model to separate these effects. For ionising sources, we vary the scatter in the escape fraction and in the SFR-halo mass relation across three source prescriptions. For the X-ray heating sources, we introduce luminosity scatter and also separately remove SFR scatter. Without recalibration, these modifications shift the mean emissivities, changing the timing of reionization and IGM heating and producing substantial differences in the 21-cm signal. Once the relevant mean emissivities are restored, however, the global ionizing and thermal histories converge and the power-spectrum differences fall below the forecast sensitivity of a deep SKA-Low observation. There are some weak changes in ionisation morphology, depending on which galaxies dominate the ionising emissivity and on their response to photoheating. Within the models and scales considered here, calibrating the mean radiative emissivities is therefore substantially more important than resolving their galaxy-to-galaxy scatter.
Read the full abstract

Commutative Algebra

Macdonald dual in the semilocal case and applications to local homology and local cohomology

Antonival Lopes do Nascimento Filho

This paper extends Macdonald's duality theory and Cuong-Nam's local homology theory – which is in some sense dual to the local cohomology theory of A. Grothendieck [Gro1] – to semilocal noetherian rings equipped with the $J$-adic topology. By overcoming the key structural obstructions of the semilocal setting via Zöschinger's structure theory and semilocal Matlis duality, this work establishes topological limit duality isomorphisms and links local homology with local cohomology. Ultimately, this framework provides the foundational machinery required to extend the classical results of Macdonald [Mac] and Cuong-Nam [C-N1] to the semilocal setting.
Read the full abstract

Number Theory

New bounds for $\mathrm{GL}(3)\times\mathrm{GL}(2)$ shifted convolution sums

Guangwei Hu and Guangshi Lü

Shifted convolution sums of $\mathrm{GL}(3)\times\mathrm{GL}(2)$ type have been extensively studied in several important settings, while a number of natural cases remain open. In this paper, we first revisit the shifted convolution sums involving $\lambda_\pi(1,n)$ and the divisor function $d(n)$, improving the recent power saving of $1/42$ obtained by Blomer and Li to $1/22$. We also treat the corresponding sums involving $\lambda_\pi(1,n)$ and $r(n,\mathcal{Q})$, settling a previously open case and obtaining a power saving of $1/22$. Here $\lambda_\pi(1,n)$ denotes the normalized Fourier coefficients of a Hecke–Maass cusp form $\pi$ on $\mathrm{SL}(3,\mathbb{Z})$, and $r(n,\mathcal{Q})$ denotes the number of representations of a positive integer $n$ by a positive definite integral binary quadratic form $\mathcal{Q}$. The proof combines two circle methods with a different organization of the major-arc analysis. A key point is that the nonzero Poisson frequencies are kept together in a single coefficient vector, rather than estimated separately, which allows them to be controlled by a finite matrix norm. The zero Poisson frequency is treated separately: its contribution is independent of the reduced residue class, and its arithmetic mean cancels the constant term in the complementary Jutila weight before absolute values are taken. We make this cancellation explicit, including the relevant normalization and local factors. The same mechanism extends to the binary quadratic-form setting, while the remaining Jutila contribution is handled by the available correlation estimates.
Read the full abstract

Sound

Noise-Aware Lombard Speech Synthesis via Flow-GRPO for Intelligibility Optimization

Zhaoyang Wang, Dingding Yao, Chengzhong Wang et al. (+1 more)

Voice-based conversational agents are increasingly deployed in real-world settings, where environmental noise often degrades the interactive experience. Therefore, robust acoustic adaptability is essential to maintain natural and intelligible human-computer communication. However, most existing systems remain acoustically insensitive and fail to emulate the human Lombard effect, in which speakers involuntarily adjust their vocal production to preserve intelligibility in the presence of noise. As a result, their communicative reliability degrades under adverse acoustic conditions. To remain intelligible in noisy environments, conversational agents require a mechanism to emulate noise-induced adaptive behavior during synthesis. We propose NIL-TTS, a noise-induced Lombard TTS framework. We integrate a feature-wise linear modulation (FiLM) mechanism into a flow-matching backbone for dynamic vocal adaptation. A two-stage training paradigm is utilized: initial supervised fine-tuning on paired Lombard corpora, followed by Flow-GRPO post-training to optimize for intelligibility. Evaluations confirm that NIL-TTS autonomously adjusts its output in response to varying noise conditions, exhibiting key correlates of the human Lombard effect. Driven by these adaptive modulations, the synthesized speech achieves significant intelligibility improvements that scale proportionally with noise intensity. Ultimately, this work presents a flow-matching and reinforcement learning-based advancement within the adaptive Lombard speech synthesis paradigm for challenging real-world scenarios.
Read the full abstract

All authors Zhaoyang Wang, Dingding Yao, Chengzhong Wang, Junfeng Li

High Energy Physics (Theory)

Finite Modular Hamiltonians and Their Singular Limits

Panyinyan Li, Jun Tao and Haitang Yang

Using the open solid torus as a finite parent, we revisit modular flow for spherical regions in conformal field theory. Its nonsingular angular evolution gives finite modular generators in a very simple form, while an exact conformal map to a cavity frame recovers the Rindler, CHM, and stationary-source ball constructions as singular limits. The same finite parent also provides a regular completion of the center obstruction in the complexified ball flow. Charged, Dirac, and inhomogeneous scalar examples are given.
Read the full abstract

Computation and Language

Cross-Model In-Context Neurofeedback Reveals (Limited) Activation Control in LLMs

Davide Baldelli, Robyn Kozierok, Mirko Bronzi et al. (+1 more)

Probes on activation vectors are widely used to detect misbehavior in language models, but their success relies on the assumption that models cannot control what their activations encode. Humans can learn, through neurofeedback, to control the firing of even single neurons, and recent studies investigated whether this ability extends to language models, reaching conflicting conclusions. No experimental design so far disentangles competence at the task from genuine access to one's own activations. We fill this gap with in-context neurofeedback sessions, in which a "subject" model is asked to maximize (or minimize) a score derived from its activations, shown back to it at every turn. In our first experiment, the sentences are generated either by the subject or by a second "writer" model, while the score always derives from the activations of the subject. We test ten open-weight models from five families, with sizes from 9B to 32B. Some models move their own scores considerably, but no more than a writer of comparable competence moves them. Is this because the task requires no privileged access, or because the subject controls its activations while processing the sentences, whoever wrote them? To answer this, we prefill a fixed sentence as the subject's reply at every turn and ask it to increase or decrease the score. Three out of ten models show small but consistent control effects. We further ask whether models can control their activations to the point of fooling an activation monitor. We make models repeat sentences a monitor would flag, such as a deceptive answer for a deception monitor, and ask them to raise or lower the monitor's score. No model learns to lower it, but one model can raise it. In-context control of a model's own activations might not yet be a threat to activation monitors, but it exists in a few models and needs to be tracked as models evolve.
Read the full abstract

All authors Davide Baldelli, Robyn Kozierok, Mirko Bronzi, Damiano Fornasiere

Probability

Entropic Inequalities and Limit Theorems in Boolean Probability

Kewei Pan

We study Boolean entropy and Fisher information in information inequalities and the Boolean central limit theorem. We prove a fractional Stam inequality for the Boolean Fisher information associated with arbitrary fractional partitions of subset sums, and derive an entropy power inequality for sums of Boolean independent random variables and its weighted version. We then establish a Boolean analogue of Barron's entropy convergence result along the Boolean central limit theorem: Boolean entropy converges to the entropy of the Rademacher law if and only if the initial entropy is finite. Furthermore, under a finite fourth moment, in the Boolean central limit theorem, we obtain \[ \Gamma(\mu_n\mid\mathbf b)= \frac{m_4(\mu)-1-m_3(\mu)^2}{2}\frac{\log n}{n} +o\!\left(\frac{\log n}{n}\right), \] with an explicit example showing sharpness. For symmetric laws with finite fourth moment and finite Boolean Fisher information, \[ \Psi(\mu_n\mid\mathbf b) =2\bigl(m_4(\mu)-1\bigr)\frac{\log n}{n} +o\left(\frac{\log n}{n}\right). \] Our approach is based on the Nevanlinna representation of Cauchy transforms.
Read the full abstract

Symplectic Geometry

Quantization of positive Lagrangians

Jake P. Solomon and Bentzion Yavetz

A Lagrangian submanifold in an almost Calabi-Yau manifold is called positive if the real part of the holomorphic volume form restricted to it is positive. It is called $\epsilon$-positive if the ratio of the restrictions of the imaginary and real parts of the holomorphic volume form is bounded by $\frac{1}{\epsilon}.$ An exact isotopy class of positive Lagrangian submanifolds admits a natural Riemannian metric. We prove that the Riemannian metric restricted to the subset of $\epsilon$-positive Lagrangians induces a path metric when the relative Kähler class of the ambient Calabi-Yau manifold lies on a line of rational slope. The proof relies on the construction of a sequence of quantization maps from a circle bundle over the space of positive Lagrangians to the dual of the space of holomorphic sections of powers of a prequantum line bundle. The rationality condition for the relative Kähler class allows us to reduce to the case of Bohr-Sommerfeld Lagrangians. Building on ideas of Tian and Nirenberg-Wells, Hörmander estimates are used to show that functions on a Bohr-Sommerfeld Lagrangian can be approximated by holomorphic sections of a sufficiently high power of the prequantum line bundle. Consequently, the quantization maps eventually separate fibers of the circle bundle.
Read the full abstract

Dynamical Systems

Reciprocal Dynamics in Congruence Jumping: Integral Orbits and Valuation Cones

Max A. Alekseyev and Dmitry I. Khomovsky

For fixed positive degrees $n,m$, alternating monic reciprocal transformations define a rational discrete dynamical system on the finite-dimensional coefficient space of polynomial pairs. This system arises from congruence jumping in polynomial divisibility problems, where a quotient move changes the polynomial state, but its coefficient dynamics can be studied independently. We prove that every state with nonzero constant terms returns after four alternating updates to its rational gauge classes and classify when this return orbit is realized over $\mathbb Z$. The local obstruction is encoded by endpoint $p$-adic valuation defects. After the first integral block, constant-term valuations evolve by an explicit matrix $M_{n,m}\in\operatorname{SL}_2(\mathbb Z)$ whose trace depends only on $s=nm$. This yields a finite-order, elliptic-type regime for $s<4$, a parabolic dilation regime for $s=4$, and a hyperbolic regime for $s>4$. In the hyperbolic case, invariant thresholds, irrational eigencones, and discrete valley formulas give finite criteria for forward and bi-infinite integral orbits; nevertheless, failure can be delayed arbitrarily long. At the pointwise level, separated auxiliary-relation transport factors through the jumped-variable projection. One jump preserves the visible support rank, while two alternating jumps collapse it to dimension at most one; saturation cannot restore rank. In the quadratic specialization, a rank-two $p$-adic relation lattice gives an exact local cancellation-depth formula and a finite criterion on periodic Vieta orbits. We classify the fixed auxiliary relations that saturate indefinitely on an irreducible six-cycle and construct arbitrarily long persistence on parabolic and fixed hyperbolic orbits.
Read the full abstract

Computation and Language

TreeSpark: Learning to Draft the Trees You Verify for Speculative Decoding

Jihao Zhao, Qianyu Jiang, Shuaishuai Zu et al. (+6 more)

Parallel drafters such as DFlash and DSpark reduce proposal latency, but early mismatches limit how much of each block can be accepted. Retaining alternatives introduces parent states and candidate choices that reference-conditioned training does not directly supervise. We introduce TreeSpark, a unified training and inference framework extending the semi-autoregressive DSpark architecture. It constructs parent-conditioned candidate trees from one backbone pass and replays their expansion, pruning, and budgeted packing during training. Survival negative log-likelihood weights retained edges by reference-derived path mass, while a capped coverage hinge promotes target-important alternatives near local selection boundaries. Both objectives reuse reference-chain target distributions without additional target forward passes. We evaluate greedy decoding with Qwen3-4B and Qwen3-8B on nine general benchmarks and six writing domains. Across targets and suites, TreeSpark improves mean acceptance length over DSpark by 23.4% to 31.6% and mean single-request throughput by 15.1% to 25.0%. Batched Qwen3-4B throughput improves by 12.6% to 17.2%. At fixed attention capacity and inference policy, the proposed supervision further improves mean acceptance length by approximately 6.4% and mean system throughput by 4.2% to 5.0% on Qwen3-4B.
Read the full abstract

All authors Jihao Zhao, Qianyu Jiang, Shuaishuai Zu, Pengfei Li, Guanghuan Fang, Ben Feng, Biao Qin, Zhibin Wang, Baochuan Yang

Machine Learning

RM-OPD: On-Policy Distillation for Generative Reward Models

Qinglin Su, Licheng Pan, Zhen Fang et al. (+5 more)

Reinforcement Learning on Mixed-domain preference data (Mix-RL) has become the mainstream approach for training general-purpose Generative Reward Models (GRMs). However, Mix-RL suffers from two critical bottlenecks: the reward sparsity induced by outcome rewards, and the cross-domain interference arising from jointly optimizing heterogeneous capabilities, which together give rise to imbalanced capability acquisition and pervasive reward hacking. Inspired by the success of On-Policy Distillation (OPD) in multi-domain capability integration, we introduce RM-OPD (Reward Modeling via On-Policy Distillation), a two-stage framework that separates specialist acquisition from capability integration. First, we cultivate domain-specialized teacher models via single-domain GRPO fine-tuning, allowing each expert to reach its performance ceiling in isolation; then we consolidate these teachers into one student model through (i) routed on-policy distillation, which provides dense signed token-level supervision on student-generated critiques; (ii) domain contribution reweighting, which reweights token guidance toward target domain shares, preventing specific domains from dominating optimization; and (iii) optional criteria context distillation, which conditions teachers on domain-specific evaluation criteria to sharpen token-level supervision. Across multiple RM benchmarks and two base models, RM-OPD consistently outperforms vanilla Mix-RL and other strong baselines, achieving overall superior performance and exhibiting an emergent "teacher-surpassing" effect. These results establish RM-OPD as a promising paradigm for building generalist GRMs.
Read the full abstract

All authors Qinglin Su, Licheng Pan, Zhen Fang, Jiakang Wang, Qingnan Ren, Yunsheng Lu, Yan Gao, Zhixuan Chu

Cryptography and Security

Secret Sharing from Non-Extractable Randomness

Divesh Aggarwal

Bosley and Dodis (TCC 2007) proved that a class of imperfect randomness sources that suffices for information-theoretic encryption of messages longer than the logarithm of the key length must admit deterministic extraction of nearly uniform bits, and asked whether the same is true of two-out-of-two secret sharing. Aggarwal, Chung, Obremski, and Ribeiro (TCC 2022) gave a positive answer for schemes that are in addition leakage-resilient or non-malleable; the original question remained open. We answer it in the negative. For every $b\ge2$ we construct a class of flat sources on $\{0,1\}^{b+2}$, each of min-entropy at least $b$, together with a perfectly correct and perfectly private two-out-of-two scheme that shares a $b$-bit secret into two $(b+1)$-bit shares using any source in the class, and we show that every deterministic map from $\{0,1\}^{b+2}$ to $\{0,1\}$ is at statistical distance at least $1/6$ from uniform on some source in the class. This implies that sharing a long enough secret does not require extractable randomness. The proof is a short parity argument. The idea of the construction extends to other thresholds: for two-out-of-$n$ and $r$-out-of-$r$ sharing we give perfect schemes with $(b+1)$-bit shares, private for classes of flat sources from which no single bit can be extracted with error below $1/(4n-2)$ and $1/(2(2^r-1))$, respectively.
Read the full abstract

Sound

MultiPartyTurnBench: Benchmarking Floor Control in Multi-Party Spoken Dialogue

Puneet Mathur and Dinesh Manocha

Full-duplex speech systems must decide when to enter, hold, or relinquish the conversational floor, even as multiple speakers overlap or backchannel without taking it. Existing evaluations study dyadic turn-taking or end-to-end agent behavior, but lack a unified, controlled evaluation of genuine floor transitions against confusable non-events in multi-party settings. We introduce MultiPartyTurnBench (MPTB), a benchmark of 240 synthetically generated, channel-separated conversations spanning two to five speakers and six interaction types. MPTB evaluates two complementary tasks: End-of-Turn (EOT) detection distinguishes genuine turn completion from temporary holds, while Interruption (INT) detection distinguishes successful floor takeovers from non-floor-taking speech. Exact event-conditioned annotations enable causal evaluation using false-positive-constrained recall and signed latency. Across acoustic, semantic, and turn-projection baselines, we find a stark asymmetry: learned systems achieve 82.5-92.6% EOT recall at 4.3-11.8% false-positive rates, while INT detection suffers from missed takeovers or excessive false activations. This asymmetry extends to five full-duplex agents which achieve 86.1-95.8% balanced accuracy on turn entry, yet perform near chance (46.9-54.7%) on selective interruption handling. Gemini-3.8-Live yields to 97.7% of backchannels, whereas MiniCPM-o-4.5 yields to only 1.4% of genuine interruptions. These findings establish selective floor control as a distinct, unresolved challenge for reliable multi-party voice interaction.
Read the full abstract

Cryptography and Security

FreeSBee: A Constant-Time Compilation Pipeline

Julian Ganz, David Knothe, David Oswald et al. (+3 more)

Constant-time execution is an important security property, especially for IoT devices, that prevents the leakage of sensitive information over timing side-channels. However, achieving constant-time execution is very hard when going down to the cycle level, at least on realistic devices. A lot of research is concerned with producing tools that either determine whether code is constant-time or that ensure it is compiled into a constant-time binary, with various degrees of accuracy. In this work, we present and evaluate a full workflow for hardening against timing side-channels during compilation. It starts with taint analysis to determine the code locations that depend on sensitive information. These are then the object of automated control-flow transformations inside the compiler to seal possible timing side-channels. We evaluate our workflow in detail on a realistic application running on a RISC-V core using both static timing analysis and statistical methods. Even though we do not reach cycle-accurate constant-time execution, our statistical results suggest that our hardened binary is safe against timing attacks.
Read the full abstract

All authors Julian Ganz, David Knothe, David Oswald, Sebastian Reiter, Thorben Schneider, Simon Wegener

Machine Learning

Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming

Thomas Pammer, Johannes Fürnkranz and Markus Sinnl

We study the problem of learning accurate and cognitively simple binary classifiers that are directly interpretable as Boolean rules. We propose mixed integer linear programming (MILP) formulations for training sparse disjunctive normal form (DNF) networks, viewed as small networks with a conjunctive hidden layer and a disjunctive output layer. Contrary to previous work that uses MILP for inductive rule learning, we start with a fixed network structure, for which we then find optimal weights. A regularized objective explicitly trades off misclassification costs against sparsity, yielding globally optimized rule sets. Our main technical contribution is an enhanced formulation that substantially reduces the number of integer decision variables and eliminates redundant constraints, while preserving exact Boolean semantics for all feasible solutions. Furthermore, we develop a combined approach based on initialization and dropout to break symmetries in the weights between rules, mitigating redundancy, and encouraging more diverse and stable rule representations. Experiments on synthetic benchmarks and UCI datasets show that the advanced formulation improves computational efficiency and achieves competitive or superior accuracy compared to established interpretable baselines, including SAT-based DNF learning and well-known rule- and tree-based learners. Overall, our approach provides a principled optimization framework for learning compact Boolean rule sets with explicit control over accuracy and sparsity trade-offs.
Read the full abstract

Optimization and Control

Continuous-Time q-Learning with Nonsmooth Value Functions: a BSDE Approach

Manuel Asero, Elena Bandini, Andrea Cosso et al. (+1 more)

We study continuous-time q-learning for entropy-regularized diffusion control without assuming classical differentiability of the value function. Rewards and admissible policies need only be Borel measurable, which allows discontinuous rewards. The diffusion coefficient $\sigma$ is action independent and locally uniformly elliptic, and the drift is $b_0 + \sigma\psi$, with $b_0$ action independent and $\psi$ bounded. In this setting the q-function involves the value only through its gradient. For each admissible policy, we represent its value by a backward stochastic differential equation (BSDE) and prove that this value is continuous before the horizon, although the data are only measurable. Replacing the gradient with a measurable representation of the martingale integrand, we define the q-function up to Lebesgue-null sets and recover the classical definition when the value function is smooth. We then prove policy improvement, when the Gibbs update is admissible, and martingale characterizations of the value and the q-function from on-policy and off-policy observations. Here the candidate value need only be continuous before the horizon, and actions are sampled on a deterministic finite grid and held constant between sampling times. Under an integrability condition on the reward, we show that a second BSDE gives an upper bound for the optimal value, attained when its Gibbs policy is admissible, and we characterize its solution in the same way.
Read the full abstract

All authors Manuel Asero, Elena Bandini, Andrea Cosso, Laura Perelli

Atomic Physics

Fine-structure Raman interferometry with metastable barium for magnetically resilient inertial sensing

Masaki Nagamori, Hiromi Hatano, Toshiyuki Hosoya et al. (+3 more)

Atom-interferometric quantum inertial sensors are promising candidates for inertial navigation systems owing to their high sensitivity and intrinsically low drift. However, their application to next-generation inertial navigation, targeting a drift below 100 m/month, requires exceptionally stringent inertial-sensor bias stability. In conventional Raman atom interferometers based on ground-state hyperfine transitions, suppressing inertial-sensor biases arising from quadratic Zeeman shifts requires highly optimized magnetic shielding. In this work, we propose a Raman atom interferometer based on fine-structure transitions between metastable states of barium, which exhibit a quadratic Zeeman sensitivity approximately three orders of magnitude lower than that of conventional alkali-atom Raman transitions. We show that the proposed scheme increases the allowable spatial gradient of the magnetic-field magnitude by a factor of several tens, thereby relaxing the magnetic-field-control requirements from highly optimized shielding and compensation systems to standard magnetic-field-control techniques. Furthermore, we experimentally demonstrate Mach–Zehnder interference between metastable fine-structure states of barium, validating a key building block of the proposed inertial-sensing scheme.
Read the full abstract

All authors Masaki Nagamori, Hiromi Hatano, Toshiyuki Hosoya, Yuki Miyazawa, Tomoya Sato, Mikio Kozuma

High Energy Physics (Phenomenology)

Correlation between gravitational wave frequencies as a signature of new black hole physics

Ngo Phuc Duc Loc

We point out that for black holes (BH) of the same mass, despite of arising from two entirely different physical arguments, the peak frequency of the coarse-grained gravitational wave (GW) spectrum of gravitons $f_\text{peak}^\text{graviton}$ is very close to the peak frequency of the GW spectrum produced from binary BH mergers $f_\text{peak}^\text{merger}$. In particular, we find that $f_\text{peak}^\text{graviton}/f_\text{peak}^\text{merger}\approx 2.3$ independent from BH mass or population. If BHs are still evaporating, the peak frequencies scale as $\propto M_i^{-1}$ with $M_i$ is the initial BH mass. If BHs evaporated in the radiation (matter) dominated era, the peak frequencies scale as $\propto M_i^{1/2}$ ($\propto M_i$) respectively after taking into account redshift effect. Any deviation from the aforementioned magical ratio would indicate new BH physics. The sweetest frequency ballpark to look for this correlation is around $10^{14}\ \rm Hz-10^{20}\ \rm Hz$ where both GW spectra are allowed to have maximum amplitude. As a demonstration of what the new physics could be, we consider a phenomenological self-similar regular BH model of Hayward featuring a single hair parameter. The observed ratio of peak frequencies can be used to directly infer the hair parameter, along with crucial properties of the novel BH solution such as Hawking temperature, horizon size, and innermost stable circular orbit. We also comment on the pros and cons of correlating frequencies among other GW sources such as formation, population fluctuation, and sudden transition of adiabatic perturbation.
Read the full abstract

Mathematical Physics

Quantized six-vertex realizations of frustration graphs

Atsuo Kuniba and Souta Oikawa

A recent graph-theoretic criterion for a Hamiltonian to exhibit a free-parafermion spectrum is that its frustration graph be an indifference graph. We show that an arbitrary finite indifference graph can be realized as a frustration graph of the quantized six-vertex model at odd roots of unity, by a suitable choice of the defining wiring diagram and parameters. This provides a concrete realization of the criterion through the quantum Frobenius property of the quantized six-vertex model, recently established from its three-dimensional integrable structure.
Read the full abstract

Number Theory

Integral Specializations of Polynomials: Hilbert Irreducibility, Coprimality, and Density

Angelot Behajaina, Pierre Dèbes and Martin Widmer

The Hilbert specialization property of a field, whereby irreducibility of polynomials is preserved under specialization of some of the variables, admits an integral counterpart where specialization takes place inside a ring and irreducibility is required over the ring. It is known that the ring $\mathcal{O}_K$ of integers of any global field $K$ satisfies this integrally Hilbertian property. Using a different approach, we reprove this result and strengthen it by showing that the corresponding integral Hilbert subsets of $\mathcal{O}_K^k$ are not only Zariski-dense, as required by integral hilbertianity, but also of positive density. A central ingredient is a generalization to the ring $\mathcal{O}_K$ of the Ekedahl–Poonen formula originally proved for $\mathbb{Z}$ and $\mathbb{F}_q[t]$. This formula, of independent interest, gives the density of tuples in $\mathcal{O}_K^k$ for which coprime polynomials in $k$ variables and with coefficients in ${\mathcal O}_K$ take coprime values. Our generalization requires estimates for the number of points of bounded height in an arithmetic progression. We obtain such estimates as our third main result, thereby contributing to the theory of heights.
Read the full abstract

High Energy Physics (Phenomenology)

The dead-cone effect in charm, bottom and top quark jets in angular and momentum space

Redamy Perez-Ramos, Stefan Kluth and Wolfgang Ochs

The dead-cone effect reflects the mass dependence of QCD radiation from heavy quarks. We investigate this phenomenon from charm and bottom to top quarks using two-dimensional angular distributions and single-particle inclusive momentum spectra. The angular distributions illustrate the depletion of radiation around the heavy-quark direction, while the momentum spectra provide a quantitative observable of the effect. For charm and bottom jets, fragmentation functions reconstructed from LEP data at the $Z$ pole exhibit a strong suppression of high-momentum particles, quantitatively described within the Modified Leading Logarithmic Approximation of perturbative QCD. The analysis is extended to $\sqrt{s}=1$ TeV using Monte Carlo simulations. A hypothetical stable-top sample isolates the primary top-quark radiation pattern, whereas the physical decay $t\to bW$ introduces coherent radiation from the $tb$ colour dipole, obscuring the primary dead cone. By exploiting the dependence on the $b$-quark emission angle $\Theta_b$, we reconstruct the primary top-quark parton momentum spectrum through extrapolation to $\Theta_b\to0$. In contrast, the small-angle gluon and parton distributions do not converge to the stable-top dead-cone pattern because of radiation from the decay $b$ quark. These findings establish the inclusive momentum spectrum as a robust quantitative observable of the top-quark dead-cone effect and provide a unified description of mass-dependent QCD radiation from charm to top.
Read the full abstract

High Energy Astrophysical Phenomena

Search for High-Energy Neutrinos from Infrared Flares

R. Abbasi, M. Ackermann, J. Adams et al. (+415 more)

IceCube has detected a diffuse flux of high-energy astrophysical neutrinos. While its astrophysical origin is unknown, a significant neutrino emission has been observed from the gamma-ray blazar TXS 0506+056 and the Seyfert galaxy NGC 1068. This suggests accretion by supermassive black holes as one of the candidate mechanisms of the diffuse neutrino flux. Indeed, external studies that made use of IceCube data identified candidate sources, including AT2019dsg, likely a stellar tidal disruption event (TDE), and AT2019fdr, associated with an active galactic nucleus (AGN). An additional study using an optically detected sample of 63 accretion flares revealed another candidate as a potential high-energy neutrino counterpart: AT2019aalc. All of these candidate transients exhibit strong dust echoes, characterized by delayed infrared (IR) emission relative to the initial optical outburst. We look for correlations between the most comprehensive catalog of infrared flares and 10 years of IceCube archival data. We perform a time-dependent stacking analysis testing 393 sources in the Northern Sky against 10 years of through-going tracks. No significant neutrino emission is observed, and we derive upper limits to constrain neutrino emission from this catalog. These limits are subsequently extrapolated to a population-level study. We find that the contribution of IR flares to the IceCube diffuse neutrino flux at 100 TeV is below 2.7% for a spectral index of 2.5.
Read the full abstract

All authors R. Abbasi, M. Ackermann, J. Adams, J. A. Aguilar, M. Ahlers, J.M. Alameddine, S. Ali, N. M. Amin, K. Andeen, C. Argüelles, S. Athanasiadou, S. N. Axani, R. Babu, X. Bai, A. Balagopal V., S. W. Barwick, V. Basu, R. Bay, J. J. Beatty, J. Becker Tjus, J. Beise, C. Bellenghi, S. Benkel, S. BenZvi, D. Berley, E. Bernardini, D. Z. Besson, E. Blaufuss, L. Bloom, S. Blot, F. Bontempo, J. Y. Book Motzkin, C. Boscolo Meneguolo, S. Böser, O. Botner, J. Braun, B. Brinson, Z. Brisson-Tsavoussis, L. Brusa, R. T. Burley, D. Butterfield, K. Carloni, J. Carpio, N. Chau, Y. C. Chen, Z. Chen, D. Chirkin, S. Choi, A. Chubarov, B. A. Clark, D. A. Coloma Borja, A. Connolly, J. M. Conrad, D. F. Cowen, C. De Clercq, J. J. DeLaunay, D. Delgado, T. Delmeulle, S. Deng, P. Desiati, K. D. de Vries, G. de Wasseige, T. DeYoung, J. C. Díaz-Vélez, S. DiKerby, T. Ding, M. Dittmer, A. Domi, L. Draper, L. Dueser, D. Durnford, K. Dutta, M. A. DuVernois, T. Ehrhardt, L. Eidenschink, A. Eimer, C. Eldridge, P. Eller, E. Ellinger, D. Elsässer, R. Engel, H. Erpenbeck, W. Esmail, S. Eulig, J. Evans, P. A. Evenson, K. L. Fan, K. Fang, K. Farrag, A. Fattorini, A. R. Fazely, A. Fedynitch, N. Feigl, C. Finley, D. Fox, A. Franckowiak, S. Fukami, P. Fürst, J. Gallagher, A. Garcia, M. Garcia, E. Genton, L. Gerhardt, A. Ghadimi, C. Glaser, T. Glüsenkamp, J. G. Gonzalez, S. Goswami, A. Granados, D. Grant, S. J. Gray, S. Griffin, S. Griswold, K. M. Groth, D. Guevel, C. Günther, P. Gutjahr, C. Ha, A. Hallgren, F. Halzen, M. Handt, K. Hanson, J. Hardin, A. A. Harnisch, P. Hatch, A. Haungs, J. Häußler, K. Helbing, J. Hellrung, B. Henke, L. Hennig, F. Henningsen, L. Heuermann, R. Hewett, N. Heyer, S. Hickford, A. Hidvegi, C. Hill, G. C. Hill, R. Hmaid, K. D. Hoffman, A. Hollnagel, D. Hooper, S. Hori, K. Hoshina, M. Hostert, W. Hou, M. Hrywniak, T. Huber, K. Hultqvist, K. Hymon, A. Ishihara, W. Iwakiri, M. Jacquart, S. Jain, O. Janik, M. Jansson, M. Jin, N. Kamp, D. Kang, W. Kang, A. Kappes, L. Kardum, T. Karg, A. Karle, A. Katil, M. Kauer, J. L. Kelley, B. Kern, M. Khanal, A. Khatee Zathul, A. Kheirandish, T. Kim, H. Kimku, F. Kirchner, J. Kiryluk, C. Klein, S. R. Klein, Y. Kobayashi, S. Koch, A. Kochocki, R. Koirala, H. Kolanoski, T. Kontrimas, L. Köpke, C. Kopper, D. J. Koskinen, P. Koundal, M. Kowalski, A. Kravka, N. Krieger, T. Krishnan, K. Kruiswijk, E. Krupczak, E. Kun, N. Kurahashi, C. Lagunas Gualda, L. Lallement Arnaud, M. J. Larson, F. Lauber, J. P. Lazar, K. Leonard DeHolton, A. Leszczyńska, C. Li, J. Liao, C. Lin, Q. R. Liu, Y. T. Liu, M. Liubarska, C. Love, L. Lu, F. Lucarelli, W. Luszczak, Y. Lyu, M. Macdonald, F. Machingal, E. Magnus, Y. Makino, E. Manao, S. Mancina, A. Mand, I. C. Mariş, S. Marka, Z. Marka, I. Martinez-Soler, R. Maruyama, J. Mauro, F. Mayhew, F. McNally, K. Meagher, A. Medina, M. Meier, Y. Merckx, L. Merten, J. M. Mitchell, S. Mondal, T. Montaruli, R. W. Moore, Y. Morii, A. Mosbrugger, D. Mousadi, E. Moyaux, T. Mukherjee, M. Nakos, U. Naumann, R. Neshat, L. Neste, M. Neumann, M. U. Nisa, K. Noda, A. Noell, A. Novikov, A. Obertacke, V. O'Dell, A. Olivas, R. Orsoe, J. Osborn, E. O'Sullivan, B. Owens, V. Palusova, H. Pandya, A. Parenti, C. Parisel, N. Park, V. Parrish, E. N. Paudel, L. Paul, C. Pérez de los Heros, T. Pernice, T. C. Petersen, J. Peterson, S. Pick, M. Plum, A. Pontén, V. Poojyam, B. Pries, R. Procter-Murphy, G. T. Przybylski, L. Pyras, C. Raab, J. Rack-Helleis, N. Rad, M. Ravn, K. Rawlins, Z. Rechav, A. Rehman, R. Reimann, I. Reistroffer, E. Resconi, C. D. Rho, W. Rhode, L. Ricca, B. Riedel, A. Rifaie, E. J. Roberts, S. Rodan, M. Rongen, A. Rosted, C. Rott, T. Ruhe, L. Ruohan, D. Ryckbosch, J. Saffer, D. Salazar-Gallegos, P. Sampathkumar, A. Sandrock, G. Sanger-Johnson, M. Santander, S. Sarkar, M. Scarnera, M. Schaufel, H. Schieler, S. Schindler, L. Schlickmann, B. Schlüter, F. Schlüter, N. Schmeisser, T. Schmidt, F. Schmitt, A. Scholz, F. G. Schröder, S. Schwirn, S. Sclafani, D. Seckel, L. Seen, M. Seikh, S. Seunarine, P. A. Sevle Myhr, R. Shah, S. Shah, A. Sharma, S. Shefali, N. Shimizu, M. Shin, R. Singh, B. Skrzypek, R. Snihur, J. Soedingrekso, D. Soldin, P. Soldin, G. Sommani, D. Song, C. Spannfellner, G. M. Spiczak, C. Spiering, J. Stachurska, M. Stamatikos, T. Stanev, T. Stezelberger, T. Stürwald, T. Stuttard, G. W. Sullivan, I. Taboada, S. Ter-Antonyan, A. Terliuk, A. Thakuri, P. Thelen, M. Thiesmeyer, W. G. Thompson, J. Thwaites, W. Tian, S. Tilav, K. Tollefson, J. A. Torres, S. Toscano, D. Tosi, K. Upshaw, A. Vaidyanathan, N. Valtonen-Mattila, J. Valverde, J. Vandenbroucke, T. Van Eeden, N. van Eijndhoven, L. Van Rootselaar, J. van Santen, J. Vara, F. Varsi, M. Velazquez, M. Venugopal, M. Vereecken, S. Vergara Carrasco, S. Verpoest, A. Vijai, J. Villarreal, C. Walck, A. Wang, E. H. S. Warrick, C. Weaver, P. Weigel, A. Weindl, J. Weldert, A. Y. Wen, C. Wendt, J. Werthebach, M. Weyrauch, N. Whitehorn, C. H. Wiebusch, D. R. Williams, L. Witthaus, J. Woodward, G. Wrede, X. W. Xu, J. P. Yanez, Y. Yao, E. Yildizci, S. Yoshida, F. Yu, S. Yu, T. Yuan, S. Yun-Cárcamo, A. Zander Jurowitzki, A. Zegarelli, S. Zhang, Z. Zhang, P. Zhelnin, P. Zilberman, C. Zilleruelo Cañas, M. Zimmermann

Computation and Language

Gap Detection: Mitigating Homogenization through Demand and Defensibility

Xucheng Yu, Haibo Jin, Xiaoqi Wang et al. (+2 more)

Language models often produce semantically similar answers to open-ended questions, even when many plausible alternatives exist. Generating useful alternatives requires identifying both what the query calls for and what existing responses already cover. We introduce Gap Detection, a framework that searches for relevant but unoccupied directions in a shared text embedding space. Its objective, GapScore, combines query relevance (Demand) with dissimilarity to the nearest existing item (Defensibility). The framework optimizes this product on the unit sphere, maps candidate directions to concept words with a language model, and uses task-specific filtering and constrained prompting to turn those concepts into text. On five curated topics and 90 INFINITY-CHAT queries, the reported mean relative reductions in output Collision are 12.8% and 10.8%, respectively. We also instantiate the framework for generative engine optimization over four product markets and for iterative synthetic-data augmentation. In the latter setting, with 100 training examples per method, Gap Detection reduces Collision from 0.324 to 0.298 relative to undirected accumulation and increases the Vendi Score from 27.16 to 29.20; the accuracy difference is one example on a 40-example test set. Component ablations expose the trade-off between semantic separation and relevance, while also showing that better direction scores do not uniformly yield better downstream metrics. These results support explicit gap search as a practical mechanism for diversifying content, with task validity checked separately from geometric novelty.
Read the full abstract

All authors Xucheng Yu, Haibo Jin, Xiaoqi Wang, Runze Kong, Haohan Wang