Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Friday 11 September

High Energy Physics (Phenomenology)

Spin-independent scattering of pseudoscalar-mediated dark matter

Dark matter with pseudoscalar couplings provides a well-motivated scenario in which direct-detection signals are suppressed at tree level, since the scattering off nuclei is both spin-dependent and momentum suppressed. While spin-independent scattering is absent at tree level, it arises at one loop and can provide the leading direct-detection signal. We revisit this scenario in a general sub-electroweak effective field theory with a light pseudoscalar mediator, including interactions through to mass-dimension-six. We compute the matching onto the quark and gluon operators relevant for direct detection to determine whether this scenario could be detectable at future experiments, while also requiring consistency with the observed dark matter relic-abundance and indirect-detection limits. We find that while models with a pseudoscalar mediator can generate spin-independent cross sections above the neutrino floor, this generally requires additional new physics below the TeV scale.
Read the full abstract

All authors Nicole F. Bell, Giorgio Busoni, Peter Cox, Laura W. Fang, John Gargalionis, Jayden L. Newstead, Ewan N. V. Wallace, Martin J. White, Anthony G. Williams

Systems and Control

Controlled Islanding and Reconnection with Stability Guarantees via Submodular Optimization

In large-scale networked dynamical systems, local faults or disturbances may propagate through the interconnections and compromise the stability of the entire network. Partitioning the network into disjoint subsystems, or islands, limits this propagation by isolating unstable portions of the network from those that remain stable. This paper studies a two-stage controlled islanding and reconnection problem. In the first stage, the network is partitioned into stable and unstable islands. In the second stage, selected stable islands are reconnected to recover connectivity lost during islanding while satisfying stability requirements. We derive Lyapunov-based sufficient conditions for both stages. To address the coupling between the islanding and reconnection decisions, we construct an augmented graph on which the bases of a graphic matroid induce exactly the admissible islanding and reconnection pairs. We further show that the resulting optimization problem admits an equivalent nonincreasing supermodular formulation. Based on this structure, we develop a local-search algorithm with counterexample refinement and establish a performance guarantee for its inner loop with fixed sample sets. Numerical studies on linear networked systems illustrate the proposed procedure, examine its sensitivity to fault locations, and compare the proposed method with a mixed-integer linear programming benchmark in terms of solution quality and computational performance.
Read the full abstract

High Energy Physics (Phenomenology)

Tensor-polarized twist-3 distribution function $f_{LT}$ of spin-1 deuteron

We investigate the twist-2 and twist-3 tensor-polarized partonic structures of the spin-1 deuteron. Using the operator product expansion with local operators, we derive a Wandzura-Wilczek (WW)-like twist-2 relation between the tensor-polarized twist-2 quark distribution $f_{1LL}$ and the twist-3 distribution $f_{LT}$, together with a Burkhardt-Cottingham (BC)-like sum rule. The local-operator formalism makes the Lorentz structure and rotational invariance manifest and provides useful constraints on the twist-3 sector. We then use a phenomenological parametrization of the twist-2 distribution $f_{1LL}(x)$, constrained by HERMES data on the deuteron structure function $b_1$ at $Q^2=2.5~\text{GeV}^2$, to quantitatively estimate the twist-3 distribution $f_{LT}(x)$ within the WW approximation. The resulting $f_{LT}(x)$ has a shape and magnitude comparable to those of $f_{1LL}(x)$, indicating that subleading-twist effects may be relevant for future experiments at relatively low $Q^2$. In this contribution, we summarize the theoretical derivation of the WW-like relation and BC-like sum rule and present a phenomenological estimate of the tensor-polarized twist-3 quark distribution $f_{LT}$ in the deuteron.
Read the full abstract

Astrophysics of Galaxies

Mapping the Quasar Main Sequence in the UV range: A Connection with the UV Fe III Emission

The quasar main sequence (MS) provides a consistent unification scheme for type I AGNs. This framework, constructed from the relative flux between the optical emissions $\mathrm{H}\beta$ and $\mathrm{Fe\,II\,(\lambda4570)}$, and the $\mathrm{FWHM(H\beta)}$, organizes quasars into two populations (B, A) and the subpopulation xA, which typically consists of highly accreting sources. We explored the use of the UV 1900 blend emissions ($\mathrm{Al\,III}\,\mathrm{\lambda}1860$, $\mathrm{Si\,III]}\,\mathrm{\lambda}1892$, $\mathrm{C\,III]}\,\mathrm{\lambda}1909$) and the $\mathrm{Fe\,III}$ and $\mathrm{Fe\,II}$ transitions between 1700 and 2200 Angstrom as tracers of this main sequence. We fitted these emissions for a sample of 69 low-$z$ ($z\lesssim0.8$) quasars observed with the Hubble Space Telescope (HST) that had available optical measurements, enabling a direct comparison with the optical quasar main sequence. We found a UV plane defined by $\mathrm{FWHM}(\mathrm{Al\,III})$ and the equivalent width of the combined emissions of $\mathrm{Fe\,III}$ and $\mathrm{Fe\,II}$. The plane defined by these parameters mirrors the quasar main sequence in the optical regime and shows a clear distinction between the different populations. The plane also enables the identification of xA sources with high completeness but a non-negligible rate of false positives. Our results suggest that the UV iron emission around the 1900 blend could be used to trace the quasar main sequence trend, offering an alternative way to study this scheme at high redshift.
Read the full abstract

All authors D. Martínez Collipal, M. L. Martínez-Aldama, A. Pandey, M. Cardenas, P. Marziani, S. Panda

Optics

Resonance-Enhanced Time Reflection at Photonic Temporal Interfaces

Photonic temporal interfaces enable dynamic control of light fields, yet strong time reflection at optical frequencies remains challenging because conventional approaches demand large material refractive-index changes on ultrafast timescales. Here, we introduce a resonance-assisted mechanism that harnesses both polarization energy accumulated in a dispersive medium prior to the temporal interface and strongly increased energy supplied by the modulation system. We show that, under a specific critical condition, rapidly increasing the material resonance frequency yields orders-of-magnitude stronger time-reflected power flux density than what conventional plasma-frequency modulation provides. To implement this mechanism in optical systems, we identify two routes based on dielectric and plasmonic structural resonances. For the plasmonic route, we develop an analytical effective-medium model of conducting-oxide cylinder arrays, in which localized surface-plasmon resonances transform the constituent Drude response into a geometrically tunable effective Lorentz response. Using cadmium oxide as a representative material, we predict an enhancement exceeding three orders of magnitude in the summed reflected-mode power coefficient relative to the same homogeneous material under the same modest plasma-frequency modulation, even in the presence of realistic losses. These findings establish spatial resonance engineering as an effective route to strong temporal scattering with reduced demands on intrinsic material tunability.
Read the full abstract

All authors Zeyuan Li, Hammam Bahurmuz, Mohamed H. Mostafa, Mohammad S. Mirmoosa, Puneet Garg, Carsten Rockstuhl, Xuchen Wang, Viktar Asadchy

Machine Learning

Simulating Disengaged Students to Evaluate LLM-based Tutors

Simulated students generated by computational models provide a practical way to evaluate tutoring strategies and pedagogical approaches used by human and AI tutors. However, such simulations should account for disengaged behaviors, including gaming the system, wheel-spinning, and off-task behavior, because tutors may need different responses for different learner states. We present Disengagement-Aware Student Simulators (DAS2), a reproducible pre-deployment protocol that models five learner-engagement states: engaged, gaming, wheel-spinning, off-task, and mixed, and evaluates AI tutor performance across these states. Using ASSISTments09, two coders independently labeled 100 sampled tutoring sessions based on anonymized interaction-log summaries. They achieved 84% agreement (Cohen's kappa = 0.78), and among agreed cases, human consensus labels matched DAS2 rule-based labels in 81% of cases (kappa = 0.75). Conditioning simulations on intended learner states reduced the correctness-rate gap between simulated and authentic sessions from 0.54 to 0.20 for gaming and from 0.51 to 0.18 for wheel-spinning. Fine-tuned Qwen2.5-7B better matched authentic response-time distributions, while prompt-only GPT-4o generated more distinguishable learner states. Evaluation of five AI tutors from the Claude, Llama, Gemini, Qwen, and GPT families showed that relative rankings remained stable across learner states and interaction lengths, while absolute performance varied, revealing state-specific differences in tutor support. Human validation further showed that automated tutor evaluation does not fully align with human judgment. DAS2 provides a pre-deployment framework for evaluating how AI tutors respond to diverse learner-engagement states before deployment.
Read the full abstract

Distributed Computing

Unleashing the Power of Equality Saturation for Tensor Program Superoptimization

Efficient GPU implementations of tensor programs often require joint optimization of high-level algebraic formulations and low-level execution strategies. However, the resulting search space grows rapidly as transformations combine across operators, making joint optimization difficult to scale. We present EqiForge, a tensor program superoptimizer based on equality saturation. Its unified IR represents high-level tensor expressions and tiled computations in a single expression language. By composing equality rules, EqiForge derives fused implementations such as FlashAttention-style kernels directly from tensor expressions. Early compaction prunes redundant partial programs before completion, while subgraph composition extends the search to larger graphs. Across tensor-program benchmarks, EqiForge achieves a geometric mean speedup of 1.32x and a maximum of 2.74x over the fastest available baseline per configuration. Its attention kernels outperform FlashAttention by up to 1.87x in decode and approach its performance in prefill. EqiForge also discovers new implementations that outperform torch.compile on various Transformer layers, including QK-normalized MLA (3.16x) and mHC (5.84x).
Read the full abstract

All authors Qi Zhan, Xing Hu, Xin Xia, Shanping Li

Materials Science

Irreducible sub-nm$^2$ ferroelectric domains by 2D-flat polar band in perovskite superlattices

Flat polar phonon bands in ferroelectrics have been envisioned to realize irreducibly small domains for achieving the highest memory density. However, such bands are extremely rare, with only a one-dimensional flat polar band discovered in ferroelectric hafnia, which gives rise to irreducibly narrow, line-type domains with a half-unit-cell width of 2.7 Angstrom. Here, we report the discovery of a two-dimensional flat polar band in the ferroelectric (BaTiO$_3$)$_1$/(BaXO$_3$)$_1$ superlattice (X = Zr, Sn), which generates extremely localized, noninteracting dipoles within its quarter-unit-cell area that form quasi-degenerate states of irreducible sub-nm$^2$ ferroelectric domains. The estimated ferroelectric domain density, exceeding 300 Tbit/cm$^2$ (assuming one bit per domain) represents a record high among the known ferroelectrics, with independently switchable dipoles at low voltages. The ferroelectric phase hosting two-dimensional flat bands constitutes a competing ground state across superlattice stacking directions and thus can be experimentally realized in both freestanding structures and epitaxial growth on MgO substrates. This discovery opens unprecedented opportunities to explore multidimensional flat polar bands in ferroelectrics and to engineer ultra-dense and low-power memory devices.
Read the full abstract

Systems and Control

The Iterative Equivariant Filter

This paper presents the iterative equivariant filter (IterEqF). The iterative extended Kalman filter (IterEKF) replaces the standard EKF correction step with an iterative correction step that is the Gauss-Newton solution to a nonlinear weighted least squares problem. The standard equivariant filter (EqF) exploits and respects the underlying symmetry of state estimation problems posed on homogeneous spaces and Lie groups. The motivation behind the IterEqF is to combine the features of both the IterEKF and the EqF, thus leading to a high-performance state estimation solution that is well suited to navigation problems. This paper derives the iteration procedure for the IterEqF update step and shows how the intrinsic nonlinearity of the approach naturally results in a `reset' of the filter's covariance into new coordinates. Monte-Carlo simulations of range-based localisation for a mobile robot demonstrate the improvement in performance relative to a standard EqF, especially during the transient convergence.
Read the full abstract

Artificial Intelligence

LoRA-RC: Reservoir Computing with Low-Rank Adaptation

Reservoir computing (RC) trains only a linear readout over a fixed recurrent layer, making it fast and data-efficient for online prediction. However, a static reservoir degrades under system drift, readout-only adaptation is then insufficient, and unconstrained reservoir adaptation can destroy the echo-state and incremental stability properties that make RC reliable. This paper proposes LoRA-RC, which adapts the recurrent matrix through a low-rank correction driven by streaming prediction errors. The base reservoir and adaptation bases are fixed offline; a small core matrix is adapted online, projected onto a spectral-norm ball, and low-pass filtered at each step. The projection guarantees that every applied recurrent matrix remains within a certified contraction set, and an incremental input-to-state stability bound is established for the reservoir along each online adaptation path, with path-independent rate and gain. On a Lorenz system with an abrupt parameter drift, LoRA-RC cuts post-drift prediction error by 56% versus a fixed RC and 51% versus readout-only adaptation; ablations over 20 seeds show that removing the projection inflates this error by more than a factor of 40.
Read the full abstract

Group Theory

On a question of Navarro on the field of values of characters of solvable groups

In a 2023 survey paper, Gabriel Navarro posed the following problem: Given an irreducible character $\chi$ of some solvable group $G$ where $\chi$ either is 2-rational or has odd degree, does there exist some $g \in G$ such that $\mathbb{Q}(\chi(g)) = \mathbb{Q}(\chi)$? The answer was recently shown to be no in general by Ulrich Thiel. However, when $\chi$ is further assumed to be factorizable as a product of $p$-special characters, we are able to show that the answer is yes.
Read the full abstract

Combinatorics

Almost Empty Monochromatic Triangles With Many Colors

Given integers $c\geq 2$ and $s\geq 0$, let $\mathsf{M}_3(c,s)$ denote the least integer such that every set of at least $\mathsf{M}_3(c,s)$ points in the plane, no three on a line, colored with $c$ colors, contains a monochromatic triangle with at most $s$ interior points. Further, let $\lambda_3(c)$ be the least integer such that $\mathsf{M}_3(c,\lambda_3(c))<\infty$. \citet{colorempty} proved that, for every $c\geq 2$, $$\left\lfloor\frac{c-1}{2}\right\rfloor \leq \lambda_3(c)\leq c-2.$$ Later, \citet{cravioto2019almost} improved the upper bound to $c-3$, for $c\geq 4$. In this paper, we refine their argument to obtain the following asymptotic improvement: $$\lambda_3(c) \leq c-\sqrt{c\log c}+o (\sqrt{c\log c} ),$$ for all sufficiently large $c$. We also show that every $c$-coloring of a sufficiently large Horton set contains a monochromatic triangle with at most $\lfloor \frac{c-1}{2} \rfloor$ interior points. This shows that the aforementioned lower bound on $\lambda_3(c)$ is sharp within the class of Horton sets. We conclude with a conjecture on the large-color asymptotics of $\lambda_3(c)$.
Read the full abstract

All authors Bhaswar B. Bhattacharya, Sandip Das, Sk Samim Islam, Aashirwad Mohapatra, Saumya Sen

Materials Science

Deterministic patterning and alignment of tellurium quantum wires using nanoscale templates

Tellurium (Te) is an intriguing one-dimensional (1D) semiconductor that has recently attracted considerable interest as a p-type channel material. However, scalable synthesis methods have lacked control over the orientation and patterning of the Te atomic chains, thus limiting its practical use. Guided by theory, we overcome this challenge using nanowire-shaped templates to achieve oriented, single-crystal growth of Te on amorphous substrates. Strong alignment of Te atomic chains is achieved as template widths are reduced to sub-20 nm. This high structural order, confirmed by 4D scanning transmission electron microscopy, enables the observation of pristine quantum transport phenomena for deterministically patterned Te. Field-effect transistors exhibit well-defined conductance plateaus at 77 K due to population of individual 1D subbands. Furthermore, Coulomb blockade emerges at 1.7 K, with the Te channel acting as a gate-tunable quantum dot. This synthesis approach provides a scalable pathway for integration of Te-based quantum materials for future electronic and quantum technologies.
Read the full abstract

All authors I K M Reaz Rahman, Scott Dhuey, Moniruzzaman Jamal, Taehoon Kim, Naoki Higashitarumizu, Selven Virasawmy, Aidar Kemelbay, Genki Ohkatsu, Aditya Nirmale, Inha Kim, Hyong Min Kim, Karen C. Bustillo, Joel W. Ager III, Daryl C. Chrzan, Mary Scott, Yutaka Majima, Ali Javey

High Energy Physics (Phenomenology)

Simple Modular $S_3$ Models for Lepton Masses and Mixing

We perform a systematic study of an economical class of modular $S_3$ lepton-flavor models without enlarging the continuous gauge symmetry or introducing right-handed neutrinos or additional flavon fields. We provide a complete classification of all inequivalent realizations allowed by the $S_3$ singlet–doublet representation structure and the admissible modular-weight assignments. Their phenomenological viability is investigated for both normal ordering (NO) and inverted ordering (IO) using Bayesian model comparison and parameter-correlation analysis based on current neutrino-oscillation data. We identify 18 viable models for NO and 19 for IO, with 14 satisfying the experimental constraints in both orderings. The viable models successfully accommodate current neutrino-oscillation data while yielding nontrivial predictions for leptonic CP violation, Majorana phases, and observables probing the absolute neutrino-mass scale. Among these predictions, the absolute-mass observables provide the clearest separation between the two orderings. At the best-fit points, the predicted ranges of the effective electron-neutrino mass ($m_\beta$) are completely disjoint between NO and IO, whereas those of the sum of neutrino masses ($\sum_i m_i$) and the effective Majorana mass ($m_{\beta\beta}$) show only partial separation. The IO models generally predict a higher absolute neutrino-mass scale and are consequently more strongly constrained by cosmological observations and more accessible to neutrinoless double-beta-decay searches. Our results show that, despite its economical field content, the modular $S_3$ framework accommodates a diverse set of phenomenologically viable lepton-flavor realizations with experimentally testable predictions.
Read the full abstract

Artificial Intelligence

Affective Agent: On-Device Personalized Intervention Reasoning for Wearable Systems

Affective computing has advanced wearable state inference, but on-device reasoning about whether, when, and how to intervene remains challenging. We present Affective Agent, a three-layer reference architecture for personalized intervention reasoning under uncertainty on wearable-class hardware. It combines a compact sub-billion-parameter language model with physiological evidence, context, and user history to decide whether, when, and how to intervene, without cloud dependency or per-user retraining. The architecture is organized into three interacting layers (perception, personalization, and reasoning), adapting to individual users through host-managed structured memory evolution rather than per-user weight updates. We instantiate Affective Agent in indoor environmental quality control and evaluate it on held-out, simulator-generated longitudinal scenarios spanning physiological variation, context, signal quality, and intervention history. Results show that memory-driven personalization and two-pass structured reasoning improve intervention decisions within this synthetic evaluation. By moving the decision layer on-device, this work demonstrates a path from wearable state inference toward closed-loop, personalized intervention on wearable-class hardware.
Read the full abstract

Probability

Expected Infimum and persistence probabilities of Log-Normal Stationary Brown-Resnick Processes

We investigate the asymptotics of the expected infimum of log-normal Brown-Resnick stationary processes, a class of processes that arise naturally in the study of extremes of Gaussian processes and max-stable processes. Specifically, we analyse the functional $$\mathcal{G}_V(T) = \mathbb{E}\left\{\inf_{t \in [0,T]}e^{ \sqrt{2}V(t)-\sigma^2_V(t)} \right\}, \quad T>0, $$ where $V$ is a centered Gaussian process with stationary increments, continuous sample paths and variance $\sigma_V^2$, and a closely related problem of the decay rate of the persistence probability $$p_V(T,C)=\mathbb{P}\left\{\inf_{t\in [0,T]} (\sqrt{2} V(t)- \sigma^2_V(t)) > C\right\}$$ for some constant $C<0$. For both $\mathcal{G}_V(T)$ and $p_V(T,C)$ we derive exact asymptotics as $T \to \infty$ for a broad class of processes $V$, including fractional Brownian motion with Hurst parameter $H \in (1/2,1]$. For the latter, the behavior in the short-range dependence regime $H \in (0, 1/2]$ is markedly different and, in general, more delicate; we find logarithmic asymptotics for a family of processes that includes this case. Our results provide sharp bounds and comparison principles, and highlight the contrast between the behavior of infimum and supremum functionals for such processes. The discrete-time analogues and connections to Pickands constants are also discussed.
Read the full abstract

Artificial Intelligence

AIM: A Privacy-Aware Interoperable Memory Framework for Multi-Agent Multi-User LLM Systems

Traditional large language models (LLMs) are scoped to individual user sessions, limiting their knowledge to a single conversation and preventing them from learning user preferences that evolve over time. Existing agentic memory systems address this limitation but generally operate at the individual-user level, restricting the public knowledge that could be shared across users to improve downstream responses. We introduce AIM (Agentic Interoperable Memory), a unified, privacy-aware memory framework that enables multi-agent, multi-user LLM systems to persistently manage private and shared memory. AIM dynamically classifies information as private, scoped to one user and inaccessible to others, or public, accessible to all users. It enforces index-level access controls so that private memories are retrievable only by their owner, protecting sensitive data while allowing beneficial shared knowledge to improve coordination and consistency. We also introduce MUMBench (Multi-User Memory Benchmark), a dataset of multi-user interactions containing private and shareable information across four domains. To our knowledge, MUMBench is the first public dataset designed to evaluate multiple memory operations, including retrieval, creation, update, and deletion, in a multi-user environment. Across three independent runs on MUMBench, AIM achieves 96.0% visibility classification accuracy, 58.8% strict operation accuracy, and 70.5% state-aware operation accuracy.
Read the full abstract

All authors Zachary Johnson, Nigel Boachie Kumankumah, Somya Chatterjee, Tejas Sathyamurthi, Min Chen, Xinyi Alice Li, Xiao Wang, Emily Morgan Gelchie, Jessica Lin, Sadid A. Hasan, Sulaiman Vesal

Algebraic Geometry

Seshadri slope $\K$-semistability for the blow-up of projective space along a linear subspace

We study a particular instance of the frame work from \cite{Grieve:CM:Line:Slope:Stab}, building on earlier work of Arezzo-et-al \cite{Arezzo:DellaVedova:LaNave}, Ross and Thomas \cite{Ross:Thomas:2007}, \cite{Ross:Thomas:2006} and others, which applies the theory of the Chow-Mumford line bundle to determine the Donaldson-Futaki invariant of deformation to the normal cone test configurations with respect to big and nef line bundles. In particular, here we consider the case of projective $n$-space blown-up along a linear subspace. Our main result arises as an application of \cite[Theorem 1.1]{Grieve:CM:Line:Slope:Stab}. It establishes $\K$-semistability of the deformation to the normal cone test configuration for the blow-up $\PP(\Osh_{\PP^s}^{\oplus r} \oplus \Osh_{\PP^s}(1)) \simeq \operatorname{Bl}_{\PP^{r-1}}(\PP^s)$ along the exceptional divisor and with respect to the tautological line bundle $\Osh_{\PP(\Osh_{\PP^s}^{\oplus r} \oplus \Osh_{\PP^s}(1))}(1)$.
Read the full abstract

Combinatorics

Automorphism groups of endomorphism monoids of $0$-unit $3$-valent circulant digraphs

Let $G=\Cay(\mathbb{Z}_n,S)$ be a $3$-valent circulant digraph with connection set $S=\{0,s,t\}$, where $s,t\in\mathbb{Z}_n^*$ and $t\neq\pm s$. Determining the automorphism group of its endomorphism monoid reduces to determining the subgroup $U_S(\mathbb{Z}_n)\leq\mathbb{Z}_n^*$ that normalizes $\End(G)$. In this paper, we first establish necessary and sufficient conditions for a unit circulant digraph without $2$-cycles to admit an endomorphism whose image induces a directed cycle. Using this characterization, we explicitly determine $U_S(\mathbb{Z}_n)$ for the previously unresolved $0$-unit $3$-valent family.
Read the full abstract

Machine Learning

Sampling via Decision-Flow: Training-Free Extraction of Improved Latent Reasoning Paths in Large Language Models

A central question in LLM reasoning is whether reinforcement learning (RL) instills genuinely new capabilities or merely reshapes how existing knowledge is expressed during inference. Building on the distribution-sharpening hypothesis, which holds that RL reallocates probability mass toward high-reward trajectories already latent in base models, we ask: can we unlock those latent paths without costly RL fine-tuning? We present Decision-Flow Sampling (DF-Sample), a training-free, data-free inference-time framework that constructs a hierarchical reasoning tree, scores terminal nodes for quality, and back-propagates utilities to inform each intermediate branching decision. Unlike conventional sampling strategies that make purely local step-wise choices, DF-Sample performs explicit global trajectory evaluation before committing to a path, recovering high-quality but low-probability reasoning chains that standard decoding overlooks. On GPQA, DF-Sample achieves 45.6% accuracy, surpassing power sampling (38.9%) and GRPO (39.9%), showing that a training-free method can outperform a trained one. Across three models and four benchmarks, DF-Sample consistently outperforms baselines, indicating substantial latent reasoning potential in pretrained base models.
Read the full abstract

Robotics

DATAFARM: Distribution-Aligned Task and Motion Planning for Fine-Tuning Vision-Language-Action Models

Collecting high-quality robot data remains a fundamental challenge for training robot foundation models. Task and motion planning (TAMP) offers a scalable way to generate demonstrations, but our experiments show that raw TAMP trajectories provide surprisingly little benefit when used to fine-tune pretrained vision-language-action (VLA) models, despite successfully solving the target tasks. We hypothesize that this failure arises from a behavioral distribution mismatch between planner-generated trajectories and the data used to pretrain the VLA. To address this mismatch, we introduce DATAFARM: Distribution-Aligned Task And motion planning for Fine-tuning A Robot foundation Model, an approach that incorporates the pretraining distribution directly into TAMP trajectory generation. DATAFARM aligns generated trajectories with the pretraining data in robot joint configurations, motion style, and temporal execution profiles. We evaluate DATAFARM on three tabletop manipulation tasks that TAMP can perform and a cloth-folding task beyond the capability of TAMP. DATAFARM achieves an average success rate of 56.7%, substantially outperforming raw TAMP (8.3%) while approaching human teleoperation (61.7%). On Deformable Object Manipulation, which is outside the fine-tuning distribution, the fine-tuned model retains 85% success, compared with 90% for the pretrained model. These results show that aligning planner-generated demonstrations with the pretraining distribution can make TAMP an effective source of data for VLA fine-tuning. Website and code: https://prpl-group.com/datafarm/
Read the full abstract

Cryptography and Security

Function Name Is All You Need to Detect Blockchain Application Attacks

Blockchain application attacks, targeting business logic bugs in decentralized applications (dApps), have been an increasing concern to their developers and users, causing significant financial loss. Existing attack detectors either rely on handcrafted rules for detection, or need difficult-to-obtain smart contract source code to analyze attack transactions. This makes them brittle and inapplicable in practice. In this paper, we argue that function name sequences suffice to capture the high-level semantics of a transaction, and hence can be used to detect blockchain application attacks. Our empirical study on transactions from 424 real-world attack incidents shows that 98.46% of call traces can be resolved to function names, whereas only 74.78% invoke contracts with available source code. Based on this observation, we propose TxLucent (pronounced "translucent"), an automated framework to detect blockchain application attacks by extracting application semantics from transaction call traces. TxLucent maps call traces to function name sequences and uses a transformer to learn semantics from such sequences. Consequently, TxLucent can detect attacks without relying on hand-coded patterns or source code. Our results show that TxLucent achieves a 1.56% false negative rate on 424 known incidents with 14,611 attack transactions, and an estimated 0.0017% false positive rate for benign transactions from over 500 million transactions on the Ethereum blockchain. Finally, TxLucent takes an average of 24.90 milliseconds to analyze a transaction, thus supporting real-time attack detection on popular blockchains.
Read the full abstract

Human-Computer Interaction

"I Felt Very Seen, But Still Very Alone": Longitudinal Trajectories of General-Purpose LLM Use for Socioemotional Support

People increasingly use general-purpose chatbots such as ChatGPT, Claude, and Gemini for mental health and emotional support. We report a multi-stage longitudinal qualitative study of 18 U.S. adults, conducted from April to December 2025, combining initial interviews, a four-week diary study, focus groups, and exit interviews. We find that socioemotional use often emerged gradually out of practical use and when other forms of support were unavailable. Participants developed routines and boundaries around chatbot use, which were disrupted by model updates, evolving public discourse about AI harms, and changes in personal circumstances. We demonstrate how longitudinal study captures factors beyond the human-AI dyad, and argue that HCI researchers and designers should account for users' histories with their chatbots and broader care ecologies when evaluating AI systems over time and introducing updates that may disrupt established sources of support.
Read the full abstract

All authors Meryl Ye, Briana Vecchione, Livia Garofalo, Ranjit Singh

Artificial Intelligence

Do Influence-Derived Data Perturbations Enable Machine Unlearning? A Controlled Study of Three Plausible Roles

We evaluate Deep Perturbation Learning (DPL), which perturbs training images and labels along influence-derived directions, in three roles in which prior work has positioned it for machine unlearning: a direct deletion signal (the strongest claim), a utility-preserving regularizer, and a warm start for adversarial unlearning. Evidence for the weaker roles has been used to support the stronger one, so we test each role separately under a matched protocol with exact-seed retraining baselines. An audit of the public implementation identifies two correctness issues: image directions are computed on augmented, normalized tensors but applied to raw images, and the label perturbation falls below float32 resolution, leaving labels unchanged. After correcting the image-perturbation pipeline, DPL fails the direct-deletion criterion on CIFAR-10/ResNet-18 in all three paired seeds. Its utility effects are inconsistent in sign across seeds, and once direction-computation time is counted it underperforms simple warm-start baselines. A one-seed Tiny ImageNet check likewise does not favor DPL as a regularizer or warm start; preprocessing inconsistencies in the released code make the direct comparison there inconclusive. These results cover random instance deletion only and do not rule out influence-based methods in other deletion regimes. We release a role-matched evaluation protocol and an audit checklist for perturbation-based deletion claims.
Read the full abstract

All authors Chenkai Wu, Chrispine Kambimbi, Qinyang Zeng, Jun Yan

Populations and Evolution

Identifiability of a Simple Model of Lateral Gene Transfer

In evolutionary biology, factors like lateral gene transfer complicate the tree of life, making inference of a species tree more difficult. The phenomenon of lateral gene transfer allows genetic material to be passed between organisms as opposed to inheritance, causing the tree for a particular gene to differ from the species tree. In this work, we define a model of lateral gene transfer on site patterns. In the case where lateral gene transfer is restricted to occur only between closely related species, we show that the unrooted topology of the species tree is identifiable from SNP data on the taxa. Our proof involves showing a connection between our lateral gene transfer model and graphical models on a related tree, and uses ranks of flattenings to identify splits in the tree. We also report on results of using the singular value decomposition on flattening matrices to identify the unrooted topology in simulated data.
Read the full abstract

Geometric Topology

Quantum Invariants indexed by Fibered Faces of the Thurston Polytope

We study the Gukov–Manolescu quantum invariant for oriented links. While this invariant is a single series in the knot case, we find that links admit multiple such series, each one consistent with the Melvin–Morton–Rozansky expansion of the colored Jones polynomials. We prove that convergent inverted state sums yield a family of multivariable Gukov–Manolescu series, indexed by monomials of the Alexander polynomial. We conjecture that these monomials correspond precisely to the fibered faces of the Thurston norm ball and provide extensive computational evidence for this correspondence. Further results concerning the leading term, the corresponding single-variable invariant, and the effect of partial Dehn surgery are also established.
Read the full abstract

Computation and Language

ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression

We describe six submissions under the team name ESTS to the unconstrained WMT26 Model Compression Shared Task for English–Simplified Chinese and English–Egyptian Arabic. We submit three compression operating points per translation direction, all derived from GPT-OSS-20B. We use task-specific routing mass to rank experts and cross-lingual routing divergence to allocate retained capacity across layers, then physically remove low-importance experts. The resulting specialists are recovery-tuned on GPT-5.1-generated synthetic translation data and further compressed by applying MXFP4 quantization to the retained expert projection weights. We additionally implement a robust inference system for the instruction-conditioned WMT26 setting, including category inference, output validation, retries, segmented fallback, and source-owned JSON reconstruction. Across our six submissions, parameter counts range from 4.186B to 7.770B and packed artifact sizes from 4.55 to 6.33~GiB. Internal xCOMET-XL evaluation using GPT-5.1 pseudo-references provides an internal comparison across the submitted compression operating points.
Read the full abstract

Software Engineering

PQLS: A High-Performance Python Library for Steady-State Simulation of Open Quantum Systems

PQLS (Parallel Quantum Liouvillian Solver) is a high-performance Python library for computing steady-state solutions of the Lindblad master equation. It provides a layered user-facing API with three levels of abstraction. The high-level API accepts physical descriptions of atomic ladder systems and automatically determines the required model parameters; the mid- level API accepts quantities such as Rabi frequencies, detunings, decay rates, and network topology; finally the low-level API allows users to directly specify the Hamiltonian and collapse operators. This layered design enables users to choose between physical convenience and direct numerical control depending on the application. PQLS is built on JAX and XLA to enable vectorized and hardware-accelerated computation of large parameter sweeps. Rather than solving the steady-state problem sequentially for each parameter configuration, PQLS represents multiple Hamiltonians as batched tensors and processes them through a JAX-compiled solver, reducing Python-level overhead and improving hardware utilization. This is particularly useful for applications requiring large or multidimensional parameter sweeps, such as computing frequency spectra, evaluating field-dependent responses, and performing Doppler averaging over atomic velocity distributions. PQLS has been benchmarked against QuTiP, QuTiP-JAX, and RydIQule on CPU and GPU platforms, demonstrating substantial computational speedups across the tested configurations.
Read the full abstract

High Energy Astrophysical Phenomena

Electromagnetic Probes of the Supernova Engine

Explosions from stellar collapse are important in nearly all aspects of astronomy: formation of compact remnants (including the potential seeds of supermassive black holes), the production of compact binaries the produce X-ray binaries and radio pulsars, the origin of many of the heavy elements (a critical aspect of galactic chemical evolution), and as standard candles to probe the early universe (including the nature of star formation). Because these explosions occur in extreme conditions, if we can understand them, they can be used to probe fundamental physics: matter at extreme densities (exceeding nuclear densities, quark formation), neutrino physics (including neutrino oscillations) and, in black-hole forming systems, the nature of relativity. Gravitational waves and neutrinos provide the most direct probe of stellar collapse, but these diagnostics require nearby events. Electromagnetic probes, from shock breakout to observations of supernova remnants, can be much more common and provide complementary diagnostics of the explosive engines behind stellar collapse. In this review, we review these electromagnetic diagnostics, the physics behind them and the theory and modeling work we require to take advantage of these probes.
Read the full abstract

Numerical Analysis

How to grade the accuracy of the BLAS

Motivated by accelerating machine learning (ML), many computer vendors and chip manufacturers are building accelerators for matrix multiplication, which save time and energy by operating in the lower precisions needed for ML. This has in turn motivated many efforts to use these accelerators to provide faster matrix multiplication implementations with the higher precision required by many other linear algebra applications. Motivated by the large design space of algorithms for approximating higher precision, with significant performance/accuracy tradeoffs, we provide a benchmark to "grade the accuracy" of a matrix multiplication implementation (or the BLAS more generally), ranging from an "A" for attaining the classic floating point error bound, to a "C" for attaining a weaker but still useful bound, that is satisfied by Strassen-like algorithms. We also propose "ungameable" tests that vendors or users can run to verify their promised accuracy, and describe how these different grades impact the accuracy of applications like LU, QR, and Cholesky decomposition. Our test code is publicly available at github.com/Reference-LAPACK/grading-the-BLAS for developers and users, and we also plan to publicly release all our test results.
Read the full abstract

All authors James Demmel, Greg Henry, Igor Kozachenko, Julien Langou, Xiaoye Sherry Li, Jason Riedy, Jackson Vanover