Artificial Intelligence
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 6 November
Empowered by Large Language Models (LLMs), deep research systems integrate online information into research reports through multi-step web searching and information synthesis. Evaluating these reports presents a significant challenge due to their long context, high information density, and domain specificity, necessitating the use of AI evaluation as a substitute for prohibitively expensive human assessment. However, few studies have examined the effectiveness of existing reward models or LLMs in understanding and evaluating such research reports. To bridge this gap, we introduce DeepResearch RewardBench, the first benchmark dedicated to evaluating reward modeling for research reports. We collect diverse report candidates from multiple agents or tool-augmented LLMs. These candidates are scored using query-specific golden rubrics, and we ensure data quality and difficulty through a combination of multi-round filtering and human verification. Our benchmark comprises preference datasets formatted as pairs, triplets, and quadruplets to test ranking capabilities across varying levels of difficulty. Extensive experiments demonstrate that while current reward models handle general coherence effectively, their performance significantly degrades when evaluating research reports. Generative LLMs acting as judges demonstrate comprehensive superiority in this domain, with comparative evaluation proving more effective than direct scoring methods. We anticipate that this benchmark will serve as a vital testbed for the evaluation and advancement of research agents.
General Economics
When Is Prosocial Behavior Intuitive? The Moderating Role of Empathic Perspective Taking
Why is prosocial behavior intuitive for some individuals but deliberative for others? We examine whether empathic perspective taking moderates the association between reflective processing and prosocial choice. We run a laboratory experiment with 652 participants, where we measure prosociality using a four-player Distribution Game that holds aggregate payoffs constant and orders allocations from self-regarding to egalitarian. We combine participants' choices with the Perspective Taking subscale of the Interpersonal Reactivity Index, a five-item Cognitive Reflection Test, and choice-stage response time in the allocation decision. Perspective taking significantly moderates the association between response time and prosociality. At high levels of perspective taking, longer response times are associated with modestly lower prosociality; at low levels, the association is positive but not statistically significant. No comparable moderation emerges for Cognitive Reflection Test performance. These results are consistent with the possibility that, when others' outcomes are readily represented (high perspective taking), prosocial choices may arise relatively quickly, while additional decision-specific processing may be associated with greater attention to their private cost. When others' outcomes are less salient (low perspective taking), additional processing may instead help bring them into consideration. The findings highlight the importance of distinguishing processing proximal to the specific decision from domain-general analytic reflection when explaining whether prosocial behavior appears intuitive.
Read the full abstract
Combinatorics
Bijecting $m$-Tamari intervals, trees, and tight gradings
Tight gradings, which are certain integer gradings on maximal Dyck paths, provide a combinatorial framework for computing all coefficients of wall-functions in rank-$2$ cluster scattering diagrams. We construct a bijection between a special subset of tight gradings and the intervals of the $m$-Tamari lattice, providing a new combinatorial proof of the central wall-function formula proved by Reineke. Inspired by the structure of these gradings, we introduce a class of plane trees called $m$-leafy trees that give a new model for $m$-Tamari intervals. The defining condition of an $m$-leafy tree is simple: in every full subtree, the number of leaves is greater than $m$ times the number of internal vertices. The leafy tree perspective gives rise to a quadratic functional equation with a catalytic parameter, providing an alternative way to enumerate $m$-Tamari intervals.
Read the full abstract
Machine Learning
Learning Across Heterogeneous Feature Spaces with Group Robustness
In clinical applications, patients undergo different combinations of tests, and machine learning models must therefore ensure accurate predictions across heterogeneous feature spaces. Current approaches upweight the patients with the highest loss, without accounting for the limited predictive information in their features preventing their loss from going any lower. Moreover, when patients' feature spaces are fully disjoint, current methods reduce to training independent models, preventing knowledge transfer across them. We propose a framework for group robust learning across heterogeneous feature spaces, where groups are patients sharing a feature space. Since a high loss may simply reflect uninformative features, we instead optimize for worst-group excess loss, the gap between a group's loss and an estimate of its achievable optimum. To allow collaboration between groups, we use group-specific encoders mapping into a shared latent space in which learnable Gaussian anchors cluster patients by outcome. On tabular and imaging medical datasets, including fully disjoint feature settings, our framework outperforms state-of-the-art methods when groups differ in their optimal losses or cannot train a good model on their own, reducing worst-group loss by up to 79% and excess loss by up to 95% with up to 18$\times$ fewer parameters.
Read the full abstract
Statistics Applications
Modeling the Joint Distribution of Box Score Statistics: A Gaussian Copula Approach
What is the probability that LeBron James records at least 27 points, 7 rebounds, and 7 assists tonight? While an interesting statistical question by itself, the rise of sports betting and multi-leg parlays has made answering this question increasingly relevant. However, coming up with an estimated likelihood is tricky, since box score statistics aren't independent. We estimate the joint probability that a player reaches a combination of thresholds in points, assists, and rebounds using a Gaussian copula, fit separately for each player using only their previous games. We compare it with two benchmarks: the product of the empirical marginal probabilities (i.e. assuming independence) and the empirical frequency of the joint event. Across ~22,000 player-observations, the Gaussian copula performs on par or better than either benchmark at lower threshold levels, with large gains at high joint exceedance levels. Because it is generative, the model also allows for probabilistic calculations for combinations that a player has rarely or never produced.
Read the full abstract
Artificial Intelligence
WorldMind: Test-Time Scaling for Visual Spatial Reasoning through Latent World Simulation
Vision-Language Models (VLMs) have continuous deficiency in spatial reasoning tasks that require mental spatial transformation, such as reasoning under hypothetical changes in viewpoint or position, known as perspective-taking tasks. Existing approaches either scale linguistic reasoning or augment VLMs with static geometric representations, but neither directly enables active transformation of the represented scene. World models offer this capability, yet existing systems typically communicate through synthesized RGB images, introducing a latent-to-pixel-to-latent bottleneck. We introduce WorldMind, a framework for test-time scaling of spatial reasoning through latent world simulation. Given a query and observation, the VLM selects useful spatial transformations, while a world model executes them and exposes aligned intermediate latent features directly to the VLM. The resulting latent evidence is incorporated into subsequent reasoning steps, enabling iterative spatial simulation without repeatedly decoding imagined states into pixels. To optimize this interaction, WorldMind further employs process-verifiable reinforcement learning that rewards reasoning progress only when the associated action and latent transition are valid. WorldMind thus scales spatial reasoning by allocating additional inference-time computation to active world simulation rather than generating more language or pixels. Across VSI-Bench, MMSI-Bench, and MindCube-Tiny, WorldMind achieves an average improvement of 16.0% over the base VLM and 8.3% over the strongest RGB-based world simulation.
Read the full abstract
Artificial Intelligence
What Is the Value of AI-Involved Art? Evidence From a Juried In-Person Exhibition
What is the value of AI-involved art, and to whom does it belong? We begin to answer these questions using data from an in-person, juried AI art exhibition. Our study involved artists, lawyers, and computer scientists (N=209) as either creators or evaluators of AI-involved artworks. We found that nearly half of evaluators reported willingness to buy an artwork, with pricing correlating with perceived artistic value—creativity and visual appeal—as well as the effort necessary to create it. In contrast, three-quarters of creators reported willingness to sell at prices correlating only with effort rather than artistic value. Both groups viewed prompting users and AI developers as deserving rewards and rights, leaving training data artists behind. Disciplinary training shaped these views: participants with legal training favored distributed reward models. These findings have implications for future research on human-AI creative collaboration and policy debates regarding AI-involved works.
Read the full abstract
Astrophysics of Galaxies
The impact of MUSE sky subtraction on unresolved stellar population end products: Study cases of NGC 1097 and NGC 1433
One of the main steps in data reduction for ground-based astronomy is sky subtraction. For extended objects, it is necessary to observe dedicated sky frames, usually with airmass values similar to the target's, using different pointing strategies. However, it is not well documented how different observing strategies can affect the science end product. In particular, with the use of the MUSE/VLT, this can yield distinct stellar population products, such as age, metallicity, and \alpha-enhancement. To address this, we considered two case studies seeking to investigate: (i) the sky stability in observations taken consecutively over a period of 2 hours with thin sky transparency and airmass values of 1.25-1.05, and (ii) the impact of different sky conditions on the same target. Among our main results, we find that for galaxies with 0.0014 < z < 0.0077, the sky line [NI] at \lambda 5200 coincides with the Mgb bandpass, which is essential for metallicity measurements, highlighting the need for proper sky subtraction. In addition, for regions with relative fluxes above 10 with respect to the sky in NGC1097, sky frames with variation up to ~18 % still retrieve consistent stellar population properties - in this case, this corresponds to flux values of >1500 x 10^-20 erg/s/cm^2 and Johnson-V ~19.0. In contrast, when comparing data from different nights for NGC1433, fainter regions suffer greatly from distinct sky conditions, especially in metallicity measurements, which range within +/- 0.2, strongly affecting the interpretation of stellar mass assembly and the origin of the gas (i.e., mergers might bring metal-poor gas, while an in-situ burst of star formation can yield metal-rich stars). This study is the first of its kind and highlights the importance of proper sky subtraction, providing insights regarding data limitations and observational strategies for future observations with MUSE.
Read the full abstract
Robotics
BATCH: Behavior Alignment for Cross-Embodiment Learning from Humans
Human demonstrations provide a scalable source of diverse manipulation data, but leveraging them for robot learning remains challenging due to differences in embodiment, viewpoint, action spaces, and temporal structure. We introduce BATCH, a cross-embodiment learning framework that explicitly aligns human and robot behaviors. BATCH learns self-supervised temporal representations with a cycle-consistency objective to establish cross-embodiment correspondences without manual annotation. We further introduce two complementary training objectives that leverage these correspondences within an autoregressive videoaction diffusion model: behavior-aligned co-training and human-to-robot style transfer. On real-world manipulation tasks, BATCH improves generalization to unseen task configurations compared with training on unaligned data and prior cross-embodiment baselines, with additional gains from the style-transfer objective. Our results demonstrate that explicit behavior alignment enables more effective transfer from human demonstrations to robot policies.
Read the full abstract
Computation and Language
Learning of Tree Automata applied to Neural Language Acceptors
We extract visibly pushdown grammars (VPGs) from neural language models in a fully black-box setting. Our work builds on the VPL* framework, which learns VPGs from recurrent networks by exploiting access to the target's internal state. First, we replace the white-box equivalence oracle with PAC sampling over trees. The resulting algorithm applies unchanged to transformer language models and, more generally, to any acceptor exposing only binary decisions. Second, we further observe that the original framework only ever queries the target on well-formed sequences that can be accepted by the most permissive VPG (called BParse) with the target's ranked alphabet. That is, VPL* is blind to the target's behavior outside BParse. Thus, we propose to sample trees directly instead of sequences and call the target with all the resulting sequences in order to provide an error estimate of how far the target deviates from being a VPG. Last but not least, we develop a learner that captures a tree automaton describing discovered behaviors of the target outside BParse, which results in a larger learnable space. We evaluate the approach on several cases, including transformers trained with Dyck grammars and synthetic targets designed to be outside BParse.
Read the full abstract
Quantum Physics
Bayes-Optimal Quantum Computational Sensing via Convex Spectral Design
Quantum computational sensing performs task-specific inference directly through coherent sensing dynamics, avoiding an intermediate estimation step. Existing designs, however, typically optimize quantum control parameters through nonconvex programs whose performance can depend strongly on initialization. This contribution formulates binary phase classification directly in terms of the sensor's measurement response. For a fixed query budget, generalized quantum signal processing (GQSP) restricts this response to a bounded trigonometric polynomial, while the Bayes classification error is linear in the response. These properties yield an exact finite-dimensional semidefinite program for the globally Bayes-optimal sensor response. Fejér–Riesz factorization and GQSP synthesis then recover a physically realizable control sequence. The same formulation extends naturally to sample-based design and additive phase uncertainty. Numerical tests show improved accuracy and optimization reliability over direct control-angle training; noise-aware redesign also lowers error under phase uncertainty.
Read the full abstract
Machine Learning
TrialJev: Open-Weight Single Pass Eligibility Scoring for Patient Trial Screening
Clinical trial screening requires assessing large numbers of patient-trial pairs to prioritize candidates for detailed eligibility review, often under restrictions on sharing patient records. We introduce TrialJev, an open-weight eligibility scorer that uses LoRA-adapted Qwen models and a pointer head to score natural-language answer options in a single forward pass, without autoregressive decoding. Trained on TREC Clinical Trials 2021 and evaluated on 6,975 eligible-or-excluded pairs from 50 unseen TREC 2022 synthetic patients, TrialJev achieves 80.97% (Qwen3.5-27B) and 81.84% (Qwen3.8-27B) accuracy, compared with 79.87% and 79.93% for LLMs fine-tuned on the same data to generate verdicts. In matched batch-1 measurements, verdict generation takes 2.45-2.70 times as long as the one-pass inference of TrialJev at 9B-27B. Under a common development-fitted decision correction, the two 27B TrialJev models reach 80.99% and 81.88% accuracy, compared with 81.28% for the proprietary hosted Jev API; paired confidence intervals include zero. TrialJev runs locally on a single GPU, requiring approximately 178 ms per pair at 27B, without transmitting patient text to an external model API. An A/B-logit ablation shows that avoiding decoding, rather than the pointer head alone, accounts for the measured efficiency advantage. TrialJev is intended for first-pass prioritization of candidate trials, complementing rather than replacing explanatory models and clinical review. Evaluation is limited to synthetic patients and implementation-specific timing, and clinical use has not been validated. Adapters, pointer heads, and model cards are released on Hugging Face.
Read the full abstract
Image and Video Processing
Dataset Collections: Challenges of Large-Scale Data Aggregation in 3D Medical Image Datasets
The development of deep learning-based medical AI depends on high-quality annotated data. Yet, dataset creation remains constrained by the need for expert radiologist annotations, limiting curated datasets to relatively small cohorts. To overcome this limitation, recent efforts increasingly aggregate disparate public datasets into large-scale dataset collections. However, combining datasets at scale often obscures data provenance, introducing substantial risks of reuse propagation, unintended leakage, bias amplification and distorted data distributions. In this work, we study these problems on a massive collection of 565k 3D radiological images across 1034 public datasets and provide practical resources to address them. Firstly, we formalize dataset collections as a distinct paradigm in medical image analysis and characterize their structural challenges, including reuse propagation, duplication, and provenance fragmentation. Secondly, we perform a large-scale empirical study of duplication on this collection and demonstrate extensive exact and near-duplicate reuse across public datasets via perceptual hashing and manual tracing, identifying a lower bound of 118k duplicate pairs. We make the perceptual hashes publicly accessible for comparison through our website, providing researchers with a standardized reference for identifying potential overlaps and shared provenance. Thirdly, since hashing alone cannot detect all duplicates, we introduce MAP (Metadata for Aggregation and Provenance), a lightweight, portable metadata profile for documenting provenance, known overlaps, and deduplication decisions in dataset collections. Our contributions help dataset collections shift from coarse aggregations to reliable and provenance-aware data scaling mechanisms in medical image analysis. A web tool supporting our work is available at https://mic-dct.dkfz.de/.
Read the full abstract
History and Overview of Mathematics
Discrete Spoilers: Introducing Abstract Undergraduate Mathematics Using Graph Theory
Students often have serious difficulties with many key concepts of the undergraduate mathematics curriculum. We argue that many of these difficulties can be eased by explicitly separating the key ideas from the technical details. To do so, we present finite graphs as a setting where the basic concepts appear more transparently, because on a graph most proofs only require basic techniques from linear algebra. We illustrate this approach in three areas that cover a large part of the curriculum. First, we explain basic operators from vector calculus, such as the gradient and the divergence, through the incidence matrix of a graph. Second, we show that the diffusion equation has a simple solution on a graph. We also solve the graph Dirichlet problem, in which the values at some vertices are fixed. Third, we introduce a second incidence matrix that illustrates homology, cohomology and the Helmholtz decomposition in a simple way. Finally, we outline how to apply this approach to other branches of mathematics.
Read the full abstract
Medical Physics
Towards in vivo photon dosimetry using radiation-sensitive mixed-core ultrasound phase-change contrast agents
Objective. Advanced radiotherapy applies complex treatment plans that require verification of the beam delivery. Therefore, in vivo dosimetry, measuring delivered radiation doses directly on the patient, has become increasingly important. Radiation-sensitive phase-change nanodroplets were recently proposed for in vivo radiation sensing. The radiation-induced conversion of superheated nanodroplets into microbubbles generates ultrasound contrast from which radiation deposition can be derived. While successful for in vivo proton range verification, previously used nanodroplet formulations were not suitable for in vivo photon dosimetry due to stringent requirements on the degree of superheat of the nanodroplet core. Here, we used a mixed-core strategy, combining different perfluorocarbons in the same nanodroplet core, to modulate the photon response and droplet stability at body temperature. Approach. Lipid-shelled nanodroplets with different cores (pure octafluoropropane, pure decafluorobutane and a 17:4 (v/v) octafluoropropane:decafluorobutane mixture) were dispersed in deionized water in phantom containers and irradiated with a photons at various temperatures. Simultaneous high-frame rate ultrasound acquisition allowed to link ultrasound contrast generation to the radiation delivery. Nanodroplets were characterized using 19F NMR spectroscopy. Main results. Photons induced nanodroplet-mediated ultrasound contrast strong enough for clinical applications at body temperature. A clear trade-off was observed between a stronger photon response and maintaining sufficient droplet stability. Neither octafluoropropane (lacking stability) nor decafluorobutane (lacking photon response) cores proved suitable. On the contrary, the mixed-core formulation exhibited a strong photon response with minimal background already at small radiation doses (0.1 Gy). This paves the way for future in vivo dosimetry.
Read the full abstract
Computation and Language
Retrofitting Unbounded Context onto Stock Language Models
A language model that never had to forget would keep every document it has read and every turn it has spoken, and answer from any of it. This paper shows that two stock models can do that on unmodified weights by manipulating their geometry. Rotary attention turns each pair of key dimensions, a plane, at its own rate. The slow planes turn once in thousands of tokens and carry the address a query matches when it reads a far key; the fast planes turn once in a few tokens, and the order of the tokens inside a span is already written into its keys and values by the time the span leaves the near window. At distance the position blurs, and what attention reads is a bag rather than a sequence. A far span is read by its words: under competition it is read only if it is visible at the query of the question's own handle word, and a read through a description rather than a word fails at distance in every model measured, a 70B included. These facts allow a trick. What sits in the far part of attention does not have to be the history as it was; it can be installed ahead of the prediction. The design keeps a near window of 8K tokens at their true positions and changes what lies beyond it with the prompt: every far key and value goes to the host exact, a word index finds the units the request's rare words name, a far window of slots inside the trained range holds them while the request is encoded, and a rare word the model generates fetches its units mid-generation with one exact re-encode. Implemented in llama.cpp on Mistral-7B and Llama-3.1-8B, the design reads planted single needles at 99 to 100 of 100 and three-key discrimination at 88 to 100 of 100 from 32K to one million tokens, scores 70.8 on RULER against 69.9 for dense attention on Llama-3.1-8B at 128K and 67.8 against 75.5 on Mistral-7B at 32K, and at 128K prefills at 1.85 times and decodes at 1.65 to 1.68 times stock speed.
Read the full abstract
High Energy Astrophysical Phenomena
Probing circumstellar interaction with spectroscopy of ZTF hydrogen-rich superluminous supernovae
Hydrogen-rich superluminous supernovae (H-rich SLSNe) are rare, exceptionally luminous transients. Most exhibit narrow Balmer emission and can be further classified as SLSNe-IIn, indicating interaction with dense circumstellar material (CSM). We present and analyse 304 optical spectra of 106 H-rich SLSNe observed by the Zwicky Transient Facility (ZTF). We examine their spectral evolution and Balmer-line properties, and model the H$\alpha$ profiles using Monte Carlo electron-scattering calculations. Of the 106 objects, 105 exhibit narrow Balmer emission consistent with SLSN-IIn classifications, including six with additional Ia-CSM-like characteristics and three with possible flash-ionisation features; the remaining one object exhibits only broad Balmer emission and is classified as a SLSN-II. Post-peak Balmer profiles range from approximately symmetric, exponential profiles to asymmetric profiles with prominent blue excesses or shoulders. Objects with Balmer line blue excess favour higher fitted photon-emitting shell velocities and tend to be fainter than those with persistently symmetric profiles. The H$\alpha$ pseudo-equivalent width (pEW) and FWHM of SLSNe-IIn overlap with those of lower-luminosity SNe IIn, while the pEW evolution is more diverse than in the broad-lined SLSN-II comparison sample. Our SLSNe-IIn sample's median Balmer decrement increases from 2.85 before $+100$ d to 4.40 afterwards, remaining consistently below that of the comparison Ia-CSM sample. We identify no Balmer-line morphology unique to SLSNe-IIn in the available comparison with lower-luminosity SNe IIn, although some SLSNe-IIn exhibit prolonged high-ionisation emission. Our results also support interaction with dense CSM formed through enhanced pre-explosion mass loss. We also highlight a possible hidden Ia-CSM population among luminous H-rich interacting SNe.
Read the full abstract
Probability
The satisfiability conjecture for random k-SAT formulas. Human-friendly exposition of the recent OpenAI resolution
The satisfiability conjecture was one of the most important open problems in discrete probability. It was resolved recently first by Carenini~\cite{GCarenini}, and shortly after by OpenAI~\cite{k-sat-OpenAI}, among other 370 problems solved by its internal model. The conjecture concerned the satisfiability property of random instances of the k-SAT problem. It posits the existence of a fixed value for the clause-to-variables ratio around which the probability that the formula is satisfiable drops sharply from nearly one to nearly zero as the ratio increases. While the case of large constant $k$ was resolved a decade ago by Ding, Sly and Sun~\cite{ding2015proof,ding2022proof} using the heavy machinery of the second moment method, the case of general $k$ eluded proof until ~\cite{GCarenini} and~\cite{k-sat-OpenAI}. In this short note we provide a ”human-friendly” and detailed exposition of the OpenAI's resolution of the conjecture.
Read the full abstract
Biological Physics
Learning the interactions underlying multicellular migration using many-body stochastic inference
Uncovering the interactions that give rise to emergent collective behavior in active and living matter remains a major challenge. Multicellular migration exemplifies this challenge: complex cell-cell interactions shape collective motion, but are difficult to measure directly. Here, we introduce a many-body stochastic inference framework that learns these interactions directly from experimental trajectories. We infer interaction models spanning different dynamical dependencies and complexities from short-time cell dynamics, and select among them based on their ability to predict emergent collective behaviors. Applied across three behaviorally distinct cell types, this approach yields minimal, predictive models of cell-cell interactions underlying multicellular migration. We find that non-cancerous epithelial cells coordinate multicellular migration through reciprocal repulsion and velocity alignment, while cancerous mesenchymal cells additionally exhibit previously unrecognized nonreciprocal interactions that couple to self-propulsion through two distinct mechanisms, which we term velocity inhibition in fibrosarcoma cells and velocity enhancement in breast cancer cells. Conceptually, these nonreciprocal interactions allow cells to tune the onset and speed of flocks, underscoring their role in collective migration and potentially enhancing cancer cell invasiveness at large cell densities. More broadly, our framework offers a general and principled route for extracting interaction laws of active and living systems directly from data.
Read the full abstract
Robotics
A Unified Inertial Encoder for Human and Humanoid Motion
Learning humanoid state estimation and language-grounded motion understanding from robot recordings ties training coverage to the robot's motion repertoire. Human motion datasets offer broader coverage, yet transferring this knowledge requires overcoming differences in body geometry, sensing, and physical disturbances. To enable this transfer, we propose a unified inertial encoder whose shared inertial interface uses kinematic IMU synthesis to represent human and humanoid motion at corresponding body sites. Within this common input space, joint training on human and kinematically retargeted humanoid motion combines velocity estimation and motion–text alignment with continuous-time encoding and sensor and motion randomization. The resulting model transfers zero-shot to a physical Unitree G1 without physics simulation, real-robot training trajectories, or fine-tuning, achieving the lowest velocity error among evaluated methods across five datasets, including 0.099m/s on real G1. Its motion-description retrieval also extends to retargeted G1 motion categories observed only in human training, demonstrating how human motion knowledge can support the generalization of humanoid inertial odometry and language-grounded motion understanding.
Read the full abstract
Formal Languages and Automata
Automata for MSO over Infinite Trees with Quantification over Borel Sets of Branches
Rabin's Tree Theorem says that the MSO theory of the infinite binary tree $2^*$ is decidable. Shelah showed that MSO logic becomes undecidable if this tree is extended to $2^{\le \omega}$, i.e. by allowing quantification over sets of infinite branches. A longstanding open problem is whether the decidability can be recovered in $2^{\le \omega}$ by restricting set quantification to Borel sets. We make some progress in this direction, by identifying a suitable automaton model, and showing that most of the automata-theoretic approach to Rabin's Theorem can be extended to the new framework. The only missing part is a conjecture about finite-memory determinacy in certain games. This paper states and explores the conjecture. We prove it in some restricted cases, and give lower bounds on the memory required in those games.
Read the full abstract
Materials Science
Controlled Oxidation of Tellurium Nanoparticles for Telecom Detection
Surface oxidation plays an important role in governing charge transport in semiconductor nanoparticle assemblies. Here, we investigate the optoelectronic properties of solution-processed tellurium (Te) nanoparticle ensembles prepared through gradual ambient-air exposure. Structural and spectroscopic analyses reveal interconnected elemental Te nanoparticles with coexisting alpha-TeO2 and hydroxyl-containing surface species, exhibiting a lower relative contribution of oxidized Te than previously investigated forced-air-oxidized ensembles. The resulting photodetectors demonstrate photoresponse at 405, 532, and 1550 nm, with maximum responsivities of 10.1, 14.4, and 8.5 mA/W, respectively. At 1550 nm, the device exhibits rise and fall times of 17.7 and 40.2 s, respectively, and an estimated shot-noise-limited specific detectivity of 1.3 * 10^6 Jones. These findings highlight the importance of controlling surface oxidation to balance interparticle charge transport and dark-current suppression, offering insights for the development of solution-processed Te-based photodetectors for short-wave infrared applications.
Read the full abstract
Cryptography and Security
Cryptanalysis of operad-based cryptography
Commutative operads have been proposed as a platform for public-key cryptography, with protocols for key agreement, key exchange, encryption and authentication, together with a masking technique and families of algebras with secret parameters intended to strengthen them. We show that breaking key agreement, key exchange and encryption reduces to two problems over the centralizer of the public operations: an interpolation problem for key agreement and a division problem for the other two. Neither problem requires recovering a secret, and the set from which secrets are drawn plays no role in these attacks. For the operad platforms in the two original papers (fields, the integers, residue rings, the tropical semiring with finite entries, convexors and the masked tropical algebras), the relevant problems can be solved in polynomial time in the input models specified here, and masking does not affect these attacks. The authentication protocol fails for a more elementary reason: as originally stated it does not bind the prover to any public key, and even with a registered public key the response expected by the verifier can be computed from public data. In the masked variant, verification requires the secret mask; on the concrete string platforms, one observed session with an invertible challenge entry suffices for forgery. We also observe that one of the auxiliary constructions satisfies the algebra axioms only in a degenerate case, although the corresponding protocol remains correct. The attacks are implemented and verified on random instances of representative platforms.
Read the full abstract
General Relativity and Quantum Cosmology
Scalable early warning for binary neutron-star mergers in third-generation gravitational-wave detectors
Third-generation gravitational-wave observatories will observe binary neutron-star coalescences for hours before merger. Over these extended time windows, Earth's motion and the long signal durations make standard search methods computationally demanding for low-latency detection. We propose an approach based on short-time Fourier transforms and convolutional neural networks that implicitly marginalizes over extrinsic parameters, effectively reducing the search to the chirp mass. The network learns the signals' nonlinear time-frequency morphology, approximating the optimal Bayesian detection statistic. Using a large-scale campaign of simulated signal injections, we compare the network's performance with a GPU-optimized matched-filtering baseline for both the ET$\Delta$ and ET2L configurations of the Einstein Telescope and signal durations from approximately 25 minutes ($f_{\rm min} = 5$ Hz) to approximately 6 hours ($f_{\rm min} = 2$ Hz). The network achieves comparable detection sensitivity while reducing computational costs by over four orders of magnitude, bringing processing times from minutes to milliseconds. This efficiency paves the way for rapid early-warning alerts for multi-messenger astronomy in the third-generation detector era.
Read the full abstract
Astrophysics of Galaxies
Classical Cepheids in the 4MOST era I: Methodology for 4MOST variable stars and NLTE abundance gradients in the Milky Way
Classical Cepheids (CCs) are key distance indicators and tracers of young stellar populations in the Milky Way. Their period-luminosity-metallicity (PLZ) relation also makes them calibrators of the Hubble constant $H_0$. Despite their importance, chemical composition, especially non-LTE (NLTE) effects, remains a significant source of systematics in PLZ applications. In preparation for the large CCs sample expected from the 4MIDABLE-HR (4MOST Consortium Survey 4: Milky Way Disc and Bulge High-Resolution) survey, we present a method for deriving Cepheid atmospheric parameters and chemical abundances in NLTE. Using the Payne artificial neural network (ANN) trained on NLTE spectra, we derive atmospheric parameters and abundances for 17 elements: O, Na, Mg, Al, Si, Ca, Ti, Mn, Fe, Co, Ni, Sr, Y, Ba, Eu in NLTE and C and Cr in LTE. We apply the new method to archival spectra of Cepheids and derive radial chemical gradients in the Milky Way. We derive robust NLTE stellar parameters and abundances for Cepheids spanning metallicities from $\sim -0.87$ to $\sim +0.35$ dex. For stars with multi-epoch spectra, we find that NLTE abundances are consistent across the pulsation cycle, whereas LTE values show larger dispersions and biases. The resulting abundance distributions and gradients align well with Galactic disk expectations from chemical evolution models. We confirm that the gradients flatten in the outer disk beyond $R_{\rm GC} \sim 15\,\mathrm{kpc}$ and provide the first NLTE gradients for Co, Ni, Y, and Eu. We show that our NLTE astrophysical analysis yields robust solutions for atmospheric parameters and chemical abundances across spectra at the resolving power of 4MOST-HR. This enables large, homogeneous studies with upcoming surveys such as 4MOST 4MIDABLE-HR and provides a solid basis for tightening constraints on Galactic chemical evolution models and Cepheid distance-scale work.
Read the full abstract
Algebraic Geometry
Adaptive visualization for enumerative problems
We present an adaptive visualization framework and its implementation in the julia package AdaptiveVisualization.jl. Our algorithm visualizes functions using an adaptive sampling procedure that concentrates the evaluation budget at interesting regions of the domain. We illustrate our package on our main application: visualizing parameter spaces of enumerative problems.
Read the full abstract
Superconductivity
Mimicry of $\textit{d}$-wave Andreev Signatures by Interlayer Pairing
The recent discovery of high-$T_{\rm c}$ superconductivity in bilayer nickelates has attracted enormous attention. Despite dominant theoretical studies pointing to interlayer pairing, the pairing symmetry is strongly debated due to conflicting evidence on nodeless bulk gaps and $d$-wave-like transport signatures. Here, we demonstrate that a nodeless interlayer pairing can naturally mimic $d$-wave Andreev signatures when the point contact is not evenly coupled to the two layers. The phenomenon can be understood by recognizing how a contact potential, which is odd under the layer-exchanging mirror symmetry, acts like an effective Zeeman field on the layer pseudospin. This pseudo field then binds in-gap Andreev states that sweep through the gap with the width and strength of the barrier. We further demonstrate in a realistic model for La$_3$Ni$_2$O$_7$ that the simulated subgap conductance for a top-layer barrier can reproduce the orientation dependent ZBCP usually interpreted as a fingerprint of $\textit{d}$-wave pairing. By revealing this mirror-selective Andreev response from an interlayer pairing state, our work offers a natural scenario to reconcile the conflicting experimental evidence in bilayer nickelates and advances our understanding of unconventional superconductivity.
Read the full abstract
Earth and Planetary Astrophysics
The deuterium-rich atmosphere of a mature super-Jupiter
The deuterium-to-hydrogen ratio (D/H) has provided key insights into the origin and evolution of the Solar System. Using the James Webb Space Telescope, we detected deuterated methane at $48\sigma$ in a mature giant exoplanet, $\epsilon$ Ind Ab. We measure the planet's D/H ratio to be $(4.4 \pm 0.7) \times 10^{-5}$, which is similar to those of Uranus and Neptune but about twice those of Jupiter, Saturn, and the protosolar nebula. $\epsilon$ Ind Ab's heavy elements (C, O, N, and S) are uniformly enriched to a level similar to Jupiter's. These measurements require $\epsilon$ Ind Ab to have accreted solids with $\mathrm{D/H} \gtrsim 1\%$, at least 30 times more deuterium-rich than typical Solar System comets. Such D/H values are found in the interstellar comet 3I/ATLAS and in organic molecules in star-forming regions, indicating that $\epsilon$ Ind Ab accreted pristine solids formed in a very cold environment.
Read the full abstract
High Energy Physics (Theory)
The Ghost and Its Secret Partner
The propagator of a ghost field acquires a pair of first-sheet complex conjugate poles above the multi-particle threshold. We show that this pole structure reveals the existence of a hidden asymptotic composite field – the ghost's secret partner – that remains coupled to the ghost asymptotically. This coupling induces persistent interference between the negative-norm one-particle state and the positive-norm composite-particle state. We analyze this phenomenon in four-derivative theories, in particular quadratic gravity, where the quantum dynamics prevents the spin-two ghost from appearing as an isolated asymptotic excitation. Although our results do not yet rule out observable negative probabilities, they may provide a dynamical basis for alternative amplitude prescriptions or superselection rules usually chosen or imposed ad hoc in theories with indefinite-norm vector spaces.
Read the full abstract
Astrophysics of Galaxies
A DOLPHOT/Euclid Study of the Resolved, Faint Outer Milky Way Globular Cluster AM-1
We present a resolved star study of AM-1, a distant ($D \simeq 120$ kpc), faint ($M_V \sim -5.1$) Milky Way (MW) globular cluster, based on Euclid Quick Data Release 1 (Q1) imaging. To perform photometry of the Euclid VIS and NISP imaging, we develop a Euclid-specific module for the crowded field stellar photometry package DOLPHOT. We construct and present empirical point spread function (PSF) models for VIS and NISP, along with a photometric calibration, entirely from the public Q1 images. Our photometry reaches $5\sigma$ point source depths of $I_E, Y_E, J_E, H_E = 27.1, 25.3, 25.5, 25.3$, $\sim$1-1.5 mag deeper than the public Q1 catalog, as expected for dedicated resolved star PSF photometry. We resolve the cluster to $\sim$1 mag below its main sequence turnoff in all bands. The Euclid photometry agrees with color-magnitude diagram (CMD) modeling of archival HST/WFPC2 imaging and with the 2MASS-anchored VISTA Hemisphere Survey to $\lesssim0.02$ mag in $I_E$, $J_E$, and $H_E$, with a $\sim$0.07 mag offset in $Y_E$. We find that AM-1 hosts an old ($\sim$10.6 Gyr), intermediate-metallicity ([Fe/H] $\sim -1$) population, with a present-day stellar mass of $\sim 2\times10^4\,M_\odot$ and $M/L_V \simeq 2.2$, and has a half-light radius of $\sim$12-16 pc with a round, low-concentration profile. These affirm earlier studies placing AM-1 among the most extended, diffuse outer-halo clusters, which dynamical work identifies as accreted. We describe and publicly release DOLPHOT v2.1, the Euclid module, and its empirical PSFs, which together enable Euclid resolved star studies ranging from faint MW systems such as AM-1 to more luminous galaxies throughout the Local Volume. We anticipate further refinements, including spatially varying PSFs, with Euclid Data Release 1.