Papers

Bhrushundi et al. (2019) introduced torus polynomial approximation as an approach to separating Majority from $ACC^0$. In this model, approximation is measured modulo integers. Krishan and Vishwanathan (2026) developed a duality framework for lower bounds and posed five open problems. We resolve two of these, concerning integer lifts and explicit polynomial approximations. An integer lift records the integer shifts that turn a torus approximation into an ordinary real approximation. We show that a degree-d torus polynomial approximating Majority with error at most 1/8 and lift contained in an interval of width $B$ must satisfy $d(B + 1) = \Omega(n)$. This rules out low-degree approximating polynomials with lifts in {-1, 0, 1}, answering their Open Problem 2. We also answer Open Problem 5 by giving, for every Boolean function on $n \geq 1$ variables, an explicit torus polynomial approximation of degree at most $n - 1$. Its error is optimal: zero when the accepting set has even cardinality and $2^{-n-1}$ otherwise. The results presented were discovered and formalized in Lean using GPT-6 Astra.

Dynamic meshes reconstructed from real-world scenes often exhibit scan-induced artifacts such as surface jitter and missing geometry, which strongly impact perceived visual quality. Objective assessment remains challenging due to the absence of artifact-free ground truth and limited perceptual understanding of artifact interactions. This paper presents a systematic and controlled perceptual study of scan-induced artifacts in dynamic meshes, focusing on surface jitter and holes. To enable systematic analysis, we introduce a physically grounded artifact synthesis framework in which holes are synthesized using a reconstruction uncertainty model derived from multi-view geometry. The study reveals asymmetric cross-artifact interactions, non-additive integration, and motion-dependent temporal masking. Based on these findings, we propose a compact, perceptually grounded no-reference quality assessment framework that encodes the observed perceptual mechanisms.

In this paper, three formulations of spin-$1/2$ Second-Order Pseudo-Hermitian quantum field theorIES (SOPHIES) are built from scratch from parity eigenstates. Pseudo-Hermiticity allows the unconstrained fields to satisfy either Fermi–Dirac or Bose–Einstein statistics, while self-conjugate Majorana-like fields are only compatible with Fermi–Dirac. The set of independent four dimension renormalizable self-interactions of the three theories are identified, and the one-loop renormalization of the unconstrained fermionic and bosonic theories coupled to an abelian gauge boson and enhanced with a Pauli term and including self-interactions is studied, together with the one-loop renormalization of the only independent quartic self-interaction of self-conjugate fields. The analysis leads to a rich set of Renormalization Group Equations with deep connections to known theories, like the Dirac formalism and the Klein–Gordon scalar through the fixed points of the Pauli coupling.

Motivated by recent advances in the synthesis of effective one-dimensional mechanical metamaterials, we develop a direct theory of strain-gradient special Cosserat rods based on the principle of virtual work. The internal virtual work is formulated intrinsically in terms of variations of the classical Cosserat strains and their spatial derivatives, without requiring a three-dimensional parent material or a stored energy. Objectivity and insensitivity to rigid virtual displacements characterize the admissible internal-work density, while localization yields the dynamic field equations, higher-order contact interactions, natural boundary conditions, and balances of linear and angular momentum. For inextensible and unshearable rods, the constrained principle determines a unique multiplier functional, although its representation by reactive force and double-force fields possesses a gauge freedom absent from classical rod theory. We also formulate objective elastic, hyperelastic, and viscoelastic constitutive laws, extend transverse hemitropy, transverse isotropy, and flip symmetry to strain-gradient rods, and derive invariant representations of symmetric stored energies. As an application, we study a planar inextensible cantilever under a compressive terminal force. We prove that the eigenvalues of the linearized problem are simple and coincide precisely with the loads at which bifurcation from the straight state occurs. In addition, each bifurcation point generates a connected global branch of nonlinear equilibria that is locally analytic near its origin.

We develop a homological model for hamiltonian reduction of Courant algebroids, parallel to the BFV formulation of hamiltonian reduction in symplectic geometry. To a hamiltonian action on a Courant algebroid, we associate a differential graded Courant–Dorfman algebra, obtained from the BFV algebra of the corresponding graded symplectic manifold; under usual regularity assumptions, its cohomology recovers the Courant–Dorfman algebra of the reduced Courant algebroid.

Clinical note generation from doctor–patient dialogue is high-volume and hallucination-intolerant: a fabricated laboratory value can create patient-safety risk, not a quality regression. We ask whether an agentic verification architecture, rather than generator scale, can make a small language model viable here. Holding a four-evaluator verification stack and refinement policy fixed and swapping only the generator between a 4B-parameter model with a QLoRA adapter and a frontier baseline, the small model reaches $0.984$ claim-level factual entailment against $0.992$ and comes within $3.3$ points on completeness, generating entirely on-device. Its deficit is confined to inter-section structural fidelity—arrangement rather than clinical comprehension, and the axis on which it also falls furthest below the corpus's reference notes. Because that deficit is an explicit verifier decision rather than a silent quality difference, it becomes a routing policy: run the small model by default, escalate what the verifier rejects. Measured end to end, that it achieved a similiar verified-rate acceptance to the frontier-only acceptance in this test set ($99.6\%$ vs.\ $98.4\%$ over $250$ encounters) while routing $62$, documenting three quarters without a frontier generation call. Ablating the loop separates architecture from scale: unrefined, the small model is accepted on $26.8\%$ of encounters and the frontier model on $74.8\%$, so verification does the most work where the generator is weakest. We also contribute section-scoped refinement, which preserves unflagged sections by construction rather than instruction, and find the baseline is not uniformly stronger—its provider refused two trauma-related transcripts the on-device model handled and certified, one facet of a provider dependency a served adapter does not carry.

Solids that are stiff to most loads and extremely compliant to one are normally built, not found. That internal atomic relaxation lowers elastic constants is textbook; what has remained untested is whether it can be driven far enough to separate the elastic scales of a crystal by an order of magnitude while leaving it mechanically stable. Density functional theory calculations of the complete in-plane elastic tensor of four non-hexagonal carbon monolayers sharing a square lattice metric show that it can: their longitudinal stiffnesses agree to within a factor of two while their shear constants span more than a factor of fifty. Twin T-graphene, previously reported as elastically isotropic from normal strains alone, is the extreme case, with $C_{66}=3.03$~N\,m$^{-1}$, nearly fifty times below graphene, at an axial stiffness typical of semiconducting monolayers, and a soft Kelvin eigenvalue a factor of $29$ below the others while the Born criteria hold. Symmetry confines the strain-induced relaxation to the $B_{2g}$ sector, a counter-torsion of the carbon sublayers about the short mid-plane bonds that removes $96\%$ of the clamped-ion shear stiffness with almost no bond stretching and leaves the semiconducting gap intact. Transverse acoustic phonons, finite-temperature dynamics, and large-deformation simulations show that the channel survives beyond the harmonic, zero-temperature limit. Because the selection rule is a point-group statement rather than a chemical one, it yields a screening criterion for other crystals and other strain components.

Large language models (LLMs) are increasingly used for code generation, yet most evaluations still focus on whether generated code passes benchmark test suites. Functional correctness alone, however, provides only a limited view of quality: benchmark tests are often incomplete, and passing them does not guarantee that code is maintainable, reliable, or well-structured. This paper {presents a large-scale static quality profiling of} LLM-generated Python code, with a focus on maintainability and reliability. Using 65{,}663 programs from 73 models in LiveCodeBench, we evaluate generated code with established static analysis tools, including Pylint, Radon, and SonarQube. Our results show systematic differences across models that are not captured by correctness alone, with maintainability emerging as the clearest source of variation. We further find that correctness often comes with structural trade-offs rather than uniformly simpler solutions. Overall, our findings motivate evaluation frameworks that assess LLM-generated code beyond pass rates alone.

Meta-evaluation of text generation metrics and LLM-judges is primarily conducted using metric-human correlations. However, prior work has shown that correlation is not an appropriate estimate of a metric's evaluation capability and can be unstable across evaluation settings, such as different datasets or languages. Therefore, we propose a regression-based meta-evaluation of metrics that learns on calibrated metric outputs to predict human ratings. We ground this measurement relative to a trivial constant-mean baseline and human agreements to provide a comparable measure of metric capability across evaluation settings. To test our meta-evaluation framework, we developed a large-scale multi-lingual style-personalized text generation (SPTG) benchmark, collecting multiple annotations across 10 topologically diverse languages and four evaluation dimensions, each with varying degrees of subjectivity. We meta-evaluated 32 metrics from three style evaluation paradigms (string-based, embedding-based, and LLM-judges) and their ensembles over these 40 evaluation settings. Our analysis revealed that even simple string- and embedding-based measurements can match strong LLM-judges when calibrated to human ratings. We also identified characteristics of human annotations that determine metric performance, as well as conditions under which a metric ensemble outperforms its individual constituents.

Many real-world questions demand not a single fact but an organized collection of them, yet existing factual knowledge benchmarks almost exclusively target single answer retrieval. We introduce ListQA, a benchmark of 9,045 human-curated, cross-validated questions (audited error rate of 1.7%) that require LLMs to recall multiple facts and compose them into structured lists of 3-10 elements. Questions range from flat lists ("Which NCAA teams went undefeated between 2000 and 2024?") to hierarchical lists with sub-attributes ("...and what was their record and result?"), spanning eight categories across 60+ countries. An evergreen design (explicit temporal constraints or historically immutable facts) keeps ground truth valid without periodic updates. For evaluation, we frame element matching as a linear sum assignment problem: optimal bipartite matching paired with LLM-based semantic grading jointly captures factual accuracy, hallucination, and format compliance. We benchmark 34 models (8 families, 42 configurations across standard and thinking enabled inference) and find: the best standard-inference model scores below 37% LLM-Judge-F1, multi-faceted questions are harder across the board, and providing the source Wikipedia page lifts scores dramatically, confirming that models can organize facts but struggle to retrieve them from parameters alone. Extended analyses cover base vs. instruction-tuned checkpoints and token efficiency. We release the dataset and evaluation code.

Mixed methods research offers established strategies for integrating qualitative and quantitative evidence, but less guidance for integrating two computational representations of the same qualitative corpus. This methodological study introduces Parallel-Then-Integrated (PTI), a four-step procedure for explicit and traceable integration. We present PTI using 1,203 interview paragraphs from 29 mothers in Nepal. The two representations were ontology-guided Patient Voice (PV) coding with PVminer and data-driven themes generated through non-negative matrix factorization (NMF). A theme-complete crosswalk documented convergence, divergence, and preprocessing-sensitive differences. Depression status, based on Patient Health Questionnaire-9 (PHQ-9) scores, served as a secondary external anchor for exploratory association checks. PTI provides a structured approach for integrating computational representations of qualitative text within mixed methods research.

Obtaining accurate estimates of the gas column density $\mathrm{N(H_2)}$ is crucial to determine the physical conditions of the molecular gas in the Interstellar Medium. This paper introduces the "Multi-Tracer Column Density Synthesis" (MT-CDS) technique, a new empirical approach to derive high dynamic-range column density maps with interferometers such as ALMA. We have compared the integrated intensity maps of two density selective tracers, namely HNC (1-0) and $\mathrm{N_2H^+}$ (1-0), obtained with IRAM-30m with ancillary Herschel column density estimates in seven star-forming regions in Orion and derived individual temperature-corrected conversion factors for each of these tracers. We applied these conversion factors obtained at low-resolution into similar high sensitivity ALMA observations to derive independent column density maps describing both the low $N_\mathrm{HNC}\mathrm{(H_2)}$ and high $N_\mathrm{N_2H^+}\mathrm{(H_2)}$ column density components at high spatial resolution (4.5"). We also combined these independent maps into estimates for the total gas column density gas $N_\mathrm{tot}\mathrm{(H_2)}$ sampling more than two orders of magnitude in column density between $\sim 10^{21.5}-10^{23.5}$ cm$^{-2}$. A comparison with independent ArTeMiS column density maps at intermediate resolution (8") shows typical deviations of $\lesssim$0.2 dex from our MT-CDS gas column density estimates. We obtained individual column density maps in five of our targets at $\sim$2000 au resolution in which ALMA observations were available, that is a factor of $\sim$5 better than originally calibrated. These results demonstrate the potential of this MT-CDS technique to retrieve accurate, high dynamic range $N\mathrm{(H_2)}$ total gas column density estimates at interferometric resolution combining multiple molecular line tracers.

Understanding the average-case complexity of $\mathsf{NP}$ is a central goal in complexity theory and cryptography. A necessary milestone is to understand the corresponding average-case versus worst-case question for the polynomial hierarchy ($\mathsf{PH}$). In a celebrated result, Hirahara [Hir21] proved that if every distributional problem in $\mathsf{PH}$ is easy on average, then every language in $\mathsf{PH}$ admits worst-case algorithms running in time $2^{O(n/\log n)}$. Remarkably, although the statement of Hirahara's theorem makes no reference to meta-complexity, all known proofs appear to use time-bounded Kolmogorov complexity and its associated meta-computational problems in an essential way. We show that the reliance on techniques and perspectives from meta-complexity is not simply a coincidence. Building on previous work of Hirahara [Hir20] and on a recent result of Hu, Lu, and Oliveira [HLO25], we prove that relating the average-case and worst-case complexities of $\mathsf{PH}$ is in fact equivalent to a purely meta-computational implication: efficient algorithms for a standard gap version of $\mathsf{MINpKT}$ with $\mathsf{PH}$ oracle access imply efficient algorithms for a mild gap version. Crucially, in these promise problems the standard gap separates a YES instance with time bound $t$ from a NO instance with time bound $\mathsf{poly}(t)$, whereas the mild gap keeps the YES threshold $t$ and moves the NO threshold to $t+\mathsf{poly}(n)$. Together with prior work, this equivalence points to the study of gap reductions as a potentially promising route toward understanding the relation between average-case and worst-case complexity. We make progress by proving that the gap in running time can be eliminated in the average-case setting. Finally, we prove that in the worst-case setting any efficient reduction between the two problems must use nonrelativizing techniques.

The automorphism groups of smooth cubic fourfolds over the complex numbers, together with actions on defining equations, are classified in previous works. In this paper, we complete the calculation of group actions on the primitive integral cohomology lattice for all of the $156$ families in the classification. We distinguish similar families, determine the remaining action-maximal cases, and use restriction to complete the list. We also determine the generic algebraic lattices for each family. The calculation depends mainly on OSCAR.

Goalball is a team sport for visually impaired athletes, played by throwing and blocking a ball while wearing eyeshades. However, existing studies provide limited guidance on which everyday training difficulties technology should address for players and coaches across team contexts. We conducted semi-structured interviews and focus groups with 18 players and 12 coaches from seven South Korean teams spanning youth, second-division, first-division, and national-team contexts. Using qualitative coding and matrix-based analysis, we describe current training practices, participants' views of these methods, the challenges they encounter, and their technological support needs. Drawing on the relationships among these four areas, we discuss implications for the design and evaluation of future support technologies that fit everyday goalball training across roles and team contexts.

Can AI agents carry out research on open scientific problems? We introduce Into the Unknown, a benchmark that provides an agent with a science paper along with a research direction to pursue and tasks it to write a follow-up paper. We work with scientists at US Department of Energy (DOE) national laboratories, who each contribute an unpublished follow-up direction from their own work together with the necessary conditions a submission must meet. While assessing the quality of a follow-up paper is challenging in general, using these conditions to form a rubric makes evaluation more tractable for research problems that have no reference answer. We have nine frontier agents write 15 follow-ups each, spanning quantum computing, plasma physics, computational chemistry, materials science, and scientific machine learning domains. None of the 135 submissions satisfies all the conditions, and 114 fail our integrity check, most often because the paper claims work that was never done or reports numbers that contradict the recorded evidence. Overall, agents readily produce papers that look like follow-ups, but none meets the standard of the domain scientist, leaving substantial headroom. Our code and data are publicly available.

LLM agents are becoming increasingly capable of handling daily workspace tasks by interacting with application environments through tool calls. However, as agents read and write sensitive user data across email, documents, and other applications, a central privacy question is whether they preserve contextual integrity (CI), that is, whether the information flows they produce comply with context-specific disclosure constraints. Existing work either focuses on model-level privacy leakage or evaluates agents in limited application environments with manually constructed disclosure constraints. In this work, we introduce AgentCIScope, a unified platform and benchmark for evaluating agent-level contextual integrity through executable, multi-step tasks across diverse, realistic application environments. AgentCIScope transforms public and synthetic privacy cases into executable agent tasks, with disclosure constraints grounded in three criteria: governing privacy policies, persona-based agreement, and a predefined sensitive-attribute taxonomy. Based on AgentCIScope, we evaluate 12 agents on 936 tasks and find that strong task-completion capability does not guarantee contextual-integrity preservation: single-run leakage rates range from 32.5% to 94.4%, while repeating each task five times further increases leakage by 4-19 percentage points. Leakage varies substantially across attributes: personal preferences and booking details have average leakage rates of 84.7% and 76.5%, compared with 17.1% for credit-card information and 29.4% for SSNs; environments: calendar has the highest average leakage rate as an information source, while WhatsApp ranks highest as an information destination; and governing privacy policies. AgentCIScope provides deep insights into agent privacy leakage and serves as a unified testbed for developing privacy-preserving agents.

Code generation models are actively being used in professional and educational settings. However, the competence of large language models (LLMs) to generate algorithmically diverse code remains largely unknown. We propose the DivAlgo benchmark, a large-scale dataset constituting algorithmically diverse solutions to programming problems. We also introduce a lightweight similarity metric to enable large-scale evaluation. Our experiments across 24 modern code generation models found that uncontrolled model-generated code is significantly less diverse than human solutions (capturing ~31% of human diversity), and even the best diverse code generation strategy could achieve only ~70% of human diversity. Moreover, while it was possible to steer the models to generate solutions based on human strategies, they were unable to reproduce the diversity of the human solutions. These findings indicate the limitations of code generation models in capturing the algorithmic diversity humans exhibit, and that overexposure to model-generated code can narrow programmers' algorithmic vocabulary.

Mobile latency is determined by how Mobile Network Operators (MNOs) route traffic: from phones, across the MNO's network to exit points where it enters the public Internet. Exit points are few, and MNOs are quiet about details, but their role prompts us to ask: What is the latency cost of sending traffic through an exit point? We develop a new method to evaluate this question, and demonstrate it on the three largest U.S. MNOs, although in principle it can apply globally. Using new analysis of existing traceroutes, we find that AT&T, T-Mobile, and Verizon funnel nationwide traffic through only 9, 5, and 9 exits, and the geographic detour they incur adds up to 62.1 ms compared to an ideal, direct-path latency. Routing delays also add latency. We show that even assuming speed-of-light paths through the exit points, latency is still up to 46.9 ms more than optimal. To understand if these costs can be overcome, we compare actual to theoretically optimal placements, finding that the latencies of current placements are 18-66% above the population-weighted optimum and 32-72% above the market-share-weighted optimum with the same number of exits. However, we show this gap is small in absolute terms: perfect placement would recover only a few milliseconds of the tens that remain. We show that geographic optimization saves little latency: the inflation users experience comes from routing and other overhead.

Human demonstrations offer a scalable alternative to costly robot data collection for dexterous manipulation, but most human datasets lack force measurements. Dexterous manipulation requires not only visual and proprioceptive understanding but also the ability to anticipate and regulate interaction forces. To scale force-aware human-robot co-training, we introduce EgoForce, a large-scale human tactile dataset capturing force-rich manipulation experience, and combine it with robot demonstrations for pretraining. We present FACT, a robot foundation model that cascades from prospective visual planning to force-aware refinement for dexterous manipulation. FACT comprises two jointly trained components. A Track-Guided Coarse Planner jointly predicts actions, future video, and visual tracks. Future video and visual tracks also provide a shared task representation that facilitates cross-embodiment knowledge transfer and coarse action generation. Building on this coarse plan, a Force-Action Joint Refiner jointly predicts future force and refines the coarse action, allowing anticipated force and action to shape each other during refinement. We evaluate the model on ten tasks requiring fine-grained coordination and force feedback, including tool use, musical instruments, and precision laboratory procedures. Using as few as 10 robot trajectories with 90 human trajectories for fine-tuning, FACT outperforms both baselines, with success-rate gains of up to 60 percentage points over T-Rex. Ablations show that force supervision in human pretraining data is crucial for downstream robot performance after fine-tuning, while visual track prediction and training with future-force prediction improve dexterous execution. Experiments varying the human-robot data ratio during fine-tuning further suggest that FACT learns from human demonstrations nearly as effectively as from robot demonstrations.

Field pea has gained interest in the Midwest U.S. and has the potential to benefit Iowa cropping systems. This study evaluated 17 accessions and 7 Canadian varieties for agronomic and seed yield performance. This study found that PI accession trials exhibit genetic diversity in phenology and canopy traits, whereas the Canadian elite varieties demonstrated great yield potential under Iowa conditions. The accession genotypes matured between 75 DAP (late June) and 95 DAP (late July) across six site-years. The top-yielding Canadian varieties averaged 1961 kg/ha across two site-years. To comprehensively assess field pea performance, this study extracted time-series Unmanned Aerial Vehicle(UAV)-derived traits, assessed UAV-derived traits, and evaluated their correlations with multiple agronomic traits and yield. The results found that UAV-derived time-series canopy area was the most yield-relevant trait (r = 0.63) among all UAV-derived canopy features, vegetation indices, and manually collected measurements. Compared with yield, UAV-derived canopy area showed higher selection reliability (H2 = 0.87, CV = 13.46%), making it suitable for phenotypic selection and accelerating Midwest field pea breeding programs. This study identifies useful UAV-derived features for pea breeding and demonstrates the value of integrating diverse germplasm, elite commercial varieties, and time-series UAV phenotyping. Combining genetic variation from the accession trial with the yield potential of high-performing commercial varieties provides valuable breeding materials and phenotyping strategies for developing locally adapted, high-yielding field pea varieties for future production in the Midwest.

In Alzheimer's disease, the amount of pathology in the brain is an imperfect predictor of how much function is lost. Pathology leads to that loss through neural circuits, but these cannot be observed neuron by neuron in patients. This study therefore simulated the central nervous system of the adult male Drosophila melanogaster: 166,700 integrate-and-fire neurons wired by the MaleCNS connectome. The simulator reproduced a published fly brain model (rate correlation 0.9997). Three kinds of damage reported in amyloid-$\beta$ flies, a lowered spike threshold, synapse loss and neuron loss, were imposed with assumed magnitudes, and circuit responses were measured in 1,424 simulated flies. The number of neurons lost was a poor guide to the deficit, which depended partly on how many synaptic contacts went with them and whether they included input neurons or one command neuron. With the same 12% of neurons removed, odour-evoked whole-brain activity was 93%, 47% or 31% of the healthy value when damage fell on poorly connected, random or highly connected neurons; these flies lost 20%, 35% and 48% of contacts. When the contacts lost were equalised, the three groups differed less; whether the remaining difference reflects where the damage fell could not be established. In 48 flies, restoring only the contacts that surviving neurons had lost recovered 62% of the whole-brain deficit, and restoring the removed neurons 39%. In flies that had lost neurons delivering the odour, these accounted for about half the deficit due to neuron loss. One neuron could decide an outcome: all 66 flies that lost the giant fibre on the stimulated side, which commands escape, had a giant fibre rate below 100 Hz, against 1 of 254 that kept it. Whole-brain activity revealed damage at milder severity than the learning index, because it changed much relative to its variation between trials.

This paper presents the iterative design and evaluation of a virtual reality (VR) nursing training system developed in collaboration with nursing education specialists to enhance procedural skill acquisition, clinical decision-making, and critical thinking among pre-licensure nursing students. Two clinical scenarios, chest pain and oxygen therapy, were created to simulate acute care settings, with design refinements informed by an expert evaluation study (n=8) and further assessed through an effectiveness study involving nursing students (n=17). Results demonstrate high user acceptance, perceived usefulness, and significant post-simulation increases in learner confidence. The VR system was found to be an accessible supplemental training tool that can foster self-directed learning. Finally, we present design recommendations for developing VR simulations for nursing education, emphasizing collaboration with domain experts, accommodating diverse learning goals, and simulating complex clinical workflows.

We study the spaces of bounded multilinear operators $\mathcal L^i(X,Y)$ endowed with $L^p_\mu$-norms, where $X,Y$ are normed spaces and $\mu$ is a measure with $p$-moments. These are norms without taking a quotient, under suitable support conditions. The topology induced coincides with the strong operator and compact-open topologies on operator-norm-bounded subsets. Furthermore, we identify the completion of these spaces with spaces of Hilbert–Schmidt operators on the reproducing kernel Hilbert space (RKHS) for $p=2$; with iterated $\gamma$-radonifying multilinear operators on the RKHS in the Gaussian case, for general $p$.

Android Application Programming Interfaces (APIs) continuously evolve, causing existing APIs to become deprecated and requiring developers to migrate their usages to recommended replacements. Although prior studies have proposed automated migration techniques, their effectiveness is often limited by the diversity and contextual complexity of API changes. This paper presents a systematic study of deprecated Android APIs, their recommended replacements, and real-world usage updates performed by developers. Based on an analysis of 1,000 deprecated APIs and 4,336 migration examples collected from GitHub, we identify six migration categories and derive recurring usage-update patterns involving Android-version checks, parameter adaptation, object initialization, data-type conversion, and changes in API interaction mechanisms. To assess their applicability, we augment the zero-shot GUPPY benchmark prompts with the migration category and corresponding usage-update pattern. On 156 benchmark tasks, the pattern-guided prompt improves correct updates from 139 to 147 and performs better in five of the six categories.

We formulate an S-matrix bootstrap in an exactly flat half-space, where boundary reflection provides geometric infrared control for four-dimensional gravity. Under the channel positivity conditions, we obtain an infrared-finite positivity bound on local four-scalar Wilson coefficients that includes a finite gravitational correction of order $G_N$ and is derived from the full UV absorptive spectrum without specifying a UV theory, imposing a spectral ansatz, requiring weak coupling or expanding the spectrum perturbatively. At fixed dimensionless external state parameters, the coefficient of the leading gravitational correction is independent of the EFT scale and requires neither an arbitrary soft resolution scale nor a model-dependent UV scale. Gaussian wave-packet scattering localizes interactions within the flat region and directly probes bulk Wilson coefficients while exponentially suppressing sensitivity to the boundary. Long-range gravitons remain sensitive to the boundary and give a finite contribution. The smearing functional passes numerical positivity tests with general half-space absorptive kernels in both crossed channels. An explicit string example shows how boundary reflection suppresses the collective contribution of string resonances to the absorptive part of the amplitude, connecting infrared control of low-energy scattering to the UV spectrum of quantum gravity in the same dispersion relation. The framework opens a route to infrared-finite constraints on effective interactions across particle physics and gravity.

In procedural tool-use domains, how much of the gap between base models is procedure rather than capability? We build an explicit program from a teacher's traces and run it in front of unchanged small models, including an 8B model. The construction, PLVR , takes the reasoning traces of a teacher, Kimi-K3, clusters their segments to locate recurring operations, has a frontier model author one typed primitive per operation, each run by a small frozen model and assembles the program by A* search over a mined rule library, under a reference-action objective. At inference the program takes one of two forms: for troubleshooting and tool search it produces the chain of thought the student would otherwise generate for itself, and the student acts on it; for multi-turn tool matching the student proposes each action and the program checks, repairs, or retries it before it goes out. On the official tau2-Bench Retail and Telecom test sets and on LiveMCPBench, with three Nemotron base models and the same program for each, the answer depends on the domain. On Telecom, Granite-4.2-8B as the student goes from 43.0% to 91.2% under the program, and every Nemotron student reaches 93.3-94.4%, against at most 58.2% for fine-tuning on the same traces, prompt optimisation of the student, or two token-matched baselines. With Granite-4.2-8B as the student, the program is within 2 to 3 points of the Nemotron students on Telecom and within 3 on Retail, and 3 to 13 points below them on LiveMCPBench: where the domain is procedural an 8B student is enough, and on tool search the student still matters. On Retail and LiveMCPBench the program is competitive with the best baseline. The spread between base models shrinks from about 7 points to about 2, averaged over the suites, with no weights changed.

Surface-related multiple elimination (SRME) is an indispensable processing step in marine seismic exploration, comprising two main components: multiple prediction and adaptive subtraction. While the multiple prediction is theoretically data-driven and relies on the data itself, the adaptive subtraction step requires a fairly subjective manual control. Our approach introduces a purely data-driven and physics-based deep learning (PDL) algorithm for improving the adaptive multiple subtraction to significantly enhanced performance. There are two main innovations in the proposed deep-learning (DL) multiple subtraction approaches: first, being data-driven and self-supervised, which removes the dependence on inflexible label preparation and model training processes; secondly, being governed by the adaptive matching filtering (AMF) equation to guide the training process. The DL module serves as a strong regression filter analogous to the traditionally known AMF. We segment a moderately sized seismic profile and an initial multiple model into a large number of patches using a patching scheme for self-contained training. The number of patches is controlled by the patch size and the shift size to enable a sufficient number of training samples required by mainstream DL platforms. The patching-based self-training and regression approach, PDL, is inherently a non-stationary matching approach that aims to obtain a globally optimal matching from the initial multiple models to the input full wavefield data through iterative patching-based training. It is also closely connected with the non-stationary regression approach, which was developed almost two decades ago. However, the PDL approach achieves even stronger regression performance due to the backpropagation-based optimization mechanism in DL, and has shown significantly improved multiple subtraction performance over a number of examples.

The processes that rapidly quench massive galaxies and maintain their quiescence remain poorly understood. Substantial molecular gas reservoirs can survive this shutdown in post-starburst galaxies, raising the question of what prevents the remaining gas from efficiently forming stars. We present deep, subarcsecond VLA 6 GHz observations of 52 massive SQuIGGLE post-starburst galaxies at $z\sim0.7$ to test the prevalence and mechanical impact of radio activity during the gas-rich post-burst phase. We detect 6 GHz continuum emission in 18/52 galaxies ($\approx$35%), spanning $\log(L_{1.4\,{\rm GHz}}/{\rm W\,Hz^{-1}})\simeq23.0$–$25.0$ and generally exceeding the contributions expected from ongoing star formation and the fading starburst, favoring an AGN origin. Ancillary data at other wavelengths provide little evidence for a population of luminous, radiatively efficient AGN, suggesting that these sources are instead dominated by comparatively weak, mechanically powered radio AGN. Radio activity is strongly concentrated within the first $\sim$200 Myr post-burst and is predominantly associated with galaxies retaining substantial gas reservoirs. Above a common radio luminosity threshold, the radio detection fraction is enhanced by a factor of $\approx$2.5 relative to the radio AGN candidate fraction in mass- and redshift-matched quiescent LEGA-C galaxies, primarily due to lower power sources. For efficient jet–ISM coupling, the inferred mechanical power could supply the approximate energy required to unbind the molecular gas on timescales comparable to its observed decline. Together, these results identify a population of radiatively inefficient AGN hosted predominantly by young, gas-rich post-starburst galaxies at $z\sim0.7$ and suggest that they may have a role in maintaining inefficient star formation.

On-policy distillation (OPD) is increasingly studied as a cheaper alternative to reinforcement learning, yet it remains a form of knowledge distillation (KD) that relies on teacher supervision. We revisit OPD from the KD perspective by asking where the teacher's advantage lies along student-generated trajectories. Controlled token flips at teacher-student disagreements directly measure the accuracy gained from teacher corrections. Single-token flips yield a modest average gain on selected failed trajectories, whereas repeated flips raise accuracy from 28.4% to 74.5%, close to the teacher's 76.9%. Gains continue to accumulate with more flips, but mass- and KL-based selection outperform random flips by 15-18 percentage points at counts of similar magnitude. Selective intervention remains useful after OPD, but greater argmax agreement does not consistently improve accuracy across model pairs. We advocate training schedules that learn more valuable teacher corrections while maintaining distributional stability. Controlled flips and fork-resolution analysis evaluate OPD beyond accuracy by measuring which teacher preferences the student acquires and what further correction can recover.