Papers

Vessel segmentation is commonly optimized as voxel-wise classification, but small local errors can strongly disrupt vascular connectivity while having little effect on overlap scores. This is particularly problematic for downstream analyses that rely on centerlines, branches, connected components, or graph structure. We propose a mesh-guided post-processing framework for repairing broken vessel segmentations produced by nnU-Net. For each predicted binary mask, a deformable template mesh is fitted to the mask surface in physical space and used as a case-specific geometric scaffold. The fitted mesh is not voxelized as the final segmentation; instead, it guides conservative reconnection of disconnected components by proposing or validating thin bridge candidates under foreground-growth constraints. We evaluated this approach in three vascular anatomies using AortaSeg24 and SEGA for the aorta, TopCoW for the Circle of Willis, and PARSE for the pulmonary arteries. Performance is measured using Dice, connected-component Dice (ccDice), and the Betti-0 number. Across these datasets, repair substantially improved connectivity while preserving overlap: Dice remained nearly unchanged, whereas ccDice increased from 0.596 to 0.992 for aorta, from 0.722 to 0.835 for TopCoW, and from 0.028 to 0.862 for PARSE. The FOMAML meta-initialization further accelerated the fitting per-case, supporting practical mesh-based repair of the vascular topology. These results suggest that explicit mesh representations can provide a useful geometric prior for correcting topological failures in otherwise accurate voxel segmentations.

In this article, we consider a wave equation on a bounded three-dimensional domain, featuring a degenerate nonlocal damping mechanism that depends on the system's energy and a source term with subquintic growth. This class of models is motivated by applications in optical physics. We establish global existence in the Shatah–Strichartz sense by combining Galerkin approximations with space-time Strichartz estimates on bounded domains. A central difficulty in the long-time analysis is the interplay between three-dimensional wave dynamics, nonlocal damping, and forcing terms that are not locally Lipschitz continuous. We prove the existence of a compact global attractor and derive an upper bound for its Kolmogorov $\varepsilon$-entropy. In the non-degenerate damping case, assuming that the domain has a smooth, strictly convex boundary, we exploit the refined Strichartz estimates of Ivanovici et al.~\cite{Ivanovici-Lascar-Lebeau-Planchon-2023}, which yield, in particular, $u\in L^3_tL^{18}_x$. Under the further assumption $f\in C^2(\mathbb R)$, this improved integrability allows us to establish quasi-stability of the associated dynamical system. Consequently, the global attractor has finite fractal dimension and higher regularity, and a generalized exponential attractor exists.

Audio classification is inherently a multi-label task, as real-world acoustic environments contain multiple simultaneous sound events. When new sound classes emerge, models must incorporate them without forgetting previously learned ones: a challenge known as class-incremental learning. Existing methods rely on storing past data and iterative gradient updates, which struggle under incomplete multi-label supervision because only the newly introduced classes are annotated at each phase, leaving old-class labels unavailable. We investigate exemplar-free analytic continual learning as a principled alternative, in which a linear classifier is updated in closed form without storing historical recordings or performing incremental back-propagation, and previously learned weights remain intact by construction. Building on analytic learning, we further propose ALMA, which addresses incomplete supervision and class imbalance through continuous old-class score estimates and frequency-based sample weighting. Experiments on a 50-class AudioSet-R benchmark across three incremental setups show that the analytic learner substantially outperforms gradient-based methods, and previously learned classes retain nearly unchanged detection performance as new classes are added. This study shows that ALMA is a simple yet effective solution to multi-label audio class-incremental learning.

We demonstrate an event-based optical camera communication system using a 120{\deg} field-of-view sensor. In daylight, three transmitting nodes achieve an aggregate 240 kbit/s, with individual links reaching 106 kbit/s at 1 m and 80 kbit/s at 320 m. Our results, establish a new baseline for event-OCC.

Large Language Models (LLMs) consume and produce a single sequence of text; hence, if text can be added to the beginning of the LLM's response, i.e., an output prefix, then all subsequent tokens will be conditioned on it. This output-prefix attack technique is a cheap black-box prompt injection. Prior work has shown this type of attack can reliably jailbreak non-reasoning models. Most reasoning models add an intermediate scratchpad reasoning step before the assistant's final response. The ability to edit this reasoning channel is exposed by some APIs and attack vectors can be leveraged for reasoning injection attacks. We present the first systematic, controlled study that isolates the scratchpad reasoning channel as an output-prefix attack vector, and the first to compare reasoning-only, output-prefix-only and reasoning-plus-output-prefix attacks across both exposed- and hidden-reasoning models. Using a factorial design of 3 prefix types $\times$ 2 reasoning injections over $1{,}800$ test cases drawn from AdvBench, we attack three 2026-era frontier models Gemini 3 Flash Preview, DeepSeek V4 Flash, and Claude Haiku 4.5. We find that injecting malicious reasoning alone is essentially inert ($\approx0\%$ attack success), but injecting the same reasoning together with a trivial output prefix raises the attack success rate to as high as $99\%$ for some models. For this type of attack we find that contextual prefixes work better than static prefixes; and that susceptibility is dependent on the model.

Auxiliary learning is an optimization paradigm in which a neural network's performance on a target task is improved by jointly training it on additional tasks. However, the mechanisms behind this improvement remain poorly understood. We study this problem using a teacher-student framework and derive a closed system of differential equations describing the dynamics of online stochastic gradient descent in the large-input limit. For linear networks, we obtain a closed-form expression for the generalization error to leading order in the learning rate, quantifying how task correlations and label noise determine the benefit of auxiliary learning. For non-linear activation functions, we develop a fluctuation-dissipation analytical theory that establishes a general relation linking the main and auxiliary errors to the corresponding single-task error. Numerical experiments support the theoretical predictions and show how auxiliary tasks improve generalization by balancing the forcing dynamics towards the optimal solution with gradient noise.

Many real-world tasks (e.g., office workflows, scientific experimentation) require LLM agents to interact repeatedly with their environments for context-dependent operations. However, such environments are often not agent-ready. First, information is often scattered and fragmented across the environment. Second, relevant evidence in the environment is often mixed with misleading information and conflicting versions. Third, environments evolve over time, introducing new noise and more challenging tasks. These challenges can substantially degrade performance for state-of-the-art AI agents (e.g., from 83.9% to 57.6%). To address these challenges, we propose Env-Rethink (a system with 27B post-trained model) that supports three main capabilities: (1) It adaptively builds Collection Maps (for organizing related files) and Event Logs (for contextualizing cross-data relationships) to supplement necessary context; (2) It further leverages the post-trained model (through offline trajectory learning) to identify underlying noise issues in the environment; (3) It ultimately evolves environments through virtual event histories that alter environmental states and evidence relationships, producing more tricky ones for further agent improvement. Experiments show that Env-Rethink can effectively improve downstream task performance (with a 15.1 percentage-point increase in mean rubric pass rate across nine models on 30 tasks).

Radar nowcasting is essential for short-term warning and emergency response, yet conventional systems mainly return future radar fields and provide limited support for operational communication and post-event verification. We formulate radar nowcasting as an evidence-grounded forecast–bulletin–audit task, in which a numerical forecaster produces both future radar fields and structured diagnostic evidence. Forecast-time bulletins use only model-available evidence, whereas post-event audits incorporate future radar truth only after the forecast horizon is observed. Based on this task formulation, WeatherDiagFlow predicts motion, growth and decay, heavy-echo risk, and uncertainty to condition rolling flow refinement, while frozen-scaffold residual calibration improves long-lead strong-echo preservation. A multi-agent layer converts the structured evidence into operational bulletins and independently generates verification audits without feeding textual outputs back into the forecaster. Experiments on FJRADAR demonstrate competitive overall performance and improved strong-echo event skill. WeatherDiagFlow therefore connects numerical prediction, evidence-grounded reporting, and auditable verification under a leakage-controlled protocol.

The astrophysical rapid neutron capture process (r-process) is a major nucleosynthesis process responsible for the production of elements heavier than copper. Prominent features in the observationally inferred r-process abundance distributions provide critical clues for astrophysical conditions and possible r-process sites. The lack of clear features for the r-process elements lighter than tellurium has led to a broad range of proposed possible scenarios and alternative processes that may produce these elements. Here we show that the enhancement found in the solar system of rhodium and ruthenium isotopes compared to neighboring isotopes can serve as such a feature. The enhancement can be traced back to sudden changes in shape of unstable rare isotopes in the path of the light element r-process. Such shape changes have been described as quantum phase transitions. This finding points to r-process scenarios with relatively high neutron densities and temperatures for the origin of rhodium and ruthenium in the solar system. Rhodium and ruthenium abundances observed in r-process enhanced stars can now be used as diagnostics for the r-process conditions that produce these lighter r-process elements. Observational data for these elements indicate that different types of r-processes may have operated in the early Galaxy, and that the conditions do not necessarily align with the weak and main r-process classifications used in the past.

Extreme-mass-ratio inspirals (EMRIs), systems in which a stellar-mass compact object spirals into a supermassive black hole ($M_1$), are prime targets for the space-based gravitational-wave interferometers and are acutely sensitive to their astrophysical environment. We quantify the joint imprint of a dark matter (DM) spike and an accretion disk on the waveforms of eccentric, equatorial Kerr EMRIs, integrating four modular effects—DM dynamical friction, DM self-gravity, Newtonian accretion-disk torques, and disk self-gravity—into a fifth post-Newtonian (5PN) augmented analytic kludge (AAK) waveform model, and assess their measurability via a Fisher-matrix analysis. The DM dynamical-friction dephasing peaks at $M_1\sim 10^6\,M_\odot$, reaching $\sim 10^2$–$10^3\,\mathrm{rad}$; there the DM spike slope is measurable to sub-percent precision and both DM channels are unambiguously detectable, while omitting DM biases the intrinsic parameters significantly. At $M_1\gtrsim 10^8\,M_\odot$ the DM and disk self-gravity channels dominate their dissipative counterparts. The disk parameters remain degenerate across the explored parameter space, since the supersonic regime dominates the inspiral and the subsonic-to-supersonic transition only mildly weakens this degeneracy without lifting it; a combined DM$+$disk analysis further reveals a cross-sector degeneracy between the DM slope and the disk parameters. This physical degeneracy, reflected in the near-singular Fisher matrix, implies that independently measuring the disk parameters requires a relativistic torque model or an electromagnetic counterpart.

LLM judges now score outputs against rubrics, in benchmarks and as training rewards. Decision models return probabilities over permitted answers with a confidence score but no free text. Two such models are Jev and an open reproduction, Perplexity's Decider. We compare each with three flash-tier LLM judges on nine panels from seven benchmarks with human judgments, giving every judge identical criterion texts. The LLM judges run in two setups, reading a whole rubric at once or one criterion at a time. Our findings suggest both decision models can often stand in for them. The LLM judges cost 16 to 357 times as much and take 5 to 350 times as long, yet in both setups each decision model trails them significantly in at most 6 of 27 comparisons. The two kinds of judge err alike, agreeing more with one another than with the human-annotated judgment labels on ordinal criteria. Over 85% of flash-tier LLM verdicts on either decision model's most confident errors repeat them, where independent errors would give about half. Each decision model's confidence ranks its errors on most ordinal panels, so each is a natural first stage for a cascade deferring uncertain verdicts to an LLM judge. Yet with flash-tier fallbacks, such cascades gained a mere 2.7 points at best over the strongest single judge, even with oracle thresholds, suggesting they mainly save cost. A cascade needs a fallback that is right where its first stage errs, but the LLM judges are mostly wrong in the same places as Jev, and the Decider is already the strongest single judge on most panels. Even each vendor's higher-tier model repeats those errors, and as a fallback it lifted no cascade more than 1.7 points above both tiers' strongest single judge. Holding across setups, effort levels, and tiers, these findings suggest designing future decision models afresh to get right what LLM judges get wrong, and vice versa.

We study how to organize Transformer depth through recurrence in flow-matching text-to-speech, varying the amount, order, and placement of weight reuse. Seven layouts perform 18 block calls per network evaluation under a common training objective and sampler. On Seed-TTS and LibriSpeech-PC, SEQUENCE applies each of nine blocks twice consecutively, retaining competitive intelligibility, speaker similarity, and predicted speech quality at 32 sampling steps with 47.1% fewer parameters than the unshared baseline. Cycling six blocks three times further reduces model size but raises 32-step word error rates relative to cycling nine blocks twice. At matched parameter counts and executed depth, reuse order and sharing position produce different quality trade-offs. These comparisons depend on sampling budget: Prefix and Suffix have similar 32-step word error rates, but Suffix is worse by 3.44 and 5.97 percentage points at four steps on the two datasets, respectively. Only Middle ranks first or second in mean word error rate at 32 and four steps on both datasets. These results show that the organization of recurrent computation affects synthesis quality, and that reuse layouts should be selected for both the sampling budget and the quality dimensions of interest.

Interferometers form a cornerstone of precision measurement, leveraging wave superposition and phase coherence to convert minute physical perturbations into measurable intensity variations. Here, we establish how paradigmatic interferometric architectures, the canonical Michelson and Mach-Zehnder interferometers, can be endowed with entirely new operating principles by replacing conventional beam splitters with judiciously designed non-Hermitian resonators. To uncover how non-Hermitian physics can assist in interferometry, we subject these canonical interferometers to a modern non-Hermitian symmetry analysis. A fundamentally transformed interference landscape then arises when the resonators and interference pathways collectively induce partially deficient spectral degeneracies, mathematically characterized as multimode versions of exceptional points. Key characteristics of these exceptional interferometers are nonanalytic destructive interference conditions, resulting in distinct bright-fringe and dark-fringe operating regimes organized by non-Hermitian winding and braiding topology. By revealing how exceptional-point-assisted interference reshapes the canonical geometries, our findings transfer central paradigms of non-Hermitian spectroscopy into the interferometric setting, and establish exceptional-point interferometry as a new platform-independent paradigm for general-purpose precision sensing, applicable from on-chip photonics to macroscopic observatories.

Cross-speaker acoustic-to-articulatory inversion requires accounting for anatomical differences between speakers. We propose a geometric adaptation framework that uses anatomical landmarks, primarily on vertebrae and dental structures,to transfer predictions from a fixed inversion model to unseen speakers. An affine transformation followed by thin-plate spline (TPS) deformation maps the predicted contours of 10 vocal-tract structures into each target speaker's geometry without retraining. Landmarks are identified in one selected /u/ frame per speaker as a common phonetic reference without assuming identical articulatory configurations across speakers, and the resulting mapping is reused across recordings. We train the model on a single-speaker rt-MRI database and evaluate adaptation on eight speakers from a separate multi-speaker rt-MRI database. We compare affine and TPS configurations using 12 or 14 landmarks. Affine12+TPS14 achieves the lowest mean point-to-closest-point error of 3.19mm. These results support the combined value of anatomical landmark information and nonrigid alignment.

Sparse direct Cholesky solvers fix one data structure for an entire matrix, but symmetric positive definite systems range from nearly dense to irregular, sometimes mixing both within one matrix. We let the data structure follow the sparsity structure, across matrices and across tiles within a matrix. Before numerical work starts, a lightweight selector captures the sparsity pattern of the Cholesky factor and routes the matrix, on one static shared-memory schedule, to one of three regimes: dense, sparse, or an intermediate semisparse regime. Dense tiles are stored in full; the semisparse regime uses a new active-column tile that keeps only the columns the factorization will touch, so banded or arrowhead-shaped structures common in spatial models still reach BLAS-3 efficiency. The approach suits the integrated nested Laplace approximation (INLA): one sparsity pattern factorized thousands of times with different values, so the one-time analysis cost is amortized and every factorization saving compounds. We evaluate the technique on 60 SPD matrices, comparing against MUMPS, PaStiX, CHOLMOD, symPACK, and Intel oneMKL PARDISO on Intel Xeon and AMD EPYC nodes; the selector alone achieves the lowest total factorization time in every regime. Summed over the suite, it beats the best fixed single-structure mode by 1.6 to 2.6x, and every alternative by 1.8 to 12.5x on Intel and 2.6 to 10.1x on AMD, with the largest gains on the most expensive factorizations. It trades more one-time analysis for less time per factorization, pulling ahead by the third factorization of a given pattern. As a first GPU extension, the dense route on one NVIDIA A100, with the factor resident on the device, runs 1.2 to 6.3x faster than on the faster CPU node, the margin widening with factor size. Solver, Python/R/Julia interfaces, benchmark suite, and results are open at https://github.com/esmail-abdulfattah/sTiles.

Traditional multivariate copulas often fail to capture complex dependence structures and may impose restrictive assumptions that limit their applicability to real-world data. To address these issues, vine copulas provide a flexible framework by constructing multivariate models from bivariate copulas. The three main types are D-vines, C-vines and R-vines. To simplify computations, the simplifying assumption is often applied, assuming conditional copulas are independent of conditioning variables. This results in some variables exerting minimal influence on the conditional copulas. Therefore, this study explores two subclasses of R-vines with node degrees capped at three, assessing its performance against existing models. The statistical results demonstrate that the proposed degree-constrained vine structures can serve as effective alternatives to existing vine copula models.

Just as motorisation made road traffic a defining noise source of the twentieth-century city, low-altitude aircraft may reshape urban acoustic exposure in the twenty-first. Yet how this noise interacts with existing sound environments and three-dimensional urban form across global cities remains unclear. We modelled identical low-altitude aircraft operations across ten urban districts spanning all inhabited continents. Horizontal exposure varied markedly across and within cities, depending on road noise conditions and urban morphology, which also strongly influenced overall attenuation. Some semi-enclosed spaces with low noise levels before low-altitude aircraft operations experienced increases above 20 dB(A), creating pronounced local contrasts in urban noise exposure. Vertically, exposure varied with height and acoustic visibility to the flight route, creating vertical exposure inequalities between storeys, façades and buildings at comparable heights. These findings show that low-altitude aircraft can reshape noise exposure across global cities, requiring route assessment to distinguish newly exposed from already exposed areas and to account for three-dimensional exposure.

We study a two-parameter family of quadratic BSDE systems of the form given by Jackson (2023) in his open questions on non-Markovian solvability. Fan, Hu, and Tang (2025) established well-posedness for every bounded terminal datum when $1/\alpha+1/\beta=1$. For every $\alpha,\beta>0$ and $M>0$, we establish existence and uniqueness in $\mathcal{S}^{\infty}\times\mathrm{BMO}$ for a dense $G_\delta$ subset of the space of terminal data satisfying $|\xi_i|\le M$ a.s., $i=1,2$, equipped with the topology of convergence in probability. We also obtain well-posedness for every terminal datum taking at most two values. Furthermore, we show that every bounded terminal problem can be reduced to one with a three-valued terminal datum while preserving its solution set. However, for every positive parameter pair with $1/\alpha+1/\beta\ne1$, we construct a three-valued terminal datum admitting at least two distinct solutions in $\mathcal{S}^{\infty}\times\mathrm{BMO}$. In particular, uniqueness can fail in Jackson's stochastic-game example $\alpha=\beta=1$.

A first-order phase transition (PT) alters a compact star on two scales: globally, via growth of a quark-matter core, and microscopically, via collision of quark bubbles, driving a simultaneous gravitational-wave (GW) emission in the kHz and MHz bands. Using a constrained ensemble of model-agnostic equations of state, we follow the accretion-driven evolution of proto-neutron stars through the PT and compute the resulting multi-band GW signal. The correlations found between the kHz emission and the MHz burst could help constrain nuclear matter and the physics of the PT. A delayed neutrino burst should accompany the signal, with the delay set by the details of the PT, thus providing a prime multi-messenger source.

Humans typically rely on vision, touch, hearing, and proprioception to perceive contact and adapt their actions during manipulation. Providing robots with comparable responsiveness therefore requires hardware that can retain and use these complementary sensory signals. Most imitation-learning systems, however, observe demonstrations primarily through vision and proprioception, limiting access to contact information that is difficult to infer visually. We present PolyUMI, an open-source platform for scalable visual–tactile–audio demonstration collection and robot deployment. Its lightweight, wireless handheld gripper records synchronized wrist-camera, optical tactile, contact-audio, and proprioceptive observations without requiring a tethered workstation. The same sensing finger can be transferred to the robot end effector, preserving the sensing geometry between demonstration collection and policy execution. To effectively use these heterogeneous observations, we further introduce VisTA, a token-level multimodal policy that integrates information across sensors and time to predict contact-aware robot actions. Experiments spanning object inference, slip control, and contact-rich manipulation show that touch and audio reveal task-relevant information beyond vision and that VisTA is competitive with or outperforms existing multimodal policies. Together, PolyUMI and VisTA provide an accessible pipeline for collecting multimodal demonstrations and learning policies that perceive physical interaction beyond vision. Project Page: https://polyumi-vista.github.io

We introduce equivariant persistent analogues of Lusternik-Schnirelmann category, topological complexity, cup length, and zero-divisors cup length for persistent spaces with group actions, and establish their stability. In particular, we investigate their behavior under equivariant interleavings and obtain bounds for the corresponding erosion distances. For Vietoris–Rips filtrations of compact metric spaces equipped with compatible group actions, we relate the erosion distances between the resulting equivariant persistent invariants to appropriate equivariant versions of the Gromov-Hausdorff distance.

Ultra-fast outflows (UFOs), launched from the inner accretion disc in active galactic nuclei (AGN), are among the prime candidates for driving AGN feedback in their host galaxies. Determining their physical properties and energetics is therefore essential for assessing their impact on galaxy evolution. This review highlights recent observational progress, with particular emphasis on the groundbreaking insights enabled by high-resolution X-ray spectroscopy. The first X-Ray Imaging and Spectroscopy Mission (XRISM)/Resolve observations of AGN winds have revealed that the innermost UFOs are kinematically structured and inhomogeneous. In the archetypal quasar PDS 456, these observations have led to a substantial upward revision of the inferred mass outflow rate, momentum flux and kinetic power. Complementary UV, optical, IR and mm observations trace the ionised, atomic and molecular phases of AGN-driven outflows from parsec to galactic scales. However, the physical coupling between the accretion-disc wind and the galactic outflows remains one of the major unresolved questions in AGN feedback. Building on the pioneering XRISM results, NewAthena's X-ray Integral Field Unit (X-IFU) will characterise the physical properties and energetics of accretion-disc winds with unprecedented precision from the local Universe up to cosmic noon, while complementary multi-wavelength observations with facilities such as the Extremely Large Telescope (ELT), James Webb Space Telescope (JWST), Atacama Large Millimeter/submillimeter Array (ALMA), and Square Kilometre Array (SKA) will determine whether and how their energy and momentum are transferred to the host galaxy.

Publicly exposed large language model (LLM) infrastructure creates a growing attack surface, yet real-world targeting remains poorly understood. We present Ollure, a low- and medium-interaction honeypot that emulates the Ollama API without a backend LLM. Spanning four deployments across cloud and university networks, Ollure operated for 84 days and recorded 290,887 interactions from 2,793 unique source IP addresses. Most of the activity consisted of automated discovery, fingerprinting, and model enumeration. However, we also observed concrete exploitation attempts against both the infrastructure and LLM layers. These included model management abuse, path traversal and SSRF probes, RCE and cryptocurrency mining payloads, resource exhaustion attempts, prompt injection, information extraction, and agent-oriented tool use. Our results provide empirical insight into real-world threats against exposed, self-hosted LLM services.

How much information can a single neuron remember? How many memories can neural networks retrieve without creating false memories? These questions are related to a basic question: how many Boolean threshold functions $f(x)=\operatorname{sgn}(a_0+\langle a,x\rangle)$, $x\in\{-1,1\}^n$, are there? In this paper, we show that the number $T_n$ of distinct Boolean threshold functions is \[ T_n=2\binom{2^n-1}{n}\bigl(1+O(n^{-99})\bigr). \] Equivalently, the capacity of a single threshold neuron is $n^2-\log_2(n!)+1+O(n^{-99})$ bits, improving the $O(n)$ error term in the result of Kahn–Komlós–Szemerédi to $O(n^{-99})$. To prove this, we show that, for $1\le r\le n-1$, and $v_1,\ldots,v_r$ are chosen at random from $\{-1,1\}^n$, \[ \mathbb P\!\left\{ \langle v_1,\ldots,v_r\rangle\cap\{-1,1\}^n =\{\pm v_1,\ldots,\pm v_r\} \right\} =1-O(n^{-99}). \] In the context of the Kanter–Sompolinsky Hamiltonian for memory retrieval, this identifies $r=n-1$ as a sharp threshold, at which, for almost every collection of $r$ memories, the only ground states are these memories and their negatives, confirming a weaker form of the Kalai–Linial–Odlyzko conjecture. It also settles a recent open problem posed by M. Anthony on the specification number of Boolean threshold functions. In addition, we show that, for every $1\le r\le n-1$, \[ \mathbb P\{v_1,\ldots,v_r\text{ are linearly dependent}\} =2\binom r2\,2^{-n}+O\!\left(2^{-n}e^{-cn}\right), \] confirming a conjecture of Kahn–Komlós–Szemerédi.

How phenotypic transformations are implemented by changes in underlying regulatory dynamics remains a central question in developmental biology. Inspired by D'Arcy Thompson's 1917 "On Growth and Form", we ask whether coherent large-scale transformations of morphology can be encoded as low-dimensional modulations of a self-organizing developmental system. We use neural cellular automata (NCAs) as bio-inspired models of distributed development, in which a shared local regulatory network grows target morphologies from a single cell. We apply low-rank adaptation (LoRA) to pretrained NCAs, representing each adapted developmental program as a low-rank modulation of a fixed regulatory scaffold. Horizontal and vertical scaling of a fully grown 2D emoji phenotype can each be implemented by rank-one adaptations. Their linear combinations parametrically control phenotype size, generalize beyond the training distribution, and compose with target-specific adapters. Strikingly, adaptations learned for one phenotype transfer zero-shot across structurally and semantically diverse phenotypes sharing the same reference scaffold, while largely preserving internal features. This suggests reusable system-level hyper-directions of scale rather than morphology-specific transformations. From approximately 25,000 independently trained phenotype-specific NCA adapters with a shared scaffold, we further identify latent low-dimensional directions that functionally control phenotypic variation including scaling, style, and symmetrical fission. Together, our results provide a computational realization of D'Arcy Thompson's remarkable grid transformations in a 2D NCA—a minimal cybernetic tissue in which variations of fully grown emoji phenotypes can be encoded, combined, and controlled through low-dimensional directions in regulatory weight space.

Current multimodal software-engineering benchmarks expose images as additional context, but do not test whether an agent can integrate evidence distributed across images into a verified repository-level repair. We present SWE-PolyVision, an executable benchmark of 92 real tasks from 36 open-source organizations, with 48 public tasks and 44 private holdouts. The release contains 402 static images and 6 videos, with at least two visual inputs per task. Each task pairs a fixed pre-fix repository with an isolated verifier and is evaluated under the supported conditions among three access modes: Text-only, Native Vision, and Tool-mediated Vision. Across eleven coding models, visual access changes which tasks are solved, but effects depend on both model and task. Two trace-linked Native Vision cases illustrate how complementary visual and textual clues can lead to source-localized, verified repairs; controlled interventions show that this conversion is not yet stable across inputs. SWE-PolyVision thus separates the availability of multi-image evidence from its successful use in repository-level repair, without treating patch success alone as proof of explicit reasoning.

While galaxies generally live in clusters and groups, particularly small groups, or other structures such as walls or filaments, some galaxies appear relatively `isolated', i.e. lacking in close neighbours. Here we use the GAMA survey, and an objective measure of isolation, to explore the most isolated galaxies. We then determine whether these have particular physical characteristics. We find that, as expected, while grouped galaxies split between blue, disk-like and star forming and red, spheroidal and quiescent, isolated galaxies are preferentially the former. However, there is a broad overlap in properties, with some isolated galaxies sharing the latter characteristics. Considering only the star-forming galaxies, when controlled by mass we find no differences between the grouped and isolated galaxies in terms of their star formation and stellar populations. Given the lack of significant interactions for the isolated sample, this implies that the evolution of present-day star-forming galaxies in all environments (at least outside rich clusters) has been, and remains, dominated by internal processes.

Distributed quantum computation requires many entangled pairs to be available simultaneously, so routing must optimize fidelity from a fixed, short-lived set of network resources rather than the rate of pairs accumulated over time. We formulate this one-shot routing problem for heterogeneous Werner-state links and jointly optimize path selection and the schedule of entanglement swapping and distillation. We show that the common purify-then-swap ordering is not optimal network-wide and that exact schedule enumeration requires $m^{\Theta(m)}$ time. We introduce Fidelity-Optimizing Local Detours (FOLD), a heuristic that starts from the highest-fidelity path and merges a detour only when distillation improves on both routes and raises end-to-end fidelity. FOLD runs in polynomial time, recovers almost all of the optimal fidelity on tractable instances, and is four orders of magnitude faster than exact optimization at circuit rank $5$. On the Internet topologies where any strategy improves on shortest-path routing, FOLD achieves the highest mean end-to-end fidelity among the polynomial baselines. We further extend the formulation to links whose fidelities are known only as distributions, where uncertainty, rather than the effort spent routing, limits the fidelity delivered. Through an example of requests competing for the same links, we show that coordination becomes a condition for feasibility, and that a shared path budget controls the fidelity trade-off between them.

Progress towards practical powder X-ray diffraction (PXRD) structure determination requires evaluation beyond small crystals and idealized patterns. X2SBench combines 152,587 simulated structure-pattern pairs with 591 curated measurements, extending evaluation to larger cells, lower-symmetry structures and diverse compositions. Fixed splits, defined inputs and common structural metrics support comparisons by crystal system, atom count and element count. Joint stratification locates weaknesses within these groups. On 558 paired targets, an X2SBench-fine-tuned model improves recovery from simulated patterns but loses accuracy on measured inputs, revealing a gap in experimental transfer. Background subtraction and smoothing improve recovery for one tested checkpoint, while augmentation and reference-lattice comparisons identify further opportunities for measurement adaptation and reliable cell estimation. An open platform provides data access and standardized result submission, establishing a shared basis for community evaluation and progress towards practical PXRD analysis.

Repeated excavation continuously reshapes pile geometry, requiring an autonomous excavator to adapt its digging targets and coordinate motion across successive excavation cycles. We present a learning-based framework for continuous autonomous excavation that integrates terrain-aware target selection with reinforcement- and imitation-learning controllers. The framework separates target-conditioned motion from local digging: a shared task-conditioned RL policy controls waypoint-guided approach and loaded transport, while an IL policy learns vision-based digging and lifting from expert demonstrations. Digging targets are selected from LiDAR elevation maps and converted into bucket-tip waypoints for motion control. The control architecture coordinates the learned policies and deterministic unloading through a shared motion interface. The complete system is deployed on a scaled hydraulic excavator with multimodal sensing and closed-loop actuator control. Offline replay and physical experiments demonstrate more consistent target selection, shorter local motion time, and increased payload compared with the respective baselines. The learned digging policy achieves a mean payload of 6.52 kg per completed cycle, compared with 2.68 kg for Fixed Dig. Three five-scoop runs further demonstrate consecutive autonomous excavation under continuously changing pile geometry.