Search

We study Online Convex Optimization (OCO) with adversarial constraints, where an online algorithm must make repeated decisions to minimize both convex loss functions and cumulative constraint violations. We focus on a setting where the algorithm has access to predictions of the loss and constraint functions. Our results show that we can improve the current best bounds of $ O(\sqrt{T}) $ regret and $ \tilde{O}(\sqrt{T}) $ cumulative constraint violations to $ O(\sqrt{E_T(f)}) $ and $ \tilde{O}(\sqrt{E_T(g)}) $, respectively, where $ E_T(f) $ and $ E_T(g) $ represent the cumulative prediction errors of the loss and constraint functions. In the worst case, where $ E_T(f) = O(T) $ and $ E_T(g) = O(T) $ (assuming bounded loss and constraint functions), our rates match the prior $ O(\sqrt{T}) $ results. However, when the loss and constraint predictions are accurate, our approach yields significantly smaller regret and cumulative constraint violations. Notably, if the constraint function remains constant over time, we achieve $ \tilde{O}(1) $ cumulative constraint violation, aligning with prior results., Comment: 12 pages

Probabilistic Ising machines (PIMs) provide a path to solving many computationally hard problems more efficiently than deterministic algorithms on von Neumann computers. Stochastic magnetic tunnel junctions (S-MTJs), which are engineered to be thermally unstable, show promise as entropy sources in PIMs. However, scaling up S-MTJ-PIMs is challenging, as it requires fine control of a small magnetic energy barrier across large numbers of devices. In addition, non-spintronic components of S-MTJ-PIMs to date have been primarily realized using general-purpose processors or field-programmable gate arrays. Reaching the ultimate performance of spintronic PIMs, however, requires co-designed application-specific integrated circuits (ASICs), combining CMOS with spintronic entropy sources. Here we demonstrate an ASIC in 130 nm foundry CMOS, which implements integer factorization as a representative hard optimization problem, using PIM-based invertible logic gates realized with 1143 probabilistic bits. The ASIC uses stochastic bit sequences read from an adjacent voltage-controlled (V-) MTJ chip. The V-MTJs are designed to be thermally stable in the absence of voltage, and generate random bits on-demand in response to 10 ns pulses using the voltage-controlled magnetic anisotropy effect. We experimentally demonstrate the chip's functionality and provide projections for designs in advanced nodes, illustrating a path to millions of probabilistic bits on a single CMOS+V-MTJ chip.

As platforms increasingly rely on learning algorithms, collectives may form and seek ways to influence these platforms to align with their own interests. This can be achieved by coordinated submission of altered data. To evaluate the potential impact of such behavior, it is essential to understand the computations that collectives must perform to impact platforms in this way. In particular, collectives need to make a priori assessments of the effect of the collective before taking action, as they may face potential risks when modifying their data. Moreover they need to develop implementable coordination algorithms based on quantities that can be inferred from observed data. We develop a framework that provides a theoretical and algorithmic treatment of these issues and present experimental results in a product evaluation domain., Comment: Code available at: https://github.com/GauthierE/statistical-collusion

Query-focused summarization (QFS) gives a summary of documents to answer a query. Past QFS work assumes queries have one answer, ignoring debatable ones (Is law school worth it?). We introduce Debatable QFS (DQFS), a task to create summaries that answer debatable queries via documents with opposing perspectives; summaries must comprehensively cover all sources and balance perspectives, favoring no side. These goals elude LLM QFS systems, which: 1) lack structured content plans, failing to guide LLMs to write balanced summaries, and 2) use the same query to retrieve contexts across documents, failing to cover all perspectives specific to each document's content. To overcome this, we design MODS, a multi-LLM framework mirroring human panel discussions. MODS treats documents as individual Speaker LLMs and has a Moderator LLM that picks speakers to respond to tailored queries for planned topics. Speakers use tailored queries to retrieve relevant contexts from their documents and supply perspectives, which are tracked in a rich outline, yielding a content plan to guide the final summary. Experiments on ConflictingQA with controversial web queries and DebateQFS, our new dataset of debate queries from Debatepedia, show MODS beats SOTA by 38-59% in topic paragraph coverage and balance, based on new citation metrics. Users also find MODS's summaries to be readable and more balanced., Comment: Accepted at NAACL 2025(main)

Due to labor shortages in specialty crop industries, a need for robotic automation to increase agricultural efficiency and productivity has arisen. Previous manipulation systems perform well in harvesting in uncluttered and structured environments. High tunnel environments are more compact and cluttered in nature, requiring a rethinking of the large form factor systems and grippers. We propose a novel codesigned framework incorporating a global detection camera and a local eye-in-hand camera that demonstrates precise localization of small fruits via closed-loop visual feedback and reliable error handling. Field experiments in high tunnels show our system can reach an average of 85.0\% of cherry tomato fruit in 10.98s on average.

The emergence of Machine Learning systems everywhere raises new challenges, such as dealing with interactions or competition between multiple learners. In that goal, we study multi-agent sequential decision-making by considering principal-agent interactions in a tree structure. In this problem, the reward of a player is influenced by the actions of her children, who are all self-interested and non-cooperative. Our main finding is that it is possible to steer all the players towards the globally optimal set of actions by simply allowing single-step contracts between them. A contract is established between a principal and one of her agents: the principal actually offers the proposed payment if the agent picks the recommended action. The analysis poses specific challenges due to the intricate interactions between the nodes of the tree. Within a bandit setup, we propose algorithmic solutions for the players to end up being no-regret with respect to the optimal pair of actions and contracts. In the long run, allowing contracts makes the players act as if they were collaborating together, although they remain non-cooperative.

Herein we report the first examples of reactive mesogenic materials (RMs) which exhibit fluid ferroelectric order based on the recently discovered ferroelectric nematic (NF) phase. We term these materials NF RMs and they provide the first steps towards the next generation of ferroelectric liquid crystalline polymer networks. We report the chemical synthesis and characterisation of the liquid crystalline properties of these materials, demonstrating that they have the lowest longitudinal molecular dipole moments of any reported NF material of 7.39 D. We go on to demonstrate a potential use case of this new class of reactive material through the polymer stabilisation of a matrix which exhibits the NF phase, increasing the phase range of the ferroelectric phase from 75 {\deg}C to 120 {\deg}C. The NF RMs reported herein are an exciting step forward in ferroelectric liquid crystal research, demonstrating that reactive NF materials are achievable, allowing for the future development of liquid crystalline ferroelectric networks, elastomers and polymers.

Machine learning classifiers often produce probabilistic predictions that are critical for accurate and interpretable decision-making in various domains. The quality of these predictions is generally evaluated with proper losses like cross-entropy, which decompose into two components: calibration error assesses general under/overconfidence, while refinement error measures the ability to distinguish different classes. In this paper, we provide theoretical and empirical evidence that these two errors are not minimized simultaneously during training. Selecting the best training epoch based on validation loss thus leads to a compromise point that is suboptimal for both calibration error and, most importantly, refinement error. To address this, we introduce a new metric for early stopping and hyperparameter tuning that makes it possible to minimize refinement error during training. The calibration error is minimized after training, using standard techniques. Our method integrates seamlessly with any architecture and consistently improves performance across diverse classification tasks.

The framework of uncoupled online learning in multiplayer games has made significant progress in recent years. In particular, the development of time-varying games has considerably expanded its modeling capabilities. However, current regret bounds quickly become vacuous when the game undergoes significant variations over time, even when these variations are easy to predict. Intuitively, the ability of players to forecast future payoffs should lead to tighter guarantees, yet existing approaches fail to incorporate this aspect. This work aims to fill this gap by introducing a novel prediction-aware framework for time-varying games, where agents can forecast future payoffs and adapt their strategies accordingly. In this framework, payoffs depend on an underlying state of nature that agents predict in an online manner. To leverage these predictions, we propose the POWMU algorithm, a contextual extension of the optimistic Multiplicative Weight Update algorithm, for which we establish theoretical guarantees on social welfare and convergence to equilibrium. Our results demonstrate that, under bounded prediction errors, the proposed framework achieves performance comparable to the static setting. Finally, we empirically demonstrate the effectiveness of POWMU in a traffic routing experiment.

This work identifies the first privacy-aware stratified sampling scheme that minimizes the variance for general private mean estimation under the Laplace, Discrete Laplace (DLap) and Truncated-Uniform-Laplace (TuLap) mechanisms within the framework of differential privacy (DP). We view stratified sampling as a subsampling operation, which amplifies the privacy guarantee; however, to have the same final privacy guarantee for each group, different nominal privacy budgets need to be used depending on the subsampling rate. Ignoring the effect of DP, traditional stratified sampling strategies risk significant variance inflation. We phrase our optimal survey design as an optimization problem, where we determine the optimal subsampling sizes for each group with the goal of minimizing the variance of the resulting estimator. We establish strong convexity of the variance objective, propose an efficient algorithm to identify the integer-optimal design, and offer insights on the structure of the optimal design.

The circumstellar liquid-water habitable zone guides our search for potentially inhabited exoplanets, but remains observationally untested. We show that the inner edge of the habitable zone can now be mapped among exoplanets using their lack of surface water, which, unlike the presence of water, can be unambiguously revealed by atmospheric sulfur species. Using coupled climate-chemistry modelling we find that the observability of sulfur-gases on exoplanets depends critically on the ultraviolet (UV) flux of their host star, a property with wide variation: most M-dwarfs have a low UV flux and thereby allow the detection of sulfur-gases as a tracer of dry planetary surfaces; however, the UV flux of Trappist-1 may be too high for sulfur to disambiguate uninhabitable from habitable surfaces on any of its planets. We generalise this result to show how a population-level search for sulfur-chemistry on M-dwarf planets can be used to empirically define the Habitable Zone in the near-future., Comment: Published in Science Advances

We present new anisotropy measurements in the near-infrared (NIR) for angular multipoles $300<\ell<10^5$ using imaging data at 1.1 $\mu$m and 1.8 $\mu$m from the fourth flight of the Cosmic Infrared Background ExpeRiment (CIBER). Using improved analysis methods and higher quality fourth flight data, we detect surface brightness fluctuations on scales $\ell<2000$ with CIBER auto-power spectra at $\sim14\sigma$ and 18$\sigma$ for 1.1 and 1.8 $\mu$m, respectively, and at $\sim10\sigma$ in cross-power spectra. The CIBER measurements pass internal consistency tests and represent a $5-10\times$ improvement in power spectrum sensitivity on several-arcminute scales relative to that of existing studies. Through cross-correlations with tracers of diffuse galactic light (DGL), we determine that scattered DGL contributes $<10\%$ to the observed fluctuation power at high confidence. On scales $\theta > 5'$, the CIBER auto- and cross-power spectra exceed predictions for integrated galactic light (IGL) and integrated stellar light (ISL) by over an order of magnitude, and are inconsistent with our baseline IGL+ISL+DGL model at high significance. We cross-correlate two of the CIBER fields with 3.6 $\mu$m and 4.5 $\mu$m mosaics from the Spitzer Deep Wide-Field Survey and find similar evidence for departures from Poisson noise in Spitzer-internal power spectra and CIBER $\times$ Spitzer cross-power spectra. A multi-wavelength analysis indicates that the auto-power of the fluctuations at low-$\ell$ is bluer than the Poisson noise from IGL and ISL; however, for $1' <\theta < 10'$, the cross-correlation coefficient $r_{\ell}$ of nearly all band combinations decreases with increasing $\theta$, disfavoring astrophysical explanations that invoke a single correlated sky component., Comment: 41 pages, 32 figures. Submitted to ApJ

Community detection methods play a central role in understanding complex networks by revealing highly connected subsets of entities. However, most community detection algorithms generate partitions of the nodes, thus (i) forcing every node to be part of a community and (ii) ignoring the possibility that some nodes may be part of multiple communities. In our work, we investigate three simple community association strength (CAS) scores and their usefulness as post-processing tools given some partition of the nodes. We show that these measures can be used to improve node partitions, detect outlier nodes (not part of any community), and help find nodes with multiple community memberships., Comment: 10 pages, 8 figures, 2 tables

Brown dwarfs with known physical properties (e.g., age and mass) are essential for constraining models of the formation and evolution of substellar objects. We present new high-contrast imaging observations of the circumbinary brown dwarf HII 1348B $\mathord{-}$ one of the few known substellar companions in the Pleiades cluster. We observed the system in the infrared (IR) $L^\prime$ band with the Large Binocular Telescope Interferometer (LBTI) in dual-aperture direct imaging mode (i.e., with the two telescope apertures used separately) on 2019 September 18. The observations attained a high signal-to-noise ratio ($\mathrm{SNR} > 150$) photometric detection and relative astrometry with uncertainties of ${\sim}5 \ \mathrm{mas}$. This work presents the first model of the companion's orbital motion using relative astrometry from five epochs across a total baseline of 23 years. Orbital fits to the compiled data show the companion's semimajor axis to be $a = 140 \substack{+130 \\ -30} \ \mathrm{au}$ with an eccentricity of $e = 0.78 \substack{+0.12 \\ -0.29}$. We infer that HII 1348B has a mass of $60\mathord{-}63 \pm 2 \ \mathrm{M_J}$ from comparison to brown dwarf evolutionary models given the well-constrained distance and age of the Pleiades. No other objects were detected in the HII 1348 system, and through synthetic planet injection and retrievals we establish detection limits at a cluster age of $112 \pm 5$ Myr down to ${\sim}10\mathord{-}30 \ \mathrm{M_J}$ for companions with projected separations of $21.5\mathord{-}280 \ \text{au}$. With this work, HII 1348B becomes the second directly imaged substellar companion in the Pleiades with measured orbital motion after HD 23514B., Comment: AASTeX v6.3.1, 29 pages with 8 figures, accepted for publication in The Astronomical Journal

We present a 3D shape analysis of both dark matter (DM) and stellar matter (SM) in simulated dwarf galaxies to determine whether stellar shape traces DM shape. Using 80 central and satellite galaxies from three simulation suites (Marvelous Massive Dwarfs, Marvelous Dwarfs, and DC Justice League) spanning stellar masses of $10^6$--$10^{10}$ M$_\odot$, we measure 3D shapes through the moment of inertia tensor at two times the effective radius to derive axis ratios ($C/A$, $B/A$) and triaxiality. We find that stellar shape does indeed follow DM halo shape for our dwarf galaxies. However, the presence of a stellar disk in more massive dwarfs (M$_* \gtrsim 10^{7.5}$ M$_\odot$) pulls the distribution of stellar $C/A$ ratios to lower values, while in lower mass galaxies the gravitational potential remains predominantly shaped by DM. Similarly, stellar triaxiality generally tracks dark matter halo triaxiality, with this relationship being particularly strong for non-disky galaxies though weaker in disky systems. This correlation is reinforced by strong alignment between SM and DM axes, particularly in disk galaxies. Further, we find no detectable difference in either SM or DM shape comparing two different SNe feedback implementations, demonstrating that shape measurements may be robust to different implementations of baryonic feedback in dwarf galaxies. We also observe that a dwarf galaxy's shape is largely unperturbed by recent mergers (with merger ratios $>4$). This comprehensive study demonstrates that stellar shape measurements can serve as a reliable tool for inferring DM shapes in dwarf galaxies., Comment: 22 pages, 14 figures, Submitted to ApJ

Intracellular networks process cellular-level information and control cell fate. They can be computationally modeled using Boolean networks, which are implicit-time causal models of discrete binary events. These networks can be embedded in computational agents to drive cellular behavior. To explore this integration, we construct a set of candidate interventions that induce apoptosis in a cell-survival network of a rare leukemia using exhaustive search simulation, stable motif control, and an individual-based mean field approach (IBMFA). Due to inherent algorithmic limitations, these interventions are most suitable for cell-level determinations, not the more realistic multicellular setting. To address these limitations, we treat the target control solutions as putative targets for therapeutic interventions and develop a pipeline to translate them to continuous-time multicellular, agent-based models. We set the discrete-to-continuous transitions between the Boolean network and multicellular model via thresholding and produce simple computational simulations designed to emulate situations in experimental and translational biology. These include a series of simulations: constant substrate gradients, global substrate pulses, and time-varying boundary conditions. We find that interventions that perform equally well in the implicit-time single-cell setting are separable in the multiscale setting in ability to impact population growth and spatial distribution. Further analysis shows that the population and spatial distribution differences arise from differences in internal dynamics (stable motif controls versus target controls) and network distance between the intervention and output nodes. This proof of concept work demonstrates the importance of accounting for internal dynamics in multicellular simulations as well as impacts on understanding of Boolean network control.

We study the sample complexity of online reinforcement learning for nonlinear dynamical systems with continuous state and action spaces. Our analysis accommodates a large class of dynamical systems ranging from a finite set of nonlinear candidate models to models with bounded and Lipschitz continuous dynamics, to systems that are parametrized by a compact and real-valued set of parameters. In the most general setting, our algorithm achieves a policy regret of $\mathcal{O}(N \epsilon^2 + \mathrm{ln}(m(\epsilon))/\epsilon^2)$, where $N$ is the time horizon, $\epsilon$ is a user-specified discretization width, and $m(\epsilon)$ measures the complexity of the function class under consideration via its packing number. In the special case where the dynamics are parametrized by a compact and real-valued set of parameters (such as neural networks, transformers, etc.), we prove a policy regret of $\mathcal{O}(\sqrt{N p})$, where $p$ denotes the number of parameters, recovering earlier sample-complexity results that were derived for linear time-invariant dynamical systems. While this article focuses on characterizing sample complexity, the proposed algorithms are likely to be useful in practice, due to their simplicity, the ability to incorporate prior knowledge, and their benign transient behavior., Comment: 18 pages, 1 figure

Techniques for preclinical intensity modulated radiation therapy are being developed to improve translation by replicating the clinical paradigm. This study presents the first treatment planning comparison between small animal IMRT (SA-IMRT) and three-dimensional conformal radiotherapy (CRT) in a model application, oxygen-guided dose painting of tumor hypoxia, using actual mouse data. A novel compensator-based platform was employed to generate SA-IMRT and CRT plans with 2-15 beam angles for seventeen mice with fibrosarcoma tumors. The whole tumor received a dose of 22.5 Gy, with a simultaneous integrated boost of 13 Gy to hypoxic voxels identified via electron paramagnetic resonance imaging. Plan quality was assessed using the Paddick conformity index (CI), uniformity, and dose volume histograms. For 3-angles, SA-IMRT yielded significantly improved dose conformity (median hypoxic CI =0.45 versus 0.17), tumor dose uniformity (11.0% versus 14.3%), and dosimetric spread between boost and non-boost targets (D50% difference = 13.0 Gy [ideal], 13.1 Gy [SA-IMRT], 7. 3 Gy [CRT]). No significant improvement in CI was associated with >3 beam angles (Wilcoxon signed-rank test, p < 0.05). This study demonstrates that SA-IMRT provides significant improvements in radiation plan quality and yields dose distributions that more closely mimic the clinical setting relative to current CRT approaches.

Scaling test-time compute is a promising axis for improving LLM capabilities. However, test-time compute can be scaled in a variety of ways, and effectively combining different approaches remains an active area of research. Here, we explore this problem in the context of solving real-world GitHub issues from the SWE-bench dataset. Our system, named CodeMonkeys, allows models to iteratively edit a codebase by jointly generating and running a testing script alongside their draft edit. We sample many of these multi-turn trajectories for every issue to generate a collection of candidate edits. This approach lets us scale "serial" test-time compute by increasing the number of iterations per trajectory and "parallel" test-time compute by increasing the number of trajectories per problem. With parallel scaling, we can amortize up-front costs across multiple downstream samples, allowing us to identify relevant codebase context using the simple method of letting an LLM read every file. In order to select between candidate edits, we combine voting using model-generated tests with a final multi-turn trajectory dedicated to selection. Overall, CodeMonkeys resolves 57.4% of issues from SWE-bench Verified using a budget of approximately 2300 USD. Our selection method can also be used to combine candidates from different sources. Selecting over an ensemble of edits from existing top SWE-bench Verified submissions obtains a score of 66.2% and outperforms the best member of the ensemble on its own. We fully release our code and data at https://scalingintelligence.stanford.edu/pubs/codemonkeys.

The tidal wave of advancements in Large Language Models (LLMs) has led to their swift integration into application-level logic. Many software systems now use prompts to interact with these black-box models, combining natural language with dynamic values interpolated at runtime, to perform tasks ranging from sentiment analysis to question answering. Due to the programmatic and structured natural language aspects of these prompts, we refer to them as Developer Prompts. Unlike traditional software artifacts, Dev Prompts blend natural language instructions with artificial languages such as programming and markup languages, thus requiring specialized tools for analysis, distinct from classical software evaluation methods. In response to this need, we introduce PromptDoctor, a tool explicitly designed to detect and correct issues of Dev Prompts. PromptDoctor identifies and addresses problems related to bias, vulnerability, and sub-optimal performance in Dev Prompts, helping mitigate their possible harms. In our analysis of 2,173 Dev Prompts, selected as a representative sample of 40,573 Dev Prompts, we found that 3.46% contained one or more forms of bias, 10.75% were vulnerable to prompt injection attacks. Additionally, 3,310 were amenable to automated prompt optimization. To address these issues, we applied PromptDoctor to the flawed Dev Prompts we discovered. PromptDoctor de-biased 68.29% of the biased Dev Prompts, hardened 41.81% of the vulnerable Dev Prompts, and improved the performance of 37.1% sub-optimal Dev Prompts. Finally, we developed a PromptDoctor VSCode extension, enabling developers to easily enhance Dev Prompts in their existing development workflows. The data and source code for this work are available at

We study the entanglement dynamics of a family of quantum collision models by analytically solving the pairwise concurrence for all qubit pairs. We introduce a diagrammatic method that offers an intuitive, frame-by-frame understanding of these dynamics. This allows us to monitor how a single collision affects the entanglement of the whole many-body system in some special cases. We focus on a class of models where the square of concurrence is a conserved quantity in the qubit collisions, aiding us to formulate general rules of entanglement propagation. In particular, among the multiple examples we will be showing, we identify a type of genuine multipartite entanglement, which we refer to as \textit{entanglement quilt}, where every qubit is entangled with every other qubit. We find that in some models, an entanglement quilt is hypersensitive to local excitation fluctuations: The presence of even a single excited qubit can destroy the entanglement quilts. We offer a detailed mathematical treatment on the phenomena, which can help us understand the disappearance of long-range entanglement in condensed matter systems above zero temperature. We also speculate about a possible property of the entanglement quilt: Every subsystem of it is entangled with every other subsystem., Comment: 14+7 pages, 10 figures

LLMs are tuned to follow instructions (aligned) by learning which of two outputs users prefer for a prompt. However, this preference data format does not convey why users prefer responses that are chosen or rejected, so LLMs trained on these datasets cannot tailor responses to varied user needs. To surface these parameters of personalization, we apply abductive reasoning to preference data, inferring needs and interests of users, i.e. personas, that may prefer each output. We test this idea in two steps: Persona Inference (PI)-abductively inferring personas of users who prefer chosen or rejected outputs-and Persona Tailoring (PT)-training models to tailor responses to personas from PI. We find: 1) LLMs infer personas accurately explaining why different users may prefer both chosen or rejected outputs; 2) Training on preference data augmented with PI personas via PT boosts personalization, enabling models to support user-written personas; and 3) Rejected response personas form harder personalization evaluations, showing PT better aids users with uncommon preferences versus typical alignment methods. We argue for an abductive view of preferences for personalization, asking not only which response is better but when, why, and for whom., Comment: In Progress Preprint

Purpose. This paper explores the capability of smartphones as computing devices for a quadcopter, specifically in terms of the ability of drones to maintain a position known as the position hold function. Image processing can be performed with the phone's sensors and powerful built-in camera. Method. Using Shi-Tomasi corner detection and the Lucas-Kanade sparse optical flow algorithms, ground features are recognized and tracked using the downward-facing camera. The position is maintained by computing quadcopter displacement from the center of the image using Euclidian distance, and the corresponding pitch and roll estimate is calculated using the PID controller. Results. Actual flights show a double standard deviation of 18.66 cm from the center for outdoor tests. With a quadcopter size of 58cm x 58cm used, it implies that 95% of the time, the quadcopter is within a diameter of 96 cm. For indoor tests, a double standard deviation of 10.55 cm means that 95% of the time, the quadcopter is within a diameter of 79 cm. Conclusion. Smartphone sensors and cameras can be used to perform optical flow position hold functions, proving their potential as computing devices for drones. Recommendations. To further improve the positioning system of the phone-based quadcopter system, it is suggested that potential sensor fusion be explored with the phone's GNSS sensor, which gives absolute positioning information for outdoor applications. Research Implications. As different devices and gadgets are integrated into the smartphone, this paper presents an opportunity for phone manufacturers and researchers to explore the potential of smartphones for a drone use-case., Comment: 13 pages

Autonomous surgical robots are a promising solution to the increasing demand for surgery amid a shortage of surgeons. Recent work has proposed learning-based approaches for the autonomous manipulation of soft tissue. However, due to variability in tissue geometries and stiffnesses, these methods do not always perform optimally, especially in out-of-distribution settings. We propose, develop, and test the first application of uncertainty quantification to learned surgical soft-tissue manipulation policies as an early identification system for task failures. We analyze two different methods of uncertainty quantification, deep ensembles and Monte Carlo dropout, and find that deep ensembles provide a stronger signal of future task success or failure. We validate our approach using the physical daVinci Research Kit (dVRK) surgical robot to perform physical soft-tissue manipulation. We show that we are able to successfully detect task failure and request human intervention when necessary while still enabling autonomous manipulation when possible. Our learned tissue manipulation policy with uncertainty-based early failure detection achieves a zero-shot sim2real performance improvement of 47.5% over the prior state of the art in learned soft-tissue manipulation. We also show that our method generalizes well to new types of tissue as well as to a bimanual soft tissue manipulation task., Comment: 8 pages, 6 figures

To take the esports scene to the next level, we introduce PandaSkill, a framework for assessing player performance and skill rating. Traditional rating systems like Elo and TrueSkill often overlook individual contributions and face challenges in professional esports due to limited game data and fragmented competitive scenes. PandaSkill leverages machine learning to estimate in-game player performance from individual player statistics. Each in-game role is modeled independently, ensuring a fair comparison between them. Then, using these performance scores, PandaSkill updates the player skill ratings using the Bayesian framework OpenSkill in a free-for-all setting. In this setting, skill ratings are updated solely based on performance scores rather than game outcomes, hightlighting individual contributions. To address the challenge of isolated rating pools that hinder cross-regional comparisons, PandaSkill introduces a dual-rating system that combines players' regional ratings with a meta-rating representing each region's overall skill level. Applying PandaSkill to five years of professional League of Legends matches worldwide, we show that our method produces skill ratings that better predict game outcomes and align more closely with expert opinions compared to existing methods.

Standard conformal prediction offers a marginal guarantee on coverage, but for prediction sets to be truly useful, they should ideally ensure coverage conditional on each test point. Unfortunately, it is impossible to achieve exact, distribution-free conditional coverage in finite samples. In this work, we propose an alternative conformal prediction algorithm that targets coverage where it matters most--in instances where a classifier is overconfident in its incorrect predictions. We start by dissecting miscoverage events in marginally-valid conformal prediction, and show that miscoverage rates vary based on the classifier's confidence and its deviation from the Bayes optimal classifier. Motivated by this insight, we develop a variant of conformal prediction that targets coverage conditional on a reduced set of two variables: the classifier's confidence in a prediction and a nonparametric trust score that measures its deviation from the Bayes classifier. Empirical evaluation on multiple image datasets shows that our method generally improves conditional coverage properties compared to standard conformal prediction, including class-conditional coverage, coverage over arbitrary subgroups, and coverage over demographic groups.

With the number of people with disabilities (PWD) increasing worldwide each year, the demand for mobility support to enable independent living and social integration is also growing. Wheelchairs commonly support the mobility of PWD in both indoor and outdoor environments. However, current powered wheelchairs (PWC) often fail to meet the needs of PWD, who may find it difficult to operate them. Furthermore, existing research on robotic wheelchairs typically focuses either on full autonomy or enhanced manual control, which can lead to reduced efficiency and user trust. To address these issues, this paper proposes a Robot Operating System (ROS)-based smart wheelchair, called CoNav Chair, that incorporates a shared control navigation algorithm and obstacle avoidance to support PWD while fostering efficiency and trust between the robot and the user. Our design consists of hardware and software components. Experimental results conducted in a typical indoor social environment demonstrate the performance and effectiveness of the smart wheelchair hardware and software design. This integrated design promotes trust and autonomy, which are crucial for the acceptance of assistive mobility technologies in the built environment., Comment: 8 pages, 9 figures

In this article we investigate the question of chromatic purity of L-theory. To do so, we utilize the theory of additive GW and L-theory in the language of Poincar\'e categories as laid out in the series of papers by Calm\`es et al. We apply this theory to chromatically localised L-theory at the prime $p=2$ and recover the L-theoretic analogues of chromatic purity for $E_1$-rings with involution. From this, we deduce that L-theory does not exhibit chromatic redshift. We deduce the higher chromatic vanishing of quadratic L-theory of arbitrary idempotent complete categories, thereby allowing the use of Hermitian trace methods to probe chromatic behaviour of GW and L-theory. Finally, we show that for $T(n+1)$-acyclic rings with involution, $T(n+1)$-local GW-theory depends only on $T(n+1)$-local K-theory and the associated duality, thereby proving a chromatic analogue of the homotopy limit problem for GW-theory., Comment: First Version, 20 pages, Comments Welcome

We collect novel data in the public service domain to evaluate the capability of the state-of-the-art automatic speech recognition (ASR) models in capturing regional differences in accents in the United Kingdom (UK), specifically focusing on two accents from Scotland with distinct dialects. This study addresses real-world problems where biased ASR models can lead to miscommunication in public services, disadvantaging individuals with regional accents particularly those in vulnerable populations. We first examine the out-of-the-box performance of the Whisper large-v3 model on a baseline dataset and our data. We then explore the impact of fine-tuning Whisper on the performance in the two UK regions and investigate the effectiveness of existing model evaluation techniques for our real-world application through manual inspection of model errors. We observe that the Whisper model has a higher word error rate (WER) on our test datasets compared to the baseline data and fine-tuning on a given data improves performance on the test dataset with the same domain and accent. The fine-tuned models also appear to show improved performance when applied to the test data outside of the region it was trained on suggesting that fine-tuned models may be transferable within parts of the UK. Our manual analysis of model outputs reveals the benefits and drawbacks of using WER as an evaluation metric and fine-tuning to adapt to regional dialects.

We present a new perspective on online learning that we refer to as gradient equilibrium: a sequence of iterates achieves gradient equilibrium if the average of gradients of losses along the sequence converges to zero. In general, this condition is not implied by nor implies sublinear regret. It turns out that gradient equilibrium is achievable by standard online learning methods such as gradient descent and mirror descent with constant step sizes (rather than decaying step sizes, as is usually required for no regret). Further, as we show through examples, gradient equilibrium translates into an interpretable and meaningful property in online prediction problems spanning regression, classification, quantile estimation, and others. Notably, we show that the gradient equilibrium framework can be used to develop a debiasing scheme for black-box predictions under arbitrary distribution shift, based on simple post hoc online descent updates. We also show that post hoc gradient updates can be used to calibrate predicted quantiles under distribution shift, and that the framework leads to unbiased Elo scores for pairwise preference prediction., Comment: Code available at https://github.com/aangelopoulos/gradient-equilibrium/

3D semantic scene completion is critical for multiple downstream tasks in autonomous systems. It estimates missing geometric and semantic information in the acquired scene data. Due to the challenging real-world conditions, this task usually demands complex models that process multi-modal data to achieve acceptable performance. We propose a unique neural model, leveraging advances from the state space and diffusion generative modeling to achieve remarkable 3D semantic scene completion performance with monocular image input. Our technique processes the data in the conditioned latent space of a variational autoencoder where diffusion modeling is carried out with an innovative state space technique. A key component of our neural network is the proposed Skimba (Skip Mamba) denoiser, which is adept at efficiently processing long-sequence data. The Skimba diffusion model is integral to our 3D scene completion network, incorporating a triple Mamba structure, dimensional decomposition residuals and varying dilations along three directions. We also adopt a variant of this network for the subsequent semantic segmentation stage of our method. Extensive evaluation on the standard SemanticKITTI and SSCBench-KITTI360 datasets show that our approach not only outperforms other monocular techniques by a large margin, it also achieves competitive performance against stereo methods. The code is available at https://github.com/xrkong/skimba, Comment: Accepted by AAAI 2025

Detection of the weak cosmological signal from high-redshift hydrogen demands careful data analysis and an understanding of the full instrument signal chain. Here we use the WODEN simulation pipeline to produce realistic data from the Murchison Widefield Array Epoch of Reionisation experiment, and test the effects of different instrumental systematics through the AusEoRPipe analysis pipeline. The simulations include a realistic full sky model, direction-independent calibration, and both random and systematic instrumental effects. Results are compared to matched real observations. We find that, (i) with a sky-based calibration and power spectrum approach we have need to subtract more than 90% of all unresolved point source flux (10mJy apparent) to recover 21-cm signal in the absence of instrumental effects; (ii) when including diffuse emission in simulations, some k-modes cannot be accessed, leading to a need for some diffuse emission removal; (iii) the single greatest cause of leakage is an incomplete sky model; (iv) other sources of errors, such as cable reflections, flagged channels and gain errors, impart comparable systematic power to one another, and less power than the incomplete skymodel., Comment: 13 pages, 11 figures, 1 table Accepted for publication to PASA

We develop a perfect foresight method to solve models with an interest rate lower bound constraint that nests OccBin/DynareOBC and \cite{Eggertsson2010}'s as well as \cite{Mertens2014}'s pen and paper solutions as special cases. Our method generalizes the pen-and-paper solutions by allowing for endogenous persistence while maintaining tractability and interpretability. We prove that our method necessarily gives stable multipliers. We use it to solve a New Keynesian model with habit formation and government spending, which we match to expectations data from the Great Recession. We find an output multiplier of government spending close to 1 for the US and Japan.

This study demonstrates the effectiveness of AFE-PUND, a revisited Positive Up Negative Down (PUND) protocol for characterizing antiferroelectric (AFE) materials, in analyzing $ZrO_2$ films across different thicknesses, revealing key trends. The proposed AFE-PUND method enables the isolation of switching currents from non-switching contributions, allowing precise extraction of remanent polarization and coercive field from hysteresis loops. The remanent polarization increases with film thickness, reflecting enhanced domain stability, while endurance cycles highlight the wake-up effect and its eventual degradation due to fatigue in thicker films. Similarly, coercive fields decrease with thickness, indicating reduced switching barriers and a clearer transition between tetragonal and orthorhombic phases. The method provides valuable insights into micro-structural influences, such as defect accumulation, grain size, and domain wall pinning, which critically affect device performance. AFE-PUND thus establishes itself as an essential tool for advancing the understanding and optimization of antiferroelectric materials.

Given a rigidly-compactly generated tensor-triangulated category whose Balmer spectrum is finite dimensional and Noetherian, we construct a torsion model for it, which is equivalent to the original tensor-triangulated category. The torsion model is determined in an adelic fashion by objects with singleton supports. This categorifies the Cousin complex from algebra, and the process of reconstructing a spectrum from its monochromatic layers in chromatic stable homotopy theory. This model is inspired by work of the second author in rational equivariant stable homotopy theory, and extends previous work of the authors from the one-dimensional setting., Comment: 48pp

Despite being the most common types of stars in the Galaxy, the physical properties of late M dwarfs are often poorly constrained. A trend of radius inflation compared to evolutionary models has been observed for earlier type M dwarfs in eclipsing binaries, possibly caused by magnetic activity. It is currently unclear whether this trend also extends to later type M dwarfs below the convective boundary. This makes the discovery of lower-mass, fully convective, M dwarfs in eclipsing binaries valuable for testing evolutionary models especially in longer-period binaries where tidal interaction between the primary and secondary is negligible. With this context, we present the discovery of the NGTS-EB-7 AB system, an eclipsing binary containing a late M dwarf secondary and an evolved G-type primary star. The secondary star has a radius of $0.125 \pm 0.006 R_\odot$ , a mass of $0.096 \pm 0.004 M_\odot$ and follows a highly eccentric $(e=0.71436 \pm 0.00085)$ orbit every $193.35875 \pm 0.00034$ days. This makes NGTS-EB-7 AB the third longest-period eclipsing binary system with a secondary smaller than $200 M_J$ with the mass and radius constrained to better than $5 \%$. In addition, NGTS-EB-7 is situated near the centre of the proposed LOPS2 southern field of the upcoming PLATO mission, allowing for detection of the secondary eclipse and measurement of the companion`s temperature. With its long-period and well-constrained physical properties - NGTS-EB-7 B will make a valuable addition to the sample of M dwarfs in eclipsing binaries and help in determining accurate empirical mass/radius relations for later M dwarf stars., Comment: Main body: 14 pages, 6 figures, 5 tables. Appendices: 7 pages, 5 figures, 5 tables

The advancement of machine learning and the availability of large-scale reaction datasets have accelerated the development of data-driven models for computer-aided synthesis planning (CASP) in the past decade. Here, we detail the newest version of ASKCOS, an open source software suite for synthesis planning that makes available several research advances in a freely available, practical tool. Four one-step retrosynthesis models form the basis of both interactive planning and automatic planning modes. Retrosynthetic planning is complemented by other modules for feasibility assessment and pathway evaluation, including reaction condition recommendation, reaction outcome prediction, and auxiliary capabilities such as solubility prediction and quantum mechanical descriptor prediction. ASKCOS has assisted hundreds of medicinal, synthetic, and process chemists in their day-to-day tasks, complementing expert decision making. It is our belief that CASP tools like ASKCOS are an important part of modern chemistry research, and that they offer ever-increasing utility and accessibility.

We present OLMo 2, the next generation of our fully open language models. OLMo 2 includes dense autoregressive models with improved architecture and training recipe, pretraining data mixtures, and instruction tuning recipes. Our modified model architecture and training recipe achieve both better training stability and improved per-token efficiency. Our updated pretraining data mixture introduces a new, specialized data mix called Dolmino Mix 1124, which significantly improves model capabilities across many downstream task benchmarks when introduced via late-stage curriculum training (i.e. specialized data during the annealing phase of pretraining). Finally, we incorporate best practices from T\"ulu 3 to develop OLMo 2-Instruct, focusing on permissive data and extending our final-stage reinforcement learning with verifiable rewards (RLVR). Our OLMo 2 base models sit at the Pareto frontier of performance to compute, often matching or outperforming open-weight only models like Llama 3.1 and Qwen 2.5 while using fewer FLOPs and with fully transparent training data, code, and recipe. Our fully open OLMo 2-Instruct models are competitive with or surpassing open-weight only models of comparable size, including Qwen 2.5, Llama 3.1 and Gemma 2. We release all OLMo 2 artifacts openly -- models at 7B and 13B scales, both pretrained and post-trained, including their full training data, training code and recipes, training logs and thousands of intermediate checkpoints. The final instruction model is available on the Ai2 Playground as a free research demo., Comment: Model demo available at playground.allenai.org

Introduction: Healthcare AI models often inherit biases from their training data. While efforts have primarily targeted bias in structured data, mental health heavily depends on unstructured data. This study aims to detect and mitigate linguistic differences related to non-biological differences in the training data of AI models designed to assist in pediatric mental health screening. Our objectives are: (1) to assess the presence of bias by evaluating outcome parity across sex subgroups, (2) to identify bias sources through textual distribution analysis, and (3) to develop a de-biasing method for mental health text data. Methods: We examined classification parity across demographic groups and assessed how gendered language influences model predictions. A data-centric de-biasing method was applied, focusing on neutralizing biased terms while retaining salient clinical information. This methodology was tested on a model for automatic anxiety detection in pediatric patients. Results: Our findings revealed a systematic under-diagnosis of female adolescent patients, with a 4% lower accuracy and a 9% higher False Negative Rate (FNR) compared to male patients, likely due to disparities in information density and linguistic differences in patient notes. Notes for male patients were on average 500 words longer, and linguistic similarity metrics indicated distinct word distributions between genders. Implementing our de-biasing approach reduced diagnostic bias by up to 27%, demonstrating its effectiveness in enhancing equity across demographic groups. Discussion: We developed a data-centric de-biasing framework to address gender-based content disparities within clinical text. By neutralizing biased language and enhancing focus on clinically essential information, our approach demonstrates an effective strategy for mitigating bias in AI healthcare models trained on text.

Vacancy-cluster color centers in lithium fluoride have been studied in detail both theoretically and experimentally for over a century, giving rise to various applications in solid-state lasers, broadband photonic devices, and radiation dosimeters. These color centers are also attractive candidate platforms for applications in quantum information science, due to their spin properties and strong coupling to the crystal lattice, which allows their properties to be easily tuned. Here we present hybrid functional calculations of common vacancy defects in lithium fluoride, including their energetic, spin, and optical properties. We show that for a wide range of hybrid functional parameters tuned to match the experimental band gap, certain defects have little variation in their predicted optical properties. We further demonstrate that the parameters needed to satisfy the generalized Koopman's theorem and correctly position defect levels within the gap, can vary dramatically, even for different charge states of the same defect. Our work establishes the accuracy of the computationally lightweight hybrid-functional approach for predicting the optical and energetic properties of color centers in polar materials., Comment: 10 pages, 5 figures

Human acceptance of social robots is greatly effected by empathy and perceived understanding. This necessitates accurate and flexible responses to various input data from the user. While systems such as this can become increasingly complex as more states or response types are included, new research in the application of large language models towards human-robot interaction has allowed for more streamlined perception and reaction pipelines. LLM-selected actions and emotional expressions can help reinforce the realism of displayed empathy and allow for improved communication between the robot and user. Beyond portraying empathy in spoken or written responses, this shows the possibilities of using LLMs in actuated, real world scenarios. In this work we extend research in LLM-driven nonverbal behavior for social robots by considering more open-ended emotional response selection leveraging new advances in vision-language models, along with emotionally aligned motion and color pattern selections that strengthen conveyance of meaning and empathy., Comment: This work was presented at the WARN, Weighing the Benefits of Autonomous Robot Personalization, workshop at the 33rd IEEE RO-MAN 2024 conference

Strong anomalous diffusion is {often} characterized by a piecewise-linear spectrum of the moments of displacement. The spectrum is characterized by slopes $\xi$ and $\zeta$ for small and large moments, respectively, and by the critical moment $\alpha$ of the crossover. The exponents $\xi$ and $\zeta$ characterize the asymptotic scaling of the bulk and the tails of the probability distribution function of displacements, respectively. Here, we adopt asymptotic theory to match the behaviors at intermediate scales. The resulting constraint explains how distributions with algebraic tails imply strong anomalous diffusion, and it relates $\alpha$ to the corresponding power law. Our theory provides novel relations between exponents characterizing strong anomalous diffusion, and it yields explicit expressions for the leading-order corrections to the asymptotic power-law behavior of the moments of displacement. They provide the time scale that must be surpassed to clearly discriminate the leading-order power law from its sub-leading corrections. This insight allows us to point out sources of systematic errors in their numerical estimates. Rather than separately fitting an exponent for each moment we devise a robust scheme to determine $\xi$, $\zeta$ and $\alpha$. The findings are supported by numerical and analytical results on five different models exhibiting strong anomalous diffusion.

Geological materials are often seen as the antithesis of soft; rocks are hard. However, during the formation of minerals and rocks, all the systems we shall discuss, indeed geological materials in general, pass through a stage where they are soft. This occurs either because they are at a high temperature -- igneous or metamorphic rock -- or because they are at a lower temperature but in the presence of water -- sedimentary rock. For this reason it is useful to introduce soft-matter concepts into the geological domain. There is a universality in the diverse instances of geological patterns that may be appreciated by looking at the common aspect in their formation of having passed through a stage as soft matter.

The EU AI Act (AIA) mandates the implementation of a risk management system (RMS) and a quality management system (QMS) for high-risk AI systems. The ISO/IEC 42001 standard provides a foundation for fulfilling these requirements but does not cover all EU-specific regulatory stipulations. To enhance the implementation of the AIA in Germany, the Federal Office for Information Security (BSI) could introduce the national standard BSI 200-5, which specifies AIA requirements and integrates existing ISMS standards, such as ISO/IEC 27001. This paper examines the interfaces between an information security management system (ISMS) and an AI management system (AIMS), demonstrating that incorporating existing ISMS controls with specific AI extensions presents an effective strategy for complying with Article 15 of the AIA. Four new AI modules are introduced, proposed for inclusion in the BSI IT Grundschutz framework to comprehensively ensure the security of AI systems. Additionally, an approach for adapting BSI's qualification and certification systems is outlined to ensure that expertise in secure AI handling is continuously developed. Finally, the paper discusses how the BSI could bridge international standards and the specific requirements of the AIA through the nationalization of ISO/IEC 42001, creating synergies and bolstering the competitiveness of the German AI landscape.

Recent advancements have underscored the potential of large language model (LLM)-based agents in financial decision-making. Despite this progress, the field currently encounters two main challenges: (1) the lack of a comprehensive LLM agent framework adaptable to a variety of financial tasks, and (2) the absence of standardized benchmarks and consistent datasets for assessing agent performance. To tackle these issues, we introduce \textsc{InvestorBench}, the first benchmark specifically designed for evaluating LLM-based agents in diverse financial decision-making contexts. InvestorBench enhances the versatility of LLM-enabled agents by providing a comprehensive suite of tasks applicable to different financial products, including single equities like stocks, cryptocurrencies and exchange-traded funds (ETFs). Additionally, we assess the reasoning and decision-making capabilities of our agent framework using thirteen different LLMs as backbone models, across various market environments and tasks. Furthermore, we have curated a diverse collection of open-source, multi-modal datasets and developed a comprehensive suite of environments for financial decision-making. This establishes a highly accessible platform for evaluating financial agents' performance across various scenarios.

Persistence effects in HgCdTe infrared detectors cause significant artifacts that can impact the quality of science observations for up to many hours after exposure to bright/saturating sources. This problem will have a substantially greater impact on viable observing modes for infrared cameras on future ELTs due to the leap in sensitivities that is expected. In this paper we present new results from an updated test system that was previously used to prototype ``on-detector guide windows'' to provide fast T/T feedback to AO systems, interleaved with simultaneous (slow) full-frame readouts for science. We now explore the possibility of continuously resetting these small regions of the detector that are illuminated with a compact source as a strategy for mitigating persistence, using two different detectors. While our results generally show promise for this observing strategy, we found for one of our detectors that the combination of fast localized resets with intense illumination can introduce a potentially problematic persistent change in local reset levels., Comment: 12 pages, 9 figures, SPIE (2024) 13103-99

To optimize interventions for improving wellness, it is essential to understand habits, which wearable devices can measure with greater precision. Using high temporal resolution biometric data taken from the Oura Gen3 ring, we examine daily and weekly sleep and activity patterns of a cohort of young adults (N=582) in their first semester of college. A high compliance rate is observed for both daily and nightly wear, with slight dips in wear compliance observed shortly after waking up and also in the evening. Most students have a late-night chronotype with a median midpoint of sleep at 5AM, with males and those with mental health impairment having more delayed sleep periods. Social jetlag, or the difference in sleep times between free days and school days, is prevalent in our sample. While sleep periods generally shift earlier on weekdays and later on weekends, sleep duration on both weekdays and weekends is shorter than during prolonged school breaks, suggesting chronic sleep debt when school is in session. Synchronized spikes in activity consistent with class schedules are also observed, suggesting that walking in between classes is a widespread behavior in our sample that substantially contributes to physical activity. Lower active calorie expenditure is associated with weekends and a delayed but longer sleep period the night before, suggesting that for our cohort, active calorie expenditure is affected less by deviations from natural circadian rhythms and more by the timing associated with activities. Our study shows that regular sleep and activity routines may be inferred from consumer wearable devices if high temporal resolution and long data collection periods are available.

The breakup of a capillary jet into drops is believed to be governed by initial disturbances on the surface of the jet that grow exponentially. The disturbances are often assumed to be due to external sources of noise, to turbulence, or to imperfections of the nozzle. However, even in conditions where external perturbations are minimal, the jet's length cannot grow indefinitely, suggesting that its fragmentation cannot be entirely attributed to such factors. Here we show that the initial disturbances are thermal capillary waves. By extrapolating the observed growth of the instability back in time, we demonstrate that the initiating disturbances must be of the order of an {\aa}ngstr\"om, consistent with fluctuations induced by thermal noise. Further, by performing many experiments with different nozzles, we find no significant variation in breakup length linked to nozzle type, shape or inner roughness. By systematically varying the jet diameter and velocity, and the surface tension, viscosity, and density of the fluid, we validate our thermal disturbance model over four orders of magnitude in jet length; seven orders of magnitude if simulations of nanojets are included., Comment: 6 pages, 5 figures

This paper presents a novel deep learning framework for estimating multivariate joint extremes of metocean variables, based on the Semi-Parametric Angular-Radial (SPAR) model. When considered in polar coordinates, the problem of modelling multivariate extremes is transformed to one of modelling an angular density, and the tail of a univariate radial variable conditioned on angle. In the SPAR approach, the tail of the radial variable is modelled using a generalised Pareto (GP) distribution, providing a natural extension of univariate extreme value theory to the multivariate setting. In this work, we show how the method can be applied in higher dimensions, using a case study for five metocean variables: wind speed, wind direction, wave height, wave period and wave direction. The angular variable is modelled empirically, while the parameters of the GP model are approximated using fully-connected deep neural networks. Our data-driven approach provides great flexibility in the dependence structures that can be represented, together with computationally efficient routines for training the model. Furthermore, the application of the method requires fewer assumptions about the underlying distribution(s) compared to existing approaches, and an asymptotically justified means for extrapolating outside the range of observations. Using various diagnostic plots, we show that the fitted models provide a good description of the joint extremes of the metocean variables considered.

We present a structural analysis of 521 galaxy candidates at 6.5 < z < 12.5, with $SNR > 10\sigma$ in the F444W filter, taken from the EPOCHS v1 sample, consisting of uniformly reduced deep JWST NIRCam data, covering the CEERS, JADES GOOD-S, NGDEEP, SMACS0723, GLASS and PEARLS surveys. We use standard software to fit single S\'ersic models to each galaxy in the rest-frame optical and extract their parametric structural parameters (S\'ersic index, half-light radius and axis-ratio), and \texttt{Morfometryka} to measure their non-parametric concentration and asymmetry parameters. We find a wide range of sizes for these early galaxies, but with a strong galaxy-size mass correlation up to $z \sim 12$ such that galaxy sizes continue to get progressively smaller in the high-redshift regime, following $R_{e} = 2.74 \pm 0.49 \left( 1 + z \right) ^{-0.79 \pm 0.08}$ kpc. Using non-parametric methods we find that galaxy merger fractions, classified through asymmetry parameters, at these redshifts remain consistent with those in literature, maintaining a value of $f_{m} \sim 0.12 \pm 0.07$ showing little dependence with redshift when combined with literature at $z > 4$. We find that galaxies which are smaller in size also appear rounder, with an excess of high axis-ratio objects. Finally, we artificially redshift a subsample of our objects to determine how robust the observational trends we see are, determining that observed trends are due to real evolutionary effects, rather than being a consequence of redshift effects., Comment: 35 pages, 13 figures. Submitted to ApJ on 19/12/2024. Comments to corresponding author welcome at lewi.westcott@manchester.ac.uk