Search

This paper claims that machine learning (ML) largely overlooks an important facet of general intelligence: robustness to a qualitatively unknown future in an open world. Such robustness relates to Knightian uncertainty (KU) in economics, i.e. uncertainty that cannot be quantified, which is excluded from consideration in ML's key formalisms. This paper aims to identify this blind spot, argue its importance, and catalyze research into addressing it, which we believe is necessary to create truly robust open-world AI. To help illuminate the blind spot, we contrast one area of ML, reinforcement learning (RL), with the process of biological evolution. Despite staggering ongoing progress, RL still struggles in open-world situations, often failing under unforeseen situations. For example, the idea of zero-shot transferring a self-driving car policy trained only in the US to the UK currently seems exceedingly ambitious. In dramatic contrast, biological evolution routinely produces agents that thrive within an open world, sometimes even to situations that are remarkably out-of-distribution (e.g. invasive species; or humans, who do undertake such zero-shot international driving). Interestingly, evolution achieves such robustness without explicit theory, formalisms, or mathematical gradients. We explore the assumptions underlying RL's typical formalisms, showing how they limit RL's engagement with the unknown unknowns characteristic of an ever-changing complex world. Further, we identify mechanisms through which evolutionary processes foster robustness to novel and unpredictable challenges, and discuss potential pathways to algorithmically embody them. The conclusion is that the intriguing remaining fragility of ML may result from blind spots in its formalisms, and that significant gains may result from direct confrontation with the challenge of KU.

As Large Language Models (LLMs) are integrated into electronic health record (EHR) workflows, validated instruments are essential to evaluate their performance before implementation. Existing instruments for provider documentation quality are often unsuitable for the complexities of LLM-generated text and lack validation on real-world data. The Provider Documentation Summarization Quality Instrument (PDSQI-9) was developed to evaluate LLM-generated clinical summaries. Multi-document summaries were generated from real-world EHR data across multiple specialties using several LLMs (GPT-4o, Mixtral 8x7b, and Llama 3-8b). Validation included Pearson correlation for substantive validity, factor analysis and Cronbach's alpha for structural validity, inter-rater reliability (ICC and Krippendorff's alpha) for generalizability, a semi-Delphi process for content validity, and comparisons of high-versus low-quality summaries for discriminant validity. Seven physician raters evaluated 779 summaries and answered 8,329 questions, achieving over 80% power for inter-rater reliability. The PDSQI-9 demonstrated strong internal consistency (Cronbach's alpha = 0.879; 95% CI: 0.867-0.891) and high inter-rater reliability (ICC = 0.867; 95% CI: 0.867-0.868), supporting structural validity and generalizability. Factor analysis identified a 4-factor model explaining 58% of the variance, representing organization, clarity, accuracy, and utility. Substantive validity was supported by correlations between note length and scores for Succinct (rho = -0.200, p = 0.029) and Organized ($\rho = -0.190$, $p = 0.037$). Discriminant validity distinguished high- from low-quality summaries ($p < 0.001$). The PDSQI-9 demonstrates robust construct validity, supporting its use in clinical practice to evaluate LLM-generated summaries and facilitate safer integration of LLMs into healthcare workflows.

Let $\mathbb{T}$ be the differential field of logarithmic-exponential transseries. We consider the expansion of $\mathbb{T}$ by the binary map that sends a real number $r$ and a positive transseries $f$ to the transseries $f^r$. Building on recent work of Aschenbrenner, van den Dries, and van der Hoeven, we show that this expansion is model complete, and we give an axiomatization of the theory of this expansion that is effective relative to the theory of the real exponential field. We show that maximal Hardy fields, equipped with the same map $(f,r)\mapsto f^r$, enjoy the same theory as $\mathbb{T}$, and we use this to establish a transfer theorem between Hardy fields and transseries., Comment: 26 pages

Promptable segmentation foundation models have emerged as a transformative approach to addressing the diverse needs in medical images, but most existing models require expensive computing, posing a big barrier to their adoption in clinical practice. In this work, we organized the first international competition dedicated to promptable medical image segmentation, featuring a large-scale dataset spanning nine common imaging modalities from over 20 different institutions. The top teams developed lightweight segmentation foundation models and implemented an efficient inference pipeline that substantially reduced computational requirements while maintaining state-of-the-art segmentation accuracy. Moreover, the post-challenge phase advanced the algorithms through the design of performance booster and reproducibility tasks, resulting in improved algorithms and validated reproducibility of the winning solution. Furthermore, the best-performing algorithms have been incorporated into the open-source software with a user-friendly interface to facilitate clinical adoption. The data and code are publicly available to foster the further development of medical image segmentation foundation models and pave the way for impactful real-world applications., Comment: CVPR 2024 MedSAM on Laptop Competition Summary: https://www.codabench.org/competitions/1847/

The 7.3 GHz (350 MHz bandwidth) Earth Observation Satellite (EOS) band, while not protected, is used for Passive Sea Surface Temperature (P-SST) measurements that provide important data for weather forecasts, coastal disaster prevention, climate modeling, and oceanographic research. The full 7 GHz band (7.125 to 8.4 GHz), which encompasses these EOS frequencies, is the largest contiguous block of potentially available mid-band spectrum and will play a significant role in meeting the anticipated demand for wireless services. A Real Time Geofenced Spectrum Sharing (RGSS) system is shown to be a practical and near term solution to spectrum sharing between P-SST measurements and 5G or 6G networks in the 7GHz band. RGSS enables IMT networks and EOS radiometers to share 350MHz of overlapping spectrum centered at 7.3 GHz. It prevents interference to P-SST measurements while simultaneously allowing IMT systems un-constrained access to the shared spectrum greater than 99.9% of the time. Subscriber impact during the less than 0.1 percent paused access time can be prevented by using 3GPP defined capabilities and O-RAN APIs to move subscribers to other frequencies. Paused access time data from a proof-of-concept RGSS system is available to academic, government, and industry researchers through a web or programmatic interface., Comment: Submitted to 2025 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN)

Utilizing machine learning techniques has always required choosing hyperparameters. This is true whether one uses a classical technique such as a KNN or very modern neural networks such as Deep Learning. Though in many applications, hyperparameters are chosen by hand, automated methods have become increasingly more common. These automated methods have become collectively known as automated machine learning, or AutoML. Several automated selection algorithms have shown similar or improved performance over state-of-the-art methods. This breakthrough has led to the development of cloud-based services like Google AutoML, which is based on Deep Learning and is widely considered to be the industry leader in AutoML services. Extreme Learning Machines (ELMs) use a fundamentally different type of neural architecture, producing better results at a significantly discounted computational cost. We benchmark the Extreme AutoML technology against Google's AutoML using several popular classification data sets from the University of California at Irvine's (UCI) repository, and several other data sets, observing significant advantages for Extreme AutoML in accuracy, Jaccard Indices, the variance of Jaccard Indices across classes (i.e. class variance) and training times., Comment: 8 pages, 7 figures

Generalized linear models are a popular tool in applied statistics, with their maximum likelihood estimators enjoying asymptotic Gaussianity and efficiency. As all models are wrong, it is desirable to understand these estimators' behaviours under model misspecification. We study semiparametric multilevel generalized linear models, where only the conditional mean of the response is taken to follow a specific parametric form. Pre-existing estimators from mixed effects models and generalized estimating equations require specificaiton of a conditional covariance, which when misspecified can result in inefficient estimates of fixed effects parameters. It is nevertheless often computationally attractive to consider a restricted, finite dimensional class of estimators, as these models naturally imply. We introduce sandwich regression, that selects the estimator of minimal variance within a parametric class of estimators over all distributions in the full semiparametric model. We demonstrate numerically on simulated and real data the attractive improvements our sandwich regression approach enjoys over classical mixed effects models and generalized estimating equations.

Tools for analyzing character portrayal in fiction are valuable for writers and literary scholars in developing and interpreting compelling stories. Existing tools, such as visualization tools for analyzing fictional characters, primarily rely on explicit textual indicators of character attributes. However, portrayal is often implicit, revealed through actions and behaviors rather than explicit statements. We address this gap by leveraging large language models (LLMs) to uncover implicit character portrayals. We start by generating a dataset for this task with greater cross-topic similarity, lexical diversity, and narrative lengths than existing narrative text corpora such as TinyStories and WritingPrompts. We then introduce LIIPA (LLMs for Inferring Implicit Portrayal for Character Analysis), a framework for prompting LLMs to uncover character portrayals. LIIPA can be configured to use various types of intermediate computation (character attribute word lists, chain-of-thought) to infer how fictional characters are portrayed in the source text. We find that LIIPA outperforms existing approaches, and is more robust to increasing character counts (number of unique persons depicted) due to its ability to utilize full narrative context. Lastly, we investigate the sensitivity of portrayal estimates to character demographics, identifying a fairness-accuracy tradeoff among methods in our LIIPA framework -- a phenomenon familiar within the algorithmic fairness literature. Despite this tradeoff, all LIIPA variants consistently outperform non-LLM baselines in both fairness and accuracy. Our work demonstrates the potential benefits of using LLMs to analyze complex characters and to better understand how implicit portrayal biases may manifest in narrative texts.

We propose to learn legged robot locomotion skills by watching thousands of wild animal videos from the internet, such as those featured in nature documentaries. Indeed, such videos offer a rich and diverse collection of plausible motion examples, which could inform how robots should move. To achieve this, we introduce Reinforcement Learning from Wild Animal Videos (RLWAV), a method to ground these motions into physical robots. We first train a video classifier on a large-scale animal video dataset to recognize actions from RGB clips of animals in their natural habitats. We then train a multi-skill policy to control a robot in a physics simulator, using the classification score of a third-person camera capturing videos of the robot's movements as a reward for reinforcement learning. Finally, we directly transfer the learned policy to a real quadruped Solo. Remarkably, despite the extreme gap in both domain and embodiment between animals in the wild and robots, our approach enables the policy to learn diverse skills such as walking, jumping, and keeping still, without relying on reference trajectories nor skill-specific rewards., Comment: Project website: https://elliotchanesane31.github.io/RLWAV/

Synthetic data generation with Large Language Models is a promising paradigm for augmenting natural data over a nearly infinite range of tasks. Given this variety, direct comparisons among synthetic data generation algorithms are scarce, making it difficult to understand where improvement comes from and what bottlenecks exist. We propose to evaluate algorithms via the makeup of synthetic data generated by each algorithm in terms of data quality, diversity, and complexity. We choose these three characteristics for their significance in open-ended processes and the impact each has on the capabilities of downstream models. We find quality to be essential for in-distribution model generalization, diversity to be essential for out-of-distribution generalization, and complexity to be beneficial for both. Further, we emphasize the existence of Quality-Diversity trade-offs in training data and the downstream effects on model performance. We then examine the effect of various components in the synthetic data pipeline on each data characteristic. This examination allows us to taxonomize and compare synthetic data generation algorithms through the components they utilize and the resulting effects on data QDC composition. This analysis extends into a discussion on the importance of balancing QDC in synthetic data for efficient reinforcement learning and self-improvement algorithms. Analogous to the QD trade-offs in training data, often there exist trade-offs between model output quality and output diversity which impact the composition of synthetic data. We observe that many models are currently evaluated and optimized only for output quality, thereby limiting output diversity and the potential for self-improvement. We argue that balancing these trade-offs is essential to the development of future self-improvement algorithms and highlight a number of works making progress in this direction.

Natural language syntax can serve as a major test for how to integrate two infamously distinct frameworks: symbolic representations and connectionist neural networks. Building on a recent neurocomputational architecture for syntax (ROSE), I discuss the prospects of reconciling the neural code for hierarchical 'vertical' syntax with linear and predictive 'horizontal' processes via a hybrid neurosymbolic model. I argue that the former can be accounted for via the higher levels of ROSE in terms of vertical phrase structure representations, while the latter can explain horizontal forms of linguistic information via the tuning of the lower levels to statistical and perceptual inferences. One prediction of this is that artificial language models will contribute to the cognitive neuroscience of horizontal morphosyntax, but much less so to hierarchically compositional structures. I claim that this perspective helps resolve many current tensions in the literature. Options for integrating these two neural codes are discussed, with particular emphasis on how predictive coding mechanisms can serve as interfaces between symbolic oscillatory phase codes and population codes for the statistics of linearized aspects of syntax. Lastly, I provide a neurosymbolic mathematical model for how to inject symbolic representations into a neural regime encoding lexico-semantic statistical features.

To accurately and confidently answer the question 'could an AI model or system increase biorisk', it is necessary to have both a sound theoretical threat model for how AI models or systems could increase biorisk and a robust method for testing that threat model. This paper provides an analysis of existing available research surrounding two AI and biorisk threat models: 1) access to information and planning via large language models (LLMs), and 2) the use of AI-enabled biological tools (BTs) in synthesizing novel biological artifacts. We find that existing studies around AI-related biorisk are nascent, often speculative in nature, or limited in terms of their methodological maturity and transparency. The available literature suggests that current LLMs and BTs do not pose an immediate risk, and more work is needed to develop rigorous approaches to understanding how future models could increase biorisks. We end with recommendations about how empirical work can be expanded to more precisely target biorisk and ensure rigor and validity of findings., Comment: Updated to correct author affiliations and clarify findings of evaluations of the o1 model

This paper is a continuation of our investigation into the Coulomb branches of twisted $A_{2n}$ of class-S. In arXiv:2411.17675, we found predictions for the contributions of twisted punctures to the graded dimensions of the Coulomb branch, based on the behaviour under nilpotent Higgsings and S-duality. While surprisingly powerful, these arguments were indirect. Here, we take a different approach: we define precisely the nature of the automorphism under which the twisted punctures are twisted (in particular, it is order-4, not order-2). From that, we find the local constraints satisfied by the Laurent coefficients of the invariant polynomials in the Higgs field, for all twisted punctures in $A_{2n}$, for all n. A crucial role is played by a new (at least, new in physics) order-reversing map on the set of nilpotent orbits in sp(n). Finally, we construct several examples of Seiberg-Witten curves for 3-punctured spheres in these theories.

In this paper, we investigate the twisted $A_{2n}$ sector of class-S theories. Heretofore, the Coulomb branches of such theories have been poorly understood. In this, and a companion paper, we make progress in our understanding of them. In particular, we find a formula for the dimension of the Coulomb branch of any twisted $A_{2n}$ class-S theory. Deferring a systematic analysis to the companion paper, we here determine many contributions of punctures to the graded Coulomb branch dimensions, and in some low rank cases, all of them. We are then able to identify a variety of known 4d $\mathcal{N}=2$ SCFTs with twisted $A_{2n}$ theories, and reproduce many of their known properties, such as S-duality amongst certain Argyres-Douglas theories., Comment: 18 pages + references

The Segment Anything Model (SAM) was originally designed for label-agnostic mask generation. Does this model also possess inherent semantic understanding, of value to broader visual tasks? In this work we follow a multi-staged approach towards exploring this question. We firstly quantify SAM's semantic capabilities by comparing base image encoder efficacy under classification tasks, in comparison with established models (CLIP and DINOv2). Our findings reveal a significant lack of semantic discriminability in SAM feature representations, limiting potential for tasks that require class differentiation. This initial result motivates our exploratory study that attempts to enable semantic information via in-context learning with lightweight fine-tuning where we observe that generalisability to unseen classes remains limited. Our observations culminate in the proposal of a training-free approach that leverages DINOv2 features, towards better endowing SAM with semantic understanding and achieving instance-level class differentiation through feature-based similarity. Our study suggests that incorporation of external semantic sources provides a promising direction for the enhancement of SAM's utility with respect to complex visual tasks that require semantic understanding., Comment: Preprint. Work in progress

Here, we present the outcomes from the second Large Language Model (LLM) Hackathon for Applications in Materials Science and Chemistry, which engaged participants across global hybrid locations, resulting in 34 team submissions. The submissions spanned seven key application areas and demonstrated the diverse utility of LLMs for applications in (1) molecular and material property prediction; (2) molecular and material design; (3) automation and novel interfaces; (4) scientific communication and education; (5) research data management and automation; (6) hypothesis generation and evaluation; and (7) knowledge extraction and reasoning from scientific literature. Each team submission is presented in a summary table with links to the code and as brief papers in the appendix. Beyond team results, we discuss the hackathon event and its hybrid format, which included physical hubs in Toronto, Montreal, San Francisco, Berlin, Lausanne, and Tokyo, alongside a global online hub to enable local and virtual collaboration. Overall, the event highlighted significant improvements in LLM capabilities since the previous year's hackathon, suggesting continued expansion of LLMs for applications in materials science and chemistry research. These outcomes demonstrate the dual utility of LLMs as both multipurpose models for diverse machine learning tasks and platforms for rapid prototyping custom applications in scientific research., Comment: Updating author information, the submission remains largely unchanged. 98 pages total

The deployment of AI in consumer products is currently focused on the use of so-called foundation models, large neural networks pre-trained on massive corpora of digital records. This emphasis on scaling up datasets and pre-training computation raises the risk of further consolidating the industry, and enabling monopolistic (or oligopolistic) behavior. Judges and regulators seeking to improve market competition may employ various remedies. This paper explores dissolution -- the breaking up of a monopolistic entity into smaller firms -- as one such remedy, focusing in particular on the technical challenges and opportunities involved in the breaking up of large models and datasets. We show how the framework of Conscious Data Contribution can enable user autonomy during under dissolution. Through a simulation study, we explore how fine-tuning and the phenomenon of "catastrophic forgetting" could actually prove beneficial as a type of machine unlearning that allows users to specify which data they want used for what purposes., Comment: This paper appeared at the 2nd Workshop on Regulatable ML at NeurIPS 2024

In the astrophysics community it is common practice to model collisionless dust, entrained in a gas flow, as a pressureless fluid. However a pressureless fluid is fundamentally different from a collisionless fluid - the latter of which generically possess a non-zero anisotropic pressure or stress tensor. In this paper we derive a fluid model for collisionless dust, entrained in a turbulent gas, starting from the equations describing the motion of individual dust grains. We adopt a covariant formulation of our model to allow for the geometry and coordinate systems prevalent in astrophysics, and provide a closure valid for the accretion disc context. We show that the continuum mechanics properties of a dust fluid corresponds to a higher-dimensional anisotropic Maxwell fluid, after the extra dimensions are averaged out, with a dynamically important rheological stress tensor. This higher-dimensional treatment has the advantage of keeping the dust velocity and velocity of the fluid seen, and their respective moments, on the same footing. This results in a simplification of the constitutive relation describing the evolution of the dust Rheological stress., Comment: 51 pages, 7 figures, accepted for publication in JfM

In recent years, signSGD has garnered interest as both a practical optimizer as well as a simple model to understand adaptive optimizers like Adam. Though there is a general consensus that signSGD acts to precondition optimization and reshapes noise, quantitatively understanding these effects in theoretically solvable settings remains difficult. We present an analysis of signSGD in a high dimensional limit, and derive a limiting SDE and ODE to describe the risk. Using this framework we quantify four effects of signSGD: effective learning rate, noise compression, diagonal preconditioning, and gradient noise reshaping. Our analysis is consistent with experimental observations but moves beyond that by quantifying the dependence of these effects on the data and noise distributions. We conclude with a conjecture on how these results might be extended to Adam.

We present the canonical analysis of different versions of unimodular gravity defined in the Pleba\'nski formalism, based on a (generally complex) SO(3) spin connection and set of (self-dual) two-forms. As in the metric formulation of unimodular gravity, one can study either a theory with fixed volume form or work in a parametrised formalism in which the cosmological constant becomes a dynamical field, constrained to be constant by the field equations. In the first case, the Hamiltonian density contains a part which is not constrained to vanish, but rather constrained to be constant, again as in the metric formulation. We also discuss reality conditions and challenges in extracting Lorentzian solutions., Comment: 10 pages, two-column revtex; v2: fixed typos, to appear in Phys. Rev. D

We introduce SymbolFit, a framework that automates parametric modeling by using symbolic regression to perform a machine-search for functions that fit the data, while simultaneously providing uncertainty estimates in a single run. Traditionally, constructing a parametric model to accurately describe binned data has been a manual and iterative process, requiring an adequate functional form to be determined before the fit can be performed. The main challenge arises when the appropriate functional forms cannot be derived from first principles, especially when there is no underlying true closed-form function for the distribution. In this work, we address this problem by utilizing symbolic regression, a machine learning technique that explores a vast space of candidate functions without needing a predefined functional form, treating the functional form itself as a trainable parameter. Our approach is demonstrated in data analysis applications in high-energy physics experiments at the CERN Large Hadron Collider (LHC). We demonstrate its effectiveness and efficiency using five real proton-proton collision datasets from new physics searches at the LHC, namely the background modeling in resonance searches for high-mass dijet, trijet, paired-dijet, diphoton, and dimuon events. We also validate the framework using several toy datasets with one and more variables., Comment: 53 pages, 35 figures. Under review

We introduce FrontierMath, a benchmark of hundreds of original, exceptionally challenging mathematics problems crafted and vetted by expert mathematicians. The questions cover most major branches of modern mathematics -- from computationally intensive problems in number theory and real analysis to abstract questions in algebraic geometry and category theory. Solving a typical problem requires multiple hours of effort from a researcher in the relevant branch of mathematics, and for the upper end questions, multiple days. FrontierMath uses new, unpublished problems and automated verification to reliably evaluate models while minimizing risk of data contamination. Current state-of-the-art AI models solve under 2% of problems, revealing a vast gap between AI capabilities and the prowess of the mathematical community. As AI systems advance toward expert-level mathematical abilities, FrontierMath offers a rigorous testbed that quantifies their progress.

Solving societal problems on a global scale requires the collection and processing of ideas and methods from diverse sets of international experts. As the number and diversity of human experts increase, so does the likelihood that elements in this collective knowledge can be combined and refined to discover novel and better solutions. However, it is difficult to identify, combine, and refine complementary information in an increasingly large and diverse knowledge base. This paper argues that artificial intelligence (AI) can play a crucial role in this process. An evolutionary AI framework, termed RHEA, fills this role by distilling knowledge from diverse models created by human experts into equivalent neural networks, which are then recombined and refined in a population-based search. The framework was implemented in a formal synthetic domain, demonstrating that it is transparent and systematic. It was then applied to the results of the XPRIZE Pandemic Response Challenge, in which over 100 teams of experts across 23 countries submitted models based on diverse methodologies to predict COVID-19 cases and suggest non-pharmaceutical intervention policies for 235 nations, states, and regions across the globe. Building upon this expert knowledge, by recombining and refining the 169 resulting policy suggestion models, RHEA discovered a broader and more effective set of policies than either AI or human experts alone, as evaluated based on real-world data. The results thus suggest that AI can play a crucial role in realizing the potential of human expertise in global problem-solving., Comment: NeurIPS 2024; Main Paper 15 pages, Appendix 11 pages

We present DisCo, a distributed algorithm for contact-rich, multi-robot tasks. DisCo is a distributed contact-implicit trajectory optimization algorithm, which allows a group of robots to optimize a time sequence of forces to objects and to their environment to accomplish tasks such as collaborative manipulation, robot team sports, and modular robot locomotion. We build our algorithm on a variant of the Alternating Direction Method of Multipliers (ADMM), where each robot computes its own contact forces and contact-switching events from a smaller single-robot, contact-implicit trajectory optimization problem, while cooperating with other robots through dual variables, enforcing constraints between robots. Each robot iterates between solving its local problem, and communicating over a wireless mesh network to enforce these consistency constraints with its neighbors, ultimately converging to a coordinated plan for the group. The local problems solved by each robot are significantly less challenging than a centralized problem with all robots' contact forces and switching events, improving the computational efficiency, while also preserving the privacy of some aspects of each robot's operation. We demonstrate the effectiveness of our algorithm in simulations of collaborative manipulation, multi-robot team sports scenarios, and in modular robot locomotion, where DisCo achieves $3$x higher success rates with a 2.5x to 5x faster computation time. Further, we provide results of hardware experiments on a modular truss robot, with three collaborating truss nodes planning individually while working together to produce a punctuated rolling-gate motion of the composite structure. Videos are available on the project page: https://disco-opt.github.io.

The innermost portions of the Milky Way's stellar halo have avoided scrutiny until recently. The lack of wide-area survey data, made it difficult to reconstruct an uninterrupted view of the density distribution of the metal-poor stars inside the Solar radius. In this study, we utilize red giant branch (RGB) stars from Gaia, with metallicities estimated using spectro-photometry from Gaia Data Release 3. Accounting for Gaia's selection function, we examine the spatial distribution of metal-poor ([M/H]<-1.3) RGB stars, from the Galactic centre (r~1 kpc) out to beyond the Solar radius (r~18 kpc). Our best-fitting single-component cored power-law model shows a vertical flattening of ~0.5 and a slope -3.4, consistent with previous studies. Motivated by the mounting evidence for two distinct stellar populations in the inner halo, we additionally test a range of two-component models. One of the components models the tidal debris from the Gaia Sausage/Enceladus merger, while the other captures the Aurora population -- stars that predate the Galactic disk formation. Our best-fit two-component model suggests that both populations contribute equally around the Solar radius, but Aurora dominates the inner halo with a steeper power-law index of -4.5, in agreement with the nitrogen-rich star distribution measured by Horta et al. (2021)., Comment: Submitted to MNRAS

We examine the solution of the Benjamin-Ono Cauchy problem for rational initial data in three types of double-scaling limits in which the dispersion tends to zero while simultaneously the independent variables either approach a point on one of the two branches of the caustic curve of the inviscid Burgers equation, or approach the critical point where the branches meet. The results reveal universal limiting profiles in each case that are independent of details of the initial data. We compare the results obtained with corresponding results for the Korteweg-de Vries equation found by Claeys-Grava in three papers. Our method is to analyze contour integrals appearing in an explicit representation of the solution of the Cauchy problem, in various limits involving coalescing saddle points., Comment: 68 pages, 12 figures

The stellar halo of the Milky Way comprises an abundance of chemical signatures from accretion events and \textit{in-situ} evolution, that form an interweaving tapestry in kinematic space. To untangle this, we consider the mixtures of chemical information, in a given region of integral of motion space, as a variant of the blind source separation problem and utilise non-negative matrix factorisation (NMF). Specifically, we examine the variation in [Fe/H], [Mg/Fe], and [Al/Fe] distributions of APOGEE DR17 stars across the $(E,L_z)$ plane of the halo. When 2 components are prescribed, the NMF algorithm splits stellar halo into low- and high-energy components in the $(E,L_z)$ plane which approximately correspond to the accreted and \textit{in-situ} halo respectively. We use these two components to define a new boundary between the \textit{in-situ} and the accreted stellar halo. Moreover, we calculate the components fractional contribution to the stellar halo as a function of energy, galactocentric spherical radius, height, and galactocentric cylindrical radius. Using a stellar halo defined by kinematic cuts, we find that the halo transitions from \textit{in-situ} dominated to accretion dominated at $E \approx -1.67 \times 10^5$ (km/s)$^2$ (using the potential in McMillan 2017), and at $(r,z,R) \approx (8.7, 3.0, 8.1)$ kpc. The low-energy component is found to span a range of [Al/Fe] that falls beyond the typically accepted \textit{in-situ} floor of [Al/Fe] $=0$. Upon prescribing more components to the NMF model, we find hints of the existence of overlapping chemical evolution sequences that other techniques struggle to find. We also examine features within these components that resemble known substructures in the halo, such as \textit{Eos} and \textit{Aurora}. This work provides insight into their origin and the part they play in the Milky Way's formation., Comment: 17 pages, 15 figures, submitted to MNRAS. Comments welcome!

The leading-order asymptotic behavior of the solution of the Cauchy initial-value problem for the Benjamin-Ono equation in $L^2(\mathbb{R})$ is obtained explicitly for generic rational initial data $u_0$. An explicit asymptotic wave profile $u^\mathrm{ZD}(t,x;\epsilon)$ is given, in terms of the branches of the multivalued solution of the inviscid Burgers equation with initial data $u_0$, such that the solution $u(t,x;\epsilon)$ of the Benjamin-Ono equation with dispersion parameter $\epsilon>0$ and initial data $u_0$ satisfies $u(t,x;\epsilon)-u^\mathrm{ZD}(t,x;\epsilon)\to 0$ in the locally uniform sense as $\epsilon\to 0$, provided a discriminant inequality holds implying that certain caustic curves in the $(t,x)$-plane are avoided. In some cases this convergence implies strong $L^2(\mathbb{R})$ convergence. The asymptotic profile $u^\mathrm{ZD}(t,x;\epsilon)$ is consistent with the modulated multi-phase wave solutions described by Dobrokhotov and Krichever., Comment: 63 pages, 18 figures

Intels efforts to build a practical quantum computer are focused on developing a scalable spin-qubit platform leveraging industrial high-volume semiconductor manufacturing expertise and 300 mm fabrication infrastructure. Here, we provide an overview of the design, fabrication, and demonstration of a new customized quantum test chip, which contains 12-quantum-dot spin-qubit linear arrays, code named Tunnel Falls. These devices are fabricated using immersion and extreme ultraviolet lithography (EUV), along with other standard high-volume manufacturing (HVM) processes, as well as production-level process control. We present key device features and fabrication details, as well as qubit characterization results confirming device functionality. These results corroborate our fabrication methods and are a crucial step towards scaling of extensible 2D qubit array schemes.

CMOS-based microelectronics are limited to ~150{\deg}C and therefore not suitable for the extreme high temperatures in aerospace, energy, and space applications. While wide bandgap semiconductors can provide high-temperature logic, nonvolatile memory devices at high temperatures have been challenging. In this work, we develop a nonvolatile electrochemical memory cell that stores and retains analog and digital information at temperatures as high as 600 {\deg}C. Through correlative electron microscopy, we show that this high-temperature information retention is a result of composition phase separation between the oxidized and reduced forms of amorphous tantalum oxide. This result demonstrates a memory concept that is resilient at extreme temperatures and reveals phase separation as the principal mechanism that enables nonvolatile information storage in these electrochemical memory cells., Comment: 19 pages, 4 figures

We show that the initial-value problem for the Benjamin-Ono equation on $\mathbb{R}$ with $L^2(\mathbb{R})$ rational initial data with only simple poles can be solved in closed form via a determinant formula involving contour integrals. The dimension of the determinant depends on the number of simple poles of the rational initial data only and the matrix elements depend explicitly on the independent variables $(t,x)$ and the dispersion coefficient $\epsilon$. This allows for various interesting asymptotic limits to be resolved quite efficiently., Comment: 22 pages, 1 figure

We present Movie Gen, a cast of foundation models that generates high-quality, 1080p HD videos with different aspect ratios and synchronized audio. We also show additional capabilities such as precise instruction-based video editing and generation of personalized videos based on a user's image. Our models set a new state-of-the-art on multiple tasks: text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation. Our largest video generation model is a 30B parameter transformer trained with a maximum context length of 73K video tokens, corresponding to a generated video of 16 seconds at 16 frames-per-second. We show multiple technical innovations and simplifications on the architecture, latent spaces, training objectives and recipes, data curation, evaluation protocols, parallelization techniques, and inference optimizations that allow us to reap the benefits of scaling pre-training data, model size, and training compute for training large scale media generation models. We hope this paper helps the research community to accelerate progress and innovation in media generation models. All videos from this paper are available at https://go.fb.me/MovieGenResearchVideos.

In this paper we discuss the definition, the construction and the implementation of \textit{generalised antenna functions} for final-state radiation up to Next-to-Next-to-Leading Order (NNLO) in QCD. Generalised antenna functions encapsulate the singular behaviour of unresolved emissions when these occur within multiple hard radiators and not just two of them, as for traditional antenna functions. The construction of such objects is possible thanks to the recently proposed algorithm for building \textit{idealised antenna functions} from a target set of infrared limits. Generalised antenna functions bring major simplifications in the assemblage of subtraction terms in the context of the antenna scheme at NNLO and beyond, as well as a substantial computational speedup of higher-order calculations. We discuss in detail the improvements on the formal and practical side for the computation of the NNLO correction to three-jet production at electron-positron colliders, providing a thorough numerical validation of the newly proposed scheme. For this calculation one can expect almost an order of magnitude speedup with respect to the original implementation.

Recent advances in generative deep learning have transformed small molecule design, but most methods lack biological systems context, focusing narrowly on specific protein pockets. We introduce a non-differentiable diffusion guidance method that integrates systems biology models, enhancing small molecule generation with pathway context. Using the Bone Morphogenetic Protein (BMP) pathway, we generate small molecules specific to one protein over competing proteins. This method enhances the precision and efficiency of small molecule drug discovery by incorporating systems biology insights into generative models., Comment: Accepted at MLSB NeurIPS 2024 workshop

In this report, we propose the implementation of national registries for frontier AI models as a foundational tool for AI governance. We explore the rationale, design, and implementation of such registries, drawing on comparisons with registries in analogous industries to make recommendations for a registry that is efficient, unintrusive, and which will bring AI governance closer to parity with the governmental insight into other high-impact industries. We explore key information that should be collected, including model architecture, model size, compute and data used during training, and we survey the viability and utility of evaluations developed specifically for AI. Our proposal is designed to provide governmental insight and enhance AI safety while fostering innovation and minimizing the regulatory burden on developers. By providing a framework that respects intellectual property concerns and safeguards sensitive information, this registry approach supports responsible AI development without impeding progress. We propose that timely and accurate registration should be encouraged primarily through injunctive action, by requiring third parties to use only registered models, and secondarily through direct financial penalties for non-compliance. By providing a comprehensive framework for AI model registries, we aim to support policymakers in developing foundational governance structures to monitor and mitigate risks associated with advanced AI systems.

By merging models, AI systems can combine the distinct strengths of separate language models, achieving a balance between multiple capabilities without requiring substantial retraining. However, the integration process can be intricate due to differences in training methods and fine-tuning, typically necessitating specialized knowledge and repeated refinement. This paper explores model merging techniques across a spectrum of complexity, examining where automated methods like evolutionary strategies stand compared to hyperparameter-driven approaches such as DARE, TIES-Merging and simpler methods like Model Soups. In addition, we introduce Differentiable Adaptive Merging (DAM), an efficient, adaptive merging approach as an alternative to evolutionary merging that optimizes model integration through scaling coefficients, minimizing computational demands. Our findings reveal that even simple averaging methods, like Model Soups, perform competitively when model similarity is high, underscoring each technique's unique strengths and limitations. We open-sourced DAM, including the implementation code and experiment pipeline, on GitHub: https://github.com/arcee-ai/DAM., Comment: 11 pages, 1 figure, and 3 tables

Tactile displays that lend tangible form to digital content could transform computing interactions. However, achieving the resolution, speed, and dynamic range needed for perceptual fidelity remains challenging. We present a tactile display that directly converts projected light into visible tactile patterns via a photomechanical surface populated with millimeter-scale optotactile pixels. The pixels transduce incident light into mechanical displacements through photostimulated thermal gas expansion, yielding millimeter scale displacements with response times of 2 to 100 milliseconds. Employing projected light for power transmission and addressing renders these displays highly scalable. We demonstrate optically driven displays with up to 1,511 addressable pixels -- several times more pixels than any prior tactile display attaining comparable performance. Perceptual studies confirm that these displays can reproduce diverse spatiotemporal tactile patterns with high fidelity. This research establishes a foundation for practical, versatile high-resolution tactile displays driven by light.

It is widely recognised that semiparametric efficient estimation can be hard to achieve in practice: estimators that are in theory efficient may require unattainable levels of accuracy for the estimation of complex nuisance functions. As a consequence, estimators deployed on real datasets are often chosen in a somewhat ad hoc fashion, and may suffer high variance. We study this gap between theory and practice in the context of a broad collection of semiparametric regression models that includes the generalised partially linear model. We advocate using estimators that are robust in the sense that they enjoy $\sqrt{n}$-consistent uniformly over a sufficiently rich class of distributions characterised by certain conditional expectations being estimable by user-chosen machine learning methods. We show that even asking for locally uniform estimation within such a class narrows down possible estimators to those parametrised by certain weight functions. Conversely, we show that such estimators do provide the desired uniform consistency and introduce a novel random forest-based procedure for estimating the optimal weights. We prove that the resulting estimator recovers a notion of $\textbf{ro}$bust $\textbf{s}$emiparametric $\textbf{e}$fficiency (ROSE) and provides a practical alternative to semiparametric efficient estimators. We demonstrate the effectiveness of our ROSE random forest estimator in a variety of semiparametric settings on simulated and real-world data.

HARMONI is the first light visible and near-IR integral field spectrograph for the ELT. It covers a large spectral range from 470nm to 2450nm with resolving powers from 3300 to 18000 and spatial sampling from 60mas to 4mas. It can operate in two Adaptive Optics modes - SCAO (including a High Contrast capability) and LTAO - or with NOAO. The project is preparing for Final Design Reviews. The integral field spectrograph is a key sub-system of HARMONI instrument, which forms the 2D spectral image and projects it onto the scientific detector. It has 40 operational modes with different platescales and gratings covering the band of 811-2450 nm with three resolution grades. In each of this configurations the as-built spectrograph wave-front error is strictly limited. We perform the sensitivity analysis for measurable and unknown errors and build the errors budget on this basis. Then we correct the values for the actual technological limits and perform a three-stage Monte-Carlo analysis combined with simulation of a few specific effect as the holographic grating wavefront error. Eventually, we show that it is possible to reach the target image quality in terms of the wavefront error and spectral resolution for the entire sub-system with practically feasible tolerances on design parameters., Comment: 20 pages, 10 figures, 1 table

In arXiv:1811.01577 the VOAs associated to 4d $\mathcal{N}=2$ class-S theories were constructed in addition to a generalization for non-simply laced Lie algebras. However, 6d (2,0) theories have an ADE classification, and therefore class-S theories which are engineered by them come in ADE types. Thus, these non-simply laced VOAs are not thought to correspond to 4d physical theories. Regardless, we analyze these VOAs and their Drinfeld-Sokolov reductions in an effort to determine their properties. We find that many of these reduced VOAs appear to correspond to actual 4d $\mathcal{N}=2$ SCFTs. Additionally, we find what appear to be $F_4$ instanton VOAs, and hence from the proposal of arXiv:2201.13435 these correspond to 3d $\mathcal{N}=4$ quiver gauge theories whose Coulomb branches are $F_4$ instanton moduli spaces. Using a construction of twisted class-S VOAs from non-simply laced ones, we find additional evidence that the $F_4$ instanton VOAs do not correspond to four-dimensional field theories and outline an analogous argument for the $G_2$ instanton case. Our method appears to be quite general and may be a promising technique for ruling out 4d origins of VOAs that are similar to those of 4d $\mathcal{N}=2$ SCFTs, in addition to constructing VOAs of actual 4d theories., Comment: 30 pages + references, 13 figures, and 8 tables

Evaluating instruction following capabilities for multimodal, multi-turn dialogue is challenging. With potentially multiple instructions in the input model context, the task is time-consuming for human raters and we show LLM based judges are biased towards answers from the same model. We propose MMMT-IF, an image based multi-turn Q$\&$A evaluation set with added global instructions between questions, constraining the answer format. This challenges models to retrieve instructions dispersed across long dialogues and reason under instruction constraints. All instructions are objectively verifiable through code execution. We introduce the Programmatic Instruction Following ($\operatorname{PIF}$) metric to measure the fraction of the instructions that are correctly followed while performing a reasoning task. The $\operatorname{PIF-N-K}$ set of metrics further evaluates robustness by measuring the fraction of samples in a corpus where, for each sample, at least K out of N generated model responses achieve a $\operatorname{PIF}$ score of one. The $\operatorname{PIF}$ metric aligns with human instruction following ratings, showing 60 percent correlation. Experiments show Gemini 1.5 Pro, GPT-4o, and Claude 3.5 Sonnet, have a $\operatorname{PIF}$ metric that drops from 0.81 on average at turn 1 across the models, to 0.64 at turn 20. Across all turns, when each response is repeated 4 times ($\operatorname{PIF-4-4}$), GPT-4o and Gemini successfully follow all instructions only $11\%$ of the time. When all the instructions are also appended to the end of the model input context, the $\operatorname{PIF}$ metric improves by 22.3 points on average, showing that the challenge with the task lies not only in following the instructions, but also in retrieving the instructions spread out in the model context. We plan to open source the MMMT-IF dataset and metric computation code., Comment: 24 pages, 16 figures

Large Language Models have advanced clinical Natural Language Generation, creating opportunities to manage the volume of medical text. However, the high-stakes nature of medicine requires reliable evaluation, which remains a challenge. In this narrative review, we assess the current evaluation state for clinical summarization tasks and propose future directions to address the resource constraints of expert human evaluation.