Search

GPT-4o is an autoregressive omni model that accepts as input any combination of text, audio, image, and video, and generates any combination of text, audio, and image outputs. It's trained end-to-end across text, vision, and audio, meaning all inputs and outputs are processed by the same neural network. GPT-4o can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, which is similar to human response time in conversation. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50\% cheaper in the API. GPT-4o is especially better at vision and audio understanding compared to existing models. In line with our commitment to building AI safely and consistent with our voluntary commitments to the White House, we are sharing the GPT-4o System Card, which includes our Preparedness Framework evaluations. In this System Card, we provide a detailed look at GPT-4o's capabilities, limitations, and safety evaluations across multiple categories, focusing on speech-to-speech while also evaluating text and image capabilities, and measures we've implemented to ensure the model is safe and aligned. We also include third-party assessments on dangerous capabilities, as well as discussion of potential societal impacts of GPT-4o's text and vision capabilities.

Moir\'e lattices are a general feature of bilayer structures, where an additional periodic superstructure is generated by either lattice mismatch or from a twist angle. They have been shown to stabilise new ground states, including unconventional superconductivity and Mott insulating phases, attributed to strong electron correlations. However, controlling these interactions requires a detailed understanding of the low energy electronic structure which is lacking so far. Probing the electronic structure is challenging due to sample inhomogeneity, the low characteristic energy scales involved and small sample sizes. Through quasiparticle interference (QPI) imaging, scanning tunnelling microscopy (STM) can overcome many of these challenges but requires detailed modelling to extract the $k$-space electronic structure. Here, we present realistic calculations of QPI in twisted bilayer structures, which accounts for the effect of the long range moir\'e lattice on the electronic structure as well as it's interaction at a defect. These calculations reveal that while the moir\'e supercell significantly reduces the size of the Brillouin zone, the QPI scattering vectors retain characteristics of the individual monolayers with distinct perturbations from the twisted geometry that can be directly linked back to the electronic structure. The procedure introduced here provides a general framework to use QPI to determine the low energy electronic structure in moir\'e lattice systems.

Matching promises transparent causal inferences for observational data, making it an intuitive approach for many applications. In practice, however, standard matching methods often perform poorly compared to modern approaches such as response-surface modeling and optimizing balancing weights. We propose Caliper Synthetic Matching (CSM) to address these challenges while preserving simple and transparent matches and match diagnostics. CSM extends Coarsened Exact Matching by incorporating general distance metrics, adaptive calipers, and locally constructed synthetic controls. We show that CSM can be viewed as a monotonic imbalance bounding matching method, so that it inherits the usual bounds on imbalance and bias enjoyed by MIB methods. We further provide a bound on a measure of joint covariate imbalance. Using a simulation study, we illustrate how CSM can even outperform modern matching methods in certain settings, and finally illustrate its use in an empirical example. Overall, we find CSM allows for many of the benefits of matching while avoiding some of the costs., Comment: 36 pages plus 10 pages of appendices, 9 figures total

Rare-earth ions are characterised by transitions with very narrow linewidths even in solid state crystals. Exceedingly long coherence times have been shown on both spin and optical transitions of rare-earth-ion doped crystals. A key factor, and generally the limitation, for such coherence times, is the effects of electronic and nuclear spins in the host crystal. Despite the attractive prospect, a low-strain, spin-free host crystal for rare-earth-ion dopants has not yet been demonstrated. The dopants experience the lowest strain when they substitute for another rare earth (including yttrium). However every stable isotope of the trivalent rare earth ions has either an electron spin, an nuclear spin, or both. The long optical coherence times reported here with erbium dopants in antiferromagnetically ordered gadolinium vanandate suggest an alternative method to achieve the quiet magnetic environment needed for long coherence times: use a magnetic host fully concentrated in electron spins and operate at temperatures low enough for these spins to be ordered. We also observe avoided crossings in the optical spectra, caused by strong coupling between the erbium ions and gadolinium magnons in the host crystal. This suggests the exciting prospect of microwave to optical quantum transduction using the rare-earth ions in these materials mediated by magnons of the host spins.

The development of large language models (LLMs) has expanded to multi-modal systems capable of processing text, images, and speech within a unified framework. Training these models demands significantly larger datasets and computational resources compared to text-only LLMs. To address the scaling challenges, we introduce Mixture-of-Transformers (MoT), a sparse multi-modal transformer architecture that significantly reduces pretraining computational costs. MoT decouples non-embedding parameters of the model by modality -- including feed-forward networks, attention matrices, and layer normalization -- enabling modality-specific processing with global self-attention over the full input sequence. We evaluate MoT across multiple settings and model scales. In the Chameleon 7B setting (autoregressive text-and-image generation), MoT matches the dense baseline's performance using only 55.8\% of the FLOPs. When extended to include speech, MoT reaches speech performance comparable to the dense baseline with only 37.2\% of the FLOPs. In the Transfusion setting, where text and image are trained with different objectives, a 7B MoT model matches the image modality performance of the dense baseline with one third of the FLOPs, and a 760M MoT model outperforms a 1.4B dense baseline across key image generation metrics. System profiling further highlights MoT's practical benefits, achieving dense baseline image quality in 47.2\% of the wall-clock time and text quality in 75.6\% of the wall-clock time (measured on AWS p4de.24xlarge instances with NVIDIA A100 GPUs).

The widespread availability of maps has enabled the development of numerous location-based applications, including navigation, ride-sharing, fitness tracking, gaming, robotics, and augmented reality. Today, the maps that power these services are predominantly controlled by a few large corporations and mostly cover outdoor spaces. As the use of these applications expands and indoor localization technologies advance, we are seeing the need for a scalable, federated location management system that can extend into private spaces. We introduce OpenFLAME (Open Federated Localization and Mapping Engine), the first federated and decentralized localization service. OpenFLAME links servers that handle localization for specific regions, providing applications with a seamless global view. Creating a federated localization system poses challenges, such as discovering the appropriate servers for a region and integrating services managed by independent providers. To address these issues and ensure scalability, we leverage Domain Name System (DNS) for service discovery and implement map abstractions to retrieve and merge locations across different maps. Our trace-driven study demonstrates that federated localization across remote servers is feasible with acceptable query latencies. To highlight the potential of the system, we developed an augmented reality navigation application for a large indoor space, showing that OpenFLAME can successfully power location-based applications.

Leakage, the occupation of any state not used in the computation, is one of the of the most devastating errors in quantum error correction. Transmons, the most common superconducting qubits, are weakly anharmonic multilevel systems, and are thus prone to this type of error. Here we demonstrate a device which reduces the lifetimes of the leakage states in the transmon by three orders of magnitude, while protecting the qubit lifetime and the single-qubit gate fidelties. To do this we attach a qubit through an on-chip seventh-order Chebyshev filter to a cold resistor. The filter is engineered such that the leakage transitions are in its passband, while the qubit transition is in its stopband. Dissipation through the filter reduces the lifetime of the transmon's $f$ state, the lowest energy leakage state, by three orders of magnitude to 33 ns, while simultaneously keeping the qubit lifetime to greater than 100 $\mu$s. Even though the $f$ state is transiently populated during a single qubit gate, no negative effect of the filter is detected with errors per gate approaching 1e-4. Modelling the filter as coupled linear harmonic oscillators, our theoretical analysis of the device corroborate our experimental findings. This leakage reduction unit turns leakage errors into errors within the qubit subspace that are correctable with traditional quantum error correction. We demonstrate the operation of the filter as leakage reduction unit in a mock-up of a single-qubit quantum error correcting cycle, showing that the filter increases the seepage rate back to the qubit subspace., Comment: 14 pages, 9 figures

This study explores the potential of off-the-shelf Vision-Language Models (VLMs) for high-level robot planning in the context of autonomous navigation. Indeed, while most of existing learning-based approaches for path planning require extensive task-specific training/fine-tuning, we demonstrate how such training can be avoided for most practical cases. To do this, we introduce Select2Plan (S2P), a novel training-free framework for high-level robot planning which completely eliminates the need for fine-tuning or specialised training. By leveraging structured Visual Question-Answering (VQA) and In-Context Learning (ICL), our approach drastically reduces the need for data collection, requiring a fraction of the task-specific data typically used by trained models, or even relying only on online data. Our method facilitates the effective use of a generally trained VLM in a flexible and cost-efficient way, and does not require additional sensing except for a simple monocular camera. We demonstrate its adaptability across various scene types, context sources, and sensing setups. We evaluate our approach in two distinct scenarios: traditional First-Person View (FPV) and infrastructure-driven Third-Person View (TPV) navigation, demonstrating the flexibility and simplicity of our method. Our technique significantly enhances the navigational capabilities of a baseline VLM of approximately 50% in TPV scenario, and is comparable to trained models in the FPV one, with as few as 20 demonstrations.

Natural-language assistants are designed to provide users with helpful responses while avoiding harmful outputs, largely achieved through alignment to human preferences. Yet there is limited understanding of whether alignment techniques may inadvertently perpetuate or even amplify harmful biases inherited from their pre-aligned base models. This issue is compounded by the choice of bias evaluation benchmarks in popular preference-finetuned models, which predominantly focus on dominant social categories, such as binary gender, thereby limiting insights into biases affecting underrepresented groups. Towards addressing this gap, we center transgender, nonbinary, and other gender-diverse identities to investigate how alignment procedures interact with pre-existing gender-diverse bias in LLMs. Our key contributions include: 1) a comprehensive survey of bias evaluation modalities across leading preference-finetuned LLMs, highlighting critical gaps in gender-diverse representation, 2) systematic evaluation of gender-diverse biases across 12 models spanning Direct Preference Optimization (DPO) stages, uncovering harms popular bias benchmarks fail to detect, and 3) a flexible framework for measuring harmful biases in implicit reward signals applicable to other social contexts. Our findings reveal that DPO-aligned models are particularly sensitive to supervised finetuning (SFT), and can amplify two forms of real-world gender-diverse harms from their base models: stigmatization and gender non-affirmative language. We conclude with recommendations tailored to DPO and broader alignment practices, advocating for the adoption of community-informed bias evaluation frameworks to more effectively identify and address underrepresented harms in LLMs., Comment: Accepted to 2024 Neurips Queer in AI Workshop

Characterization of breast parenchyma in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a challenging task owing to the complexity of underlying tissue structures. Existing quantitative approaches, like radiomics and deep learning models, lack explicit quantification of intricate and subtle parenchymal structures, including fibroglandular tissue. To address this, we propose a novel topological approach that explicitly extracts multi-scale topological structures to better approximate breast parenchymal structures, and then incorporates these structures into a deep-learning-based prediction model via an attention mechanism. Our topology-informed deep learning model, \emph{TopoTxR}, leverages topology to provide enhanced insights into tissues critical for disease pathophysiology and treatment response. We empirically validate \emph{TopoTxR} using the VICTRE phantom breast dataset, showing that the topological structures extracted by our model effectively approximate the breast parenchymal structures. We further demonstrate \emph{TopoTxR}'s efficacy in predicting response to neoadjuvant chemotherapy. Our qualitative and quantitative analyses suggest differential topological behavior of breast tissue in treatment-na\"ive imaging, in patients who respond favorably to therapy as achieving pathological complete response (pCR) versus those who do not. In a comparative analysis with several baselines on the publicly available I-SPY 1 dataset (N=161, including 47 patients with pCR and 114 without) and the Rutgers proprietary dataset (N=120, with 69 patients achieving pCR and 51 not), \emph{TopoTxR} demonstrates a notable improvement, achieving a 2.6\% increase in accuracy and a 4.6\% enhancement in AUC compared to the state-of-the-art method., Comment: 22 pages, 8 figures, 8 tables, accepted by Medical Image Analysis ( https://www.sciencedirect.com/science/article/abs/pii/S1361841524002986 )

The time until black hole formation in a binary neutron-star (NS) merger contains invaluable information about the nuclear equation of state (EoS) but has thus far been difficult to measure. We propose a new way to constrain the merger remnant's NS lifetime, which is based on the tendency of the NS remnant neutrino-driven winds to enrich the ejected material with helium. Based on the He I $\lambda 1083.3$ nm line, we show that the feature around 800-1200 nm in AT2017gfo at 4.4 days seems inconsistent with a helium mass fraction of $X_{\mathrm{He}} \gtrsim 0.05$ in the polar ejecta. Recent neutrino-hydrodynamic simulations of merger remnants are only compatible with this limit if the NS remnant collapses within 20-30 ms. Such a short lifetime implies that the total binary mass of GW170817, $M_\mathrm{\rm tot}$, lay close to the threshold binary mass for direct gravitational collapse, $M_\mathrm{thres}$, for which we estimate $M_{\mathrm{thres}}\lesssim 2.93 M_\odot$. This upper bound on $M_\mathrm{thres}$ yields upper limits on the radii and maximum mass of cold, non-rotating NSs, which rule out simultaneously large values for both quantities. In combination with causality arguments, this result implies a maximum NS mass of $M_\mathrm{max}\lesssim2.3 M_\odot$. The combination of all limits constrains the radii of 1.6 M$_\odot$ NSs to about 12$\pm$1 km for $M_\mathrm{max}$ = 2.0 M$_\odot$ and 11.5$\pm$1 km for $M_\mathrm{max}$ = 2.15 M$_\odot$. This $\sim2$ km allowable range then tightens significantly for $M_\mathrm{max}$ above $\approx2.15$ M$_\odot$. This rules out a significant number of current EoS models. The short NS lifetime also implies that a black-hole torus, not a highly magnetized NS, was the central engine powering the relativistic jet of GRB170817A. Our work motivates future developments... [abridged], Comment: 30 pages, 16 figures, 2 tables, submitted to PRX

We present a revised mass measurement for HD 119130 b (aka K2-292 b), a transiting planet ($P = 17$ days, $R_\mathrm{p} = 2.63^{+0.11}_{-0.10}$ $R_\mathrm{\oplus}$) orbiting a chromospherically inactive G dwarf, previously thought to be one of the densest sub-Neptunes known. Our follow-up Doppler observations with HARPS, HARPS-N, and HIRES reveal that HD 119130 b is, in fact, nearly one-third as massive as originally suggested by its initial confirmation paper. Our revised analysis finds $M_\mathrm{p} = 8.8 \pm 3.2$ $M_\mathrm{\oplus}$ ($M_\mathrm{p} < 15.4$ $M_\mathrm{\oplus}$ at 98\% confidence) compared to the previously reported $M_\mathrm{p} = 24.5 \pm 4.4$ $M_\mathrm{\oplus}$. While the true cause of the original mass measurement's inaccuracy remains uncertain, we present the plausible explanation that the planet's radial velocity (RV) semi-amplitude was inflated due to constructive interference with a second, untreated sinusoidal signal in the data (possibly rotational modulation from the star). HD 119130 b illustrates the complexities of interpreting the RV orbits of small transiting planets. While RV mass measurements of such planets may be precise, they are not necessarily guaranteed to be accurate. This system serves as a cautionary tale as observers and theorists alike look to the exoplanet mass-radius diagram for insights into the physics of small planet formation., Comment: 15 pages, 9 figures, 3 tables. Accepted for publication in The Astronomical Journal

Here we show a photonic computing accelerator utilizing a system-level thin-film lithium niobate circuit which overcomes this limitation. Leveraging the strong electro-optic (Pockels) effect and the scalability of this platform, we demonstrate photonic computation at speeds up to 1.36 TOPS while consuming 0.057 pJ/OP. Our system features more than 100 thin-film lithium niobate high-performance components working synergistically, surpassing state-of-the-art systems on this platform. We further demonstrate binary-classification, handwritten-digit classification, and image classification with remarkable accuracy, showcasing our system's capability of executing real algorithms. Finally, we investigate the opportunities offered by combining our system with a hybrid-integrated distributed feedback laser source and a heterogeneous-integrated modified uni-traveling carrier photodiode. Our results illustrate the promise of thin-film lithium niobate as a computational platform, addressing current bottlenecks in both electronic and photonic computation. Its unique properties of high-performance electro-optic weight encoding and conversion, wafer-scale scalability, and compatibility with integrated lasers and detectors, position thin-film lithium niobate photonics as a valuable complement to silicon photonics, with extensions to applications in ultrafast and power-efficient signal processing and ranging.

Chemical vapour deposition enables large-domain growth of ideal graphene, yet many applications of graphene require the controlled inclusion of specific defects. We present a one-step chemical vapour deposition procedure aimed at retaining the precursor topology when incorporated into the grown carbonaceous film. When azupyrene, the molecular analogue of the Stone-Wales defect in graphene, is used as a precursor, carbonaceous monolayers with a range of morphologies are produced as a function of the copper substrate growth temperature. The higher the substrate temperature during deposition, the closer the resulting monolayer is to ideal graphene. Analysis, with a set of complementary materials characterisation techniques, reveals morphological changes closely correlated with changes in the atomic adsorption heights, network topology, and concentration of 5-/7-membered carbon rings. The engineered defective carbon monolayers can be transferred to different substrates, potentially enabling applications in nanoelectronics, sensorics, and catalysis., Comment: manuscript and SI provided

Data theft and leakage, caused by external adversaries and insiders, demonstrate the need for protecting user data. Trusted Execution Environments (TEEs) offer a promising solution by creating secure environments that protect data and code from such threats. The rise of confidential computing on cloud platforms facilitates the deployment of TEE-enabled server applications, which are expected to be widely adopted in web services such as privacy-preserving LLM inference and secure data logging. One key feature is Remote Attestation (RA), which enables integrity verification of a TEE. However, $\textit{compatibility}$ issues with RA verification arise as no browsers natively support this feature, making prior solutions cumbersome and risky. To address these challenges, we propose $\texttt{RA-WEBs}$ ($\textbf{R}$emote $\textbf{A}$ttestation for $\textbf{Web}$ $\textbf{s}$ervices), a novel RA protocol designed for high compatibility with the current web ecosystem. $\texttt{RA-WEBs}$ leverages established web mechanisms for immediate deployability, enabling RA verification on existing browsers. We conduct a comprehensive security analysis, demonstrating $\texttt{RA-WEBs}$'s resilience against various threats. Our contributions include the $\texttt{RA-WEBs}$ proposal, a proof-of-concept implementation, an in-depth security analysis, and publicly available code for reproducible research.

Sparse Autoencoders (SAEs) have shown promise in improving the interpretability of neural network activations, but can learn features that are not features of the input, limiting their effectiveness. We propose \textsc{Mutual Feature Regularization} \textbf{(MFR)}, a regularization technique for improving feature learning by encouraging SAEs trained in parallel to learn similar features. We motivate \textsc{MFR} by showing that features learned by multiple SAEs are more likely to correlate with features of the input. By training on synthetic data with known features of the input, we show that \textsc{MFR} can help SAEs learn those features, as we can directly compare the features learned by the SAE with the input features for the synthetic data. We then scale \textsc{MFR} to SAEs that are trained to denoise electroencephalography (EEG) data and SAEs that are trained to reconstruct GPT-2 Small activations. We show that \textsc{MFR} can improve the reconstruction loss of SAEs by up to 21.21\% on GPT-2 Small, and 6.67\% on EEG data. Our results suggest that the similarity between features learned by different SAEs can be leveraged to improve SAE training, thereby enhancing performance and the usefulness of SAEs for model interpretability.

Intense electromagnetic fields can result in dramatic changes in the electronic properties of solids. These changes are commonly studied using optical probes of the modified electronic structure. Here we use optical-scanning tunneling microscopy (optical-STM) equipped with near-field continuous wave (CW) laser excitation to directly measure the electronic structure of light-dressed states in a monolayer transition metal dichalcogenide (TMD) semiconductor, WSe2. We find that the effective tunneling gap and tunneling density of states are strongly influenced by the intensity of the electromagnetic field when the applied field is resonant with the bandgap of the semiconductor. Our findings indicate that CW laser excitation can be used to generate light-dressed electronic states of quantum materials when confined strongly to the nanoscale.

Magnon-condensation, which emerges in pumped bosonic systems at room temperature, continues to garner great interest for its long-lived coherence. While traditionally formulated in terms of Bose-Einstein condensation, which typically occurs at ultra-low temperatures, it could potentially also be explained by Fr\"ohlich-condensation, a hypothesis of Bose-Einstein-like condensation in living systems at ambient temperatures. Here, we elucidate the essential features of magnon-condensation in an open quantum system (OQS) formulation, wherein magnons dissipatively interact with a phonon bath. Our derived equations of motion for expected magnon occupations turns out to be similar in form to the rate equations governing Fr\"ohlich-condensation. Provided that specific system parameters result in correlations amplifying or diminishing the condensation effects, we thereby posit that our treatment offers a better description of high-temperature condensation as opposed to traditional descriptions using equilibrium thermodynamics. By comparing our OQS derivation with the original uncorrelated and previous semi-classical rate equations, we furthermore highlight how both classical anti-correlations and quantum correlations alter the bosonic occupation distribution., Comment: 7 pages, 2 figures, plus supplementary material (8 pages)

The dynamics of viscous thin-film particle-laden flows down inclined surfaces are commonly modeled with one of two approaches: a diffusive flux model or a suspension balance model. The diffusive flux model assumes that the particles migrate via a diffusive flux induced by gradients in both the particle concentration and the effective suspension viscosity. The suspension balance model introduces non-Newtonian bulk stress with shear-induced normal stresses, the gradients of which cause particle migration. Both models have appeared in the literature of particle-laden flow with virtually no comparison between the two models. For particle-laden viscous flow on an incline, in a thin-film geometry, one can use lubrication theory to derive a compact dynamic model in the form of a $2\times 2$ system of conservation laws. We can then directly compare the two theories side by side by looking at similarities and differences in the flux functions for the conservation laws, and in exact and numerical simulations of the equations. We compare the flux profiles over a range of parameters, showing fairly good agreement between the models, with the biggest difference involving the behavior at the free surface. We also consider less dense suspensions at lower inclination angles where the dynamics involve two shock waves that can be clearly measured in experiments. In this context the solutions differ by no more than about 10%, suggesting that either model could be used for this configuration.

The discovery and identification of molecules in biological and environmental samples is crucial for advancing biomedical and chemical sciences. Tandem mass spectrometry (MS/MS) is the leading technique for high-throughput elucidation of molecular structures. However, decoding a molecular structure from its mass spectrum is exceptionally challenging, even when performed by human experts. As a result, the vast majority of acquired MS/MS spectra remain uninterpreted, thereby limiting our understanding of the underlying (bio)chemical processes. Despite decades of progress in machine learning applications for predicting molecular structures from MS/MS spectra, the development of new methods is severely hindered by the lack of standard datasets and evaluation protocols. To address this problem, we propose MassSpecGym -- the first comprehensive benchmark for the discovery and identification of molecules from MS/MS data. Our benchmark comprises the largest publicly available collection of high-quality labeled MS/MS spectra and defines three MS/MS annotation challenges: \textit{de novo} molecular structure generation, molecule retrieval, and spectrum simulation. It includes new evaluation metrics and a generalization-demanding data split, therefore standardizing the MS/MS annotation tasks and rendering the problem accessible to the broad machine learning community. MassSpecGym is publicly available at \url{https://github.com/pluskal-lab/MassSpecGym}.

Structured light plays an important role in metrology, optical trapping and manipulation, communications, quantum technologies, nonlinear optics and provides a rich playground for addressing new optical phenomena. Here we demonstrate a novel approach for manipulating vector vortex beams carrying longitudinal field components using metamaterials with extreme anisotropy. Implementing vectorial spectroscopy, we show that the propagation of complex beams with inhomogeneous polarisation is strongly affected by the interplay of the metamaterial anisotropy with the transverse and longitudinal field structure of the beam. This phenomenon is especially pronounced in the epsilon-near-zero regime, exclusively realised for light polarised along the metamaterial optical axis, strongly influencing the interaction of longitudinal fields with the metamaterial. The requirements on the balance between the transverse and longitudinal fields to maintain a polarisation singularity at the beam axis allow control of the beam modal content, filtering diffraction effects and tailoring spatial polarisation distribution. The proposed approach offers important capabilities for wavefront shaping as well as local spatial polarisation engineering. The understanding of the interaction of vector beams with metamaterials opens new opportunities for applications in microscopy, information encoding, biochemical sensing and quantum technologies.

Many printed electronic applications require strain-independent electrical properties to ensure deformation-independent performance. Thus, developing printed, flexible devices using 2D and other nanomaterials will require an understanding of the effect of strain on the electrical properties of nano-networks. Here we introduce novel AC electrical techniques to fully characterise the effect of strain on the resistance of high mobility printed networks, fabricated from of electrochemically exfoliated MoS2 nanosheets. These devices were initially characterised using DC piezoresistance measurements and showed good cyclability and a linear strain response, consistent with a low gauge factor of G~3. However, AC impedance spectroscopy measurements, performed as a function of strain, allowed the measurement of the effects of strain on both the nanosheets and the inter-nanosheet junctions separately. The junction resistance was found to increase linearly with strain, while the nanosheet resistance remained constant. This response is consistent with strain-induced sliding of the highly-aligned nanosheets past one another, without any strain being transferred to the sheets themselves. Our approach allows us to individually estimate the contributions of dimensional factors (G~1.4) and intrinsic factors (G~1.9) to the total gauge factor. This novel technique may provide insight into other piezoresistive systems., Comment: 3 Figures

Electroaerodynamic propulsion, where force is produced via the momentum-transferring collisions between accelerated ions and neutral air molecules, is a promising alternative mechanism for flight at the micro air vehicle scale due to its silent and solid-state nature. Its relatively low efficiency, however, has thus far precluded its use in a power-autonomous vehicle; leveraging the efficiency benefits of operation close to a fixed surface is a potential solution. While proximity effects like the ground and ceiling effects have been well-investigated for rotorcraft and flapping wing micro air vehicles, they have not been for electroaerodynamically-propelled fliers. In this work, we investigate the change in performance when centimeter-scale thrusters are operated close to a "ceiling" plane about the inlet. We show a surprising and, until now, unreported effect; a major electrostatic attractive component, analogous to electroadhesive pressure but instead mediated by a stable atmospheric plasma. The isolated electrostatic and fluid dynamic components of the ceiling effect are shown for different distances from the plane and for different materials. We further show that a flange attached to the inlet can vastly increase both components of force. A peak efficiency improvement of 600% is shown close to the ceiling. This work points the way towards effective use of the ceiling effect for power autonomous vehicles, extending flight duration, or as a perching mechanism.

We study variants of a stochastic game inspired by backgammon where players may propose to double the stake, with the game state dictated by a one-dimensional random walk. Our variants allow for different numbers of proposals and different multipliers to the stake. We determine the optimal game state for proposing and accepting, giving analytic solutions in many variants. We also introduce a 3-player generalization of the game and prove basic results about its behavior, in addition to providing a simulation.

Undergraduate graders are frequently important contributors to the teaching team in post-secondary education settings. This study set out to investigate agreement for a team of undergraduate graders as they acquired training and experience for scoring responses to open-ended tasks. Results demonstrate compelling evidence that undergraduate students can develop the ability to establish and sustain substantial agreement with an instructor, especially when equipped with proper training and a high-quality scoring rubric., Comment: Supplemental materials for the preprint available by request from the first author

Molten salts are promising candidates in numerous clean energy applications, where challenges in experimental methods limit knowledge of their safety-critical temperature-properties correlations. Herein, we developed and employed machine learning interatomic potentials (MLIP) to study AlCl3 molten salt across varied thermodynamic conditions. The MLIP accurately predicted the existence of Al2Cl6 dimers in this molten salt as informed by Raman spectra and neutron structure factor. The MLIP is validated using available experimental data for temperature correlations with viscosities, surface tension, as well as liquid and vapor densities evaluated from two-phase coexistence simulations. In doing so, we closely predicted the critical temperature and critical density compared to reported experimental values for AlCl3. The demonstrated approach for MLIP training in closely predicting phase equilibrium in this study can be useful towards screening nuclear reactors-relevant salt compositions, helping to mitigate safety concerns., Comment: 37 pages, 9 figures

In this work, we introduce a planning neural operator (PNO) for predicting the value function of a motion planning problem. We recast value function approximation as learning a single operator from the cost function space to the value function space, which is defined by an Eikonal partial differential equation (PDE). Specifically, we recast computing value functions as learning a single operator across continuous function spaces which prove is equivalent to solving an Eikonal PDE. Through this reformulation, our learned PNO is able to generalize to new motion planning problems without retraining. Therefore, our PNO model, despite being trained with a finite number of samples at coarse resolution, inherits the zero-shot super-resolution property of neural operators. We demonstrate accurate value function approximation at 16 times the training resolution on the MovingAI lab's 2D city dataset and compare with state-of-the-art neural value function predictors on 3D scenes from the iGibson building dataset. Lastly, we investigate employing the value function output of PNO as a heuristic function to accelerate motion planning. We show theoretically that the PNO heuristic is $\epsilon$-consistent by introducing an inductive bias layer that guarantees our value functions satisfy the triangle inequality. With our heuristic, we achieve a 30% decrease in nodes visited while obtaining near optimal path lengths on the MovingAI lab 2D city dataset, compared to classical planning methods (A*, RRT*).

This paper focuses on creating synthetic data to improve the quality of image captions. Existing works typically have two shortcomings. First, they caption images from scratch, ignoring existing alt-text metadata, and second, lack transparency if the captioners' training data (e.g. GPT) is unknown. In this paper, we study a principled approach Altogether based on the key idea to edit and re-align existing alt-texts associated with the images. To generate training data, we perform human annotation where annotators start with the existing alt-text and re-align it to the image content in multiple rounds, consequently constructing captions with rich visual concepts. This differs from prior work that carries out human annotation as a one-time description task solely based on images and annotator knowledge. We train a captioner on this data that generalizes the process of re-aligning alt-texts at scale. Our results show our Altogether approach leads to richer image captions that also improve text-to-image generation and zero-shot image classification tasks., Comment: accepted by EMNLP 2024; MetaCLIPv2

Passive acoustic monitoring (PAM) data is often weakly labelled, audited at the scale of detection presence or absence on timescales of minutes to hours. Moreover, this data exhibits great variability from one deployment to the next, due to differences in ambient noise and the signals across sources and geographies. This study proposes a two-step solution to leverage weakly annotated data for training Deep Learning (DL) detection models. Our case study involves binary classification of the presence/absence of sperm whale (\textit{Physeter macrocephalus}) click trains in 4-minute-long recordings from a dataset comprising diverse sources and deployment conditions to maximise generalisability. We tested methods for extracting acoustic features from lengthy audio segments and integrated Temporal Convolutional Networks (TCNs) trained on the extracted features for sequence classification. For feature extraction, we introduced a new approach using Variational AutoEncoders (VAEs) to extract information from both waveforms and spectrograms, which eliminates the necessity for manual threshold setting or time-consuming strong labelling. For classification, TCNs were trained separately on sequences of either VAE embeddings or handpicked acoustic features extracted from the waveform and spectrogram representations using classical methods, to compare the efficacy of the two approaches. The TCN demonstrated robust classification capabilities on a validation set, achieving accuracies exceeding 85\% when applied to 4-minute acoustic recordings. Notably, TCNs trained on handpicked acoustic features exhibited greater variability in performance across recordings from diverse deployment conditions, whereas those trained on VAEs showed a more consistent performance, highlighting the robust transferability of VAEs for feature extraction across different deployment conditions.

This paper explores the nuanced landscape of generative AI (genAI), particularly focusing on neural network-based models like Large Language Models (LLMs). While genAI garners both optimistic enthusiasm and sceptical criticism, this work seeks to provide a balanced examination of its capabilities, limitations, and the profound impact it may have on societal functions and personal interactions. The first section demystifies language-based genAI through detailed discussions on how LLMs learn, their computational needs, distinguishing features from supporting technologies, and the inherent limitations in their accuracy and reliability. Real-world examples illustrate the practical applications and implications of these technologies. The latter part of the paper adopts a systems perspective, evaluating how the integration of LLMs with existing technologies can enhance productivity and address emerging concerns. It highlights the need for significant investment to understand the implications of recent advancements, advocating for a well-informed dialogue to ethically and responsibly integrate genAI into diverse sectors. The paper concludes with prospective developments and recommendations, emphasizing a forward-looking approach to harnessing genAI`s potential while mitigating its risks.

Expert imitation, behavioral diversity, and fairness preferences give rise to preferences in sequential decision making domains that do not decompose additively across time. We introduce the class of convex Markov games that allow general convex preferences over occupancy measures. Despite infinite time horizon and strictly higher generality than Markov games, pure strategy Nash equilibria exist under strict convexity. Furthermore, equilibria can be approximated efficiently by performing gradient descent on an upper bound of exploitability. Our experiments imitate human choices in ultimatum games, reveal novel solutions to the repeated prisoner's dilemma, and find fair solutions in a repeated asymmetric coordination game. In the prisoner's dilemma, our algorithm finds a policy profile that deviates from observed human play only slightly, yet achieves higher per-player utility while also being three orders of magnitude less exploitable.

We describe observations of low energy excess (LEE) events (background events observed in all light dark matter direct detection calorimeters) and noise in a two-channel silicon athermal phonon detector with 375 meV baseline energy resolution. We measure two distinct LEE populations: ``shared'' multichannel events with a pulse shape consistent with athermal phonon events, and sub-eV events which couple nearly exclusively to a single channel with a significantly faster pulse shape. These ``singles'' are consistent with events occurring within the aluminum athermal phonon collection fins. Similarly, our measured detector noise is higher than the theoretical expectation. Measured noise can be split into an uncorrelated component, consistent with shot noise from small energy depositions within the athermal phonon sensor itself, and a correlated component, consistent with shot noise from energy depositions within the silicon crystal's phonon system., Comment: 6 pages, 5 figures

We investigate the fundamental dynamical process of an initial quench of the chemical potential of the two-dimensional $t$-$J$ model. Depending on the ground state phase, sharply different dynamical behavior of the charge distribution and entanglement properties are observed. In the stripe phase, the intertwining of the spin and charge density waves remains stable under time evolution. A ballistic spreading of the charge density is observed with a propagation speed that is only weakly dependent on the coupling ratio, $J/t$. Moreover, in the phase-separated regime for large $J/t$, we report a complete dynamical freezing of charge degrees of freedom within, where even under long time evolution the entanglement entropy remains bounded. Our results are obtained by combining large-scale exact diagonalizations and matrix product state techniques for time evolution., Comment: 10 pages, 8 figures

We construct the first integrable models based on the Haagerup fusion category $H_3$. We introduce a Haagerup version of the anyonic spin chain and use the boost operator formalism to identify two integrable Hamiltonians of PXP type on this chain. The first of these is an analogue of the golden chain; it has a topological symmetry based on $H_3$ and satisfies the Temperley-Lieb algebra with parameter $\delta=(3+\sqrt{13})/2$. We prove its integrability using a Lax formalism, and construct the corresponding solution to the Yang--Baxter equation. We present numerical evidence that this model is gapless with a dynamical critical exponent $z\neq 1$. The second integrable model we find breaks the topological symmetry. We present numerical evidence that this model reduces to a CFT in the large volume limit with central charge $c\sim3/2$., Comment: 6 pages, 3 figures

Concept Bottleneck Models (CBMs) have been proposed as a compromise between white-box and black-box models, aiming to achieve interpretability without sacrificing accuracy. The standard training procedure for CBMs is to predefine a candidate set of human-interpretable concepts, extract their values from the training data, and identify a sparse subset as inputs to a transparent prediction model. However, such approaches are often hampered by the tradeoff between enumerating a sufficiently large set of concepts to include those that are truly relevant versus controlling the cost of obtaining concept extractions. This work investigates a novel approach that sidesteps these challenges: BC-LLM iteratively searches over a potentially infinite set of concepts within a Bayesian framework, in which Large Language Models (LLMs) serve as both a concept extraction mechanism and prior. BC-LLM is broadly applicable and multi-modal. Despite imperfections in LLMs, we prove that BC-LLM can provide rigorous statistical inference and uncertainty quantification. In experiments, it outperforms comparator methods including black-box models, converges more rapidly towards relevant concepts and away from spuriously correlated ones, and is more robust to out-of-distribution samples.

The Workflows Community Summit gathered 111 participants from 18 countries to discuss emerging trends and challenges in scientific workflows, focusing on six key areas: time-sensitive workflows, AI-HPC convergence, multi-facility workflows, heterogeneous HPC environments, user experience, and FAIR computational workflows. The integration of AI and exascale computing has revolutionized scientific workflows, enabling higher-fidelity models and complex, time-sensitive processes, while introducing challenges in managing heterogeneous environments and multi-facility data dependencies. The rise of large language models is driving computational demands to zettaflop scales, necessitating modular, adaptable systems and cloud-service models to optimize resource utilization and ensure reproducibility. Multi-facility workflows present challenges in data movement, curation, and overcoming institutional silos, while diverse hardware architectures require integrating workflow considerations into early system design and developing standardized resource management tools. The summit emphasized improving user experience in workflow systems and ensuring FAIR workflows to enhance collaboration and accelerate scientific discovery. Key recommendations include developing standardized metrics for time-sensitive workflows, creating frameworks for cloud-HPC integration, implementing distributed-by-design workflow modeling, establishing multi-facility authentication protocols, and accelerating AI integration in HPC workflow management. The summit also called for comprehensive workflow benchmarks, workflow-specific UX principles, and a FAIR workflow maturity model, highlighting the need for continued collaboration in addressing the complex challenges posed by the convergence of AI, HPC, and multi-facility research environments.

Misinformation spreads rapidly on social media, confusing the truth and targetting potentially vulnerable people. To effectively mitigate the negative impact of misinformation, it must first be accurately detected before applying a mitigation strategy, such as X's community notes, which is currently a manual process. This study takes a knowledge-based approach to misinformation detection, modelling the problem similarly to one of natural language inference. The EffiARA annotation framework is introduced, aiming to utilise inter- and intra-annotator agreement to understand the reliability of each annotator and influence the training of large language models for classification based on annotator reliability. In assessing the EffiARA annotation framework, the Russo-Ukrainian Conflict Knowledge-Based Misinformation Classification Dataset (RUC-MCD) was developed and made publicly available. This study finds that sample weighting using annotator reliability performs the best, utilising both inter- and intra-annotator agreement and soft-label training. The highest classification performance achieved using Llama-3.2-1B was a macro-F1 of 0.757 and 0.740 using TwHIN-BERT-large., Comment: 8 pages, 3 figures, 3 tables. Code available here: https://github.com/MiniEggz/ruc-misinfo

We utilize serendipitous observations from the Asteroid Terrestrial-impact Last Alert System (ATLAS) and the Zwicky Transient Facility (ZTF) in addition to targeted follow-up observations from the Las Cumbres Observatory (LCO) and Liverpool Telescope to analyze the first observed instance of cometary activity by the newly-discovered Jupiter-family comet C/2023 RN3 (ATLAS), whose orbital dynamics place it close to residing on a Centaur-like orbit. Across our 7-month baseline, we observe an epoch of cometary activity commencing in August 2023 with an increase in brightness of >5.4 mag. The lightcurve of 2023 RN3 indicates the presence of continuous cometary activity across our observations, suggesting the onset of a new period of sustained activity. We find no evidence of any outbursts on top of the observed brightening, nor do we find any significant color evolution across our observations. 2023 RN3 is visibly extended in LCO and Liverpool Telescope observations, indicating the presence of a spatially-extended coma. Numerical integration of 2023 RN3's orbit reveals the comet to have recently undergone a slight increase in semimajor axis due to a planetary encounter with Jupiter, however whether this orbital change could trigger 2023 RN3's cometary activity is unclear. Our estimate for the maximum dust production metric of Afrho ~400 cm is consistent with previous measurements for the Jupiter-family comet and Centaur populations., Comment: 20 pages, 9 figures

Estimating steady state properties of open quantum systems is a crucial problem in quantum technology. In this work, we show how to derive in a scalable way using semi-definite programming certified bounds on the expectation value of an arbitrary observable of interest on the steady state of Lindbladian dynamics. We illustrate our method on a series of many-body systems, including a one-dimensional chain and a two-dimensional ladder, and benchmark it with state-of-the-art tensor-network approaches. For the tested models, only modest computational effort is needed to obtain certified bounds whose precision is comparable to variational methods, Comment: See also arXiv:2410.07384

We examine the language capabilities of language models (LMs) from the critical perspective of human language acquisition. Building on classical language development theories, we propose a three-stage framework to assess the abilities of LMs, ranging from preliminary word understanding to complex grammar and complex logical reasoning. Using this framework, we evaluate the generative capacities of LMs using methods from linguistic research. Results indicate that although recent LMs outperform earlier models in overall performance, their developmental trajectory does not strictly follow the path of human language acquisition. Notably, in generation tasks, LMs are more similar to human performance in areas where information is easier to extract from the corpus, such as average word length, clauses, and auxiliary verbs. Newer LMs did not exhibit significant progress in terms of specific dimensions, such as clauses and auxiliary verbs, where the variation across corpora is relatively limited. Register theory offers a plausible explanation for these observations, suggesting that the linguistic features of the training data have a substantial impact on the models' abilities.

Previous studies on two-timescale stochastic approximation (SA) mainly focused on bounding mean-squared errors under diminishing stepsize schemes. In this work, we investigate {\it constant} stpesize schemes through the lens of Markov processes, proving that the iterates of both timescales converge to a unique joint stationary distribution in Wasserstein metric. We derive explicit geometric and non-asymptotic convergence rates, as well as the variance and bias introduced by constant stepsizes in the presence of Markovian noise. Specifically, with two constant stepsizes $\alpha < \beta$, we show that the biases scale linearly with both stepsizes as $\Theta(\alpha)+\Theta(\beta)$ up to higher-order terms, while the variance of the slower iterate (resp., faster iterate) scales only with its own stepsize as $O(\alpha)$ (resp., $O(\beta)$). Unlike previous work, our results require no additional assumptions such as $\beta^2 \ll \alpha$ nor extra dependence on dimensions. These fine-grained characterizations allow tail-averaging and extrapolation techniques to reduce variance and bias, improving mean-squared error bound to $O(\beta^4 + \frac{1}{t})$ for both iterates.

Linear temporal logic (LTL) and, more generally, $\omega$-regular objectives are alternatives to the traditional discount sum and average reward objectives in reinforcement learning (RL), offering the advantage of greater comprehensibility and hence explainability. In this work, we study the relationship between these objectives. Our main result is that each RL problem for $\omega$-regular objectives can be reduced to a limit-average reward problem in an optimality-preserving fashion, via (finite-memory) reward machines. Furthermore, we demonstrate the efficacy of this approach by showing that optimal policies for limit-average problems can be found asymptotically by solving a sequence of discount-sum problems approximately. Consequently, we resolve an open problem: optimal policies for LTL and $\omega$-regular objectives can be learned asymptotically.

Our information and communication environment has fallen short of the ideals that networked global communication might have served. Identifying all the causes of its pathologies is difficult, but existing recommender systems very likely play a contributing role. In this paper, which draws on the normative tools of philosophy of computing, informed by empirical and technical insights from natural language processing and recommender systems, we make the moral case for an alternative approach. We argue that existing recommenders incentivise mass surveillance, concentrate power, fall prey to narrow behaviourism, and compromise user agency. Rather than just trying to avoid algorithms entirely, or to make incremental improvements to the current paradigm, researchers and engineers should explore an alternative paradigm: the use of language model (LM) agents to source and curate content that matches users' preferences and values, expressed in natural language. The use of LM agents for recommendation poses its own challenges, including those related to candidate generation, computational efficiency, preference modelling, and prompt injection. Nonetheless, if implemented successfully LM agents could: guide us through the digital public sphere without relying on mass surveillance; shift power away from platforms towards users; optimise for what matters instead of just for behavioural proxies; and scaffold our agency instead of undermining it.

The Clifford hierarchy is a fundamental structure in quantum computation, classifying unitary operators based on their commutation relations with the Pauli group. Despite its significance, the mathematical structure of the hierarchy is not well understood at the third level and higher. In this work, we study permutations in the hierarchy: gates which permute the $2^n$ basis states. We fully characterize all the semi-Clifford permutation gates. Moreover, we prove that any permutation gate in the third level, not necessarily semi-Clifford, must be a product of Toffoli gates in what we define as staircase form, up to left and right multiplications of Clifford permutations. Finally, we show that the smallest number of qubits for which there exists a non-semi-Clifford permutation in the third level is $7$., Comment: 15 pages, 3 figures

We introduce Latent Action Pretraining for general Action models (LAPA), an unsupervised method for pretraining Vision-Language-Action (VLA) models without ground-truth robot action labels. Existing Vision-Language-Action models require action labels typically collected by human teleoperators during pretraining, which significantly limits possible data sources and scale. In this work, we propose a method to learn from internet-scale videos that do not have robot action labels. We first train an action quantization model leveraging VQ-VAE-based objective to learn discrete latent actions between image frames, then pretrain a latent VLA model to predict these latent actions from observations and task descriptions, and finally finetune the VLA on small-scale robot manipulation data to map from latent to robot actions. Experimental results demonstrate that our method significantly outperforms existing techniques that train robot manipulation policies from large-scale videos. Furthermore, it outperforms the state-of-the-art VLA model trained with robotic action labels on real-world manipulation tasks that require language conditioning, generalization to unseen objects, and semantic generalization to unseen instructions. Training only on human manipulation videos also shows positive transfer, opening up the potential for leveraging web-scale data for robotics foundation model., Comment: Website: https://latentactionpretraining.github.io

By enabling capturing of 3D point clouds that reflect the geometry of the immediate environment, LiDAR has emerged as a primary sensor for autonomous systems. If a LiDAR scan is too sparse, occluded by obstacles, or too small in range, enhancing the point cloud scan by while respecting the geometry of the scene is useful for downstream tasks. Motivated by the explosive growth of interest in generative methods in vision, conditional LiDAR generation is starting to take off. This paper proposes a novel simultaneous diffusion sampling methodology to generate point clouds conditioned on the 3D structure of the scene as seen from multiple views. The key idea is to impose multi-view geometric constraints on the generation process, exploiting mutual information for enhanced results. Our method begins by recasting the input scan to multiple new viewpoints around the scan, thus creating multiple synthetic LiDAR scans. Then, the synthetic and input LiDAR scans simultaneously undergo conditional generation according to our methodology. Results show that our method can produce accurate and geometrically consistent enhancements to point cloud scans, allowing it to outperform existing methods by a large margin in a variety of benchmarks.

Sample quantiles, such as the median, are often better suited than the sample mean for summarising location characteristics of a data set. Similarly, linear combinations of sample quantiles and ratios of such linear combinations, e.g. the interquartile range and quantile-based skewness measures, are often used to quantify characteristics such as spread and skew. While often reported, it is uncommon to accompany quantile estimates with confidence intervals or standard errors. The rquest package provides a simple way to conduct hypothesis tests and derive confidence intervals for quantiles, linear combinations of quantiles, ratios of dependent linear combinations (e.g., Bowley's measure of skewness) and differences and ratios of all of the above for comparisons between independent samples. Many commonly used measures based on quantiles are included, although it is also very simple for users to define their own. Additionally, quantile-based measures of inequality are also considered. The methods are based on recent research showing that reliable distribution-free confidence intervals can be obtained, even for moderate sample sizes. Several examples are provided herein., Comment: 13 pages