Search

Planar Josephson junctions in a magnetic field exhibit the superconducting diode effect, by which the critical current magnitude depends on the polarity of the transport current. A number of different mechanisms for the effect have been proposed.Here, we study symmetric, T-shaped planar Josephson junctions with semiconducting weak links in an in-plane magnetic field perpendicular to an applied current bias. In particular, we vary the longitudinal width (i.e.\ parallel to the current) of the superconducting contacts and the voltage of an electrostatic gate. We observe an increase in both critical current and diode efficiency with increasing contact width and relate the critical current behavior to the induced coherence length of the Andreev bound states that mediate the supercurrent flow through the junction. We further observe a linear trend, with respect to inverse contact width, of the field at which the diode efficiency is maximized, which saturates as the contact width becomes large compared to the coherence length. The smaller field at which the critical current is maximized additionally exhibits a strong gate dependence. We interpret these observations in the context of multiple underlying mechanisms, including spin--orbit coupling and orbital effects.

Large Language Models (LLMs) benefit from training on ever larger amounts of textual data, but as a result, they increasingly incur the risk of leaking private information. The ability to selectively remove knowledge from LLMs is, therefore, a highly desirable capability. In this paper, we propose LUNAR, a novel unlearning methodology grounded in the Linear Representation Hypothesis. LUNAR operates by redirecting the representations of unlearned data to regions that trigger the model's inherent ability to express its inability to answer. LUNAR achieves state-of-the-art unlearning performance while significantly enhancing the controllability of the unlearned model during inference. Specifically, LUNAR achieves between 2.9x to 11.7x improvements on combined "unlearning efficacy" and "model utility" score ("Deviation Score") on the PISTOL dataset across various base models. We also demonstrate, through quantitative analysis and qualitative examples, LUNAR's superior controllability in generating coherent and contextually aware responses, mitigating undesired side effects of existing methods. Moreover, we demonstrate that LUNAR is robust against white-box adversarial attacks and versatile in handling real-world scenarios, such as processing sequential unlearning requests.

This work revisits optimal response-adaptive designs from a type-I error rate perspective, highlighting when and how much these allocations exacerbate type-I error rate inflation - an issue previously undocumented. We explore a range of approaches from the literature that can be applied to reduce type-I error rate inflation. However, we found that all of these approaches fail to give a robust solution to the problem. To address this, we derive two optimal proportions, incorporating the more robust score test (instead of the Wald test) with finite sample estimators (instead of the unknown true values) in the formulation of the optimization problem. One proportion optimizes statistical power and the other minimizes the total number failures in a trail while maintaining a predefined power level. Through simulations based on an early-phase and a confirmatory trial we provide crucial practical insight into how these new optimal proportion designs can offer substantial patient outcomes advantages while controlling type-I error rate. While we focused on binary outcomes, the framework offers valuable insights that naturally extend to other outcome types, multi-armed trials and alternative measures of interest., Comment: 14 pages, 3 figures, 5 tables

Active galactic nuclei (AGNs) are a key component of galaxy evolution due to feedback on the host from its supermassive black hole. The morphology of warm, in- and outflowing dusty material can reveal the nature of the onset of feedback, AGN feeding, and the unified model of AGN. Here we use the Large Binocular Telescope Interferometer (LBTI) to image the dense, obscuring disk and extended dusty outflow region of NGC 1068. In Fizeau imaging mode the LBTI synthesizes the equivalent resolution of a 22.8 m telescope. The 8.7 $\mu$m Fizeau images of NGC 1068 {have an effective resolution of $47\times90$ mas ($3.3\times6.2$ pc)} in a 5" field of view after performing PSF deconvolution techniques described here. This is the only extragalactic source to be Fizeau imaged using the LBTI, and the images bridge the scales measured with the Very Large Telescope Interferometer (VLTI; 0.5-5 pc) and those of single telescopes such as JWST and Keck ($>15$ pc). The images detect and spatially resolve the low surface brightness mid-infrared (MIR) features in the AGN disk/wind region that are over-resolved by the VLTI. The images show strong correlation between MIR dust emission and near-infrared (NIR) emission of highly excited atomic lines observed by SINFONI. Such LBTI imaging is a precursor to infrared imaging using the upcoming generation of extremely large telescopes, with angular resolutions up to 6x better than JWST, the largest space telescope in orbit., Comment: 26 Pages, Accepted for publication in Nature Astronomy

We present a parallel scan (prefix sum) algorithm in the Tensor Core Unit (TCU) model of computation. The TCU model assumes that multiplication between two square matrices of constant size $s$ is a basic operation. In the $(s^2, \ell)$-TCU model, we show that for inputs of size $n$, the algorithm has depth at most $2\lfloor \log_s (n)\rfloor$ and runs in $O(n(1 + \ell /s^2)/p + (s^2 + \ell) \log_s (n))$ time assuming $p$ tensor core units. Equivalently, the algorithm performs $O(n/s^2)$ multiplications of square matrices of size s., Comment: 14 pages, published in 29th International European Conference on Parallel and Distributed Computing (EuroPar 2023)

Associative memory architectures such as the Hopfield network have long been important conceptual and theoretical models for neuroscience and artificial intelligence. However, translating these abstract models into spiking neural networks has been surprisingly difficult. Indeed, much previous work has been restricted to storing a small number of primarily non-overlapping memories in large networks, thereby limiting their scalability. Here, we revisit the associative memory problem in light of recent advances in understanding spike-based computation. Using a recently-established geometric framework, we show that the spiking activity for a large class of all-inhibitory networks is situated on a low-dimensional, convex, and piecewise-linear manifold, with dynamics that move along the manifold. We then map the associative memory problem onto these dynamics, and demonstrate how the vertices of a hypercubic manifold can be used to store stable, overlapping activity patterns with a direct correspondence to the original Hopfield model. We propose several learning rules, and demonstrate a linear scaling of the storage capacity with the number of neurons, as well as robust pattern completion abilities. Overall, this work serves as a case study to demonstrate the effectiveness of using a geometrical perspective to design dynamics on neural manifolds, with implications for neuroscience and machine learning., Comment: 15 pages, 5 figures; accepted to NeurIPS 2024 Workshop on Symmetry and Geometry in Neural Representations (NeurReps 2024)

As large language models (LLMs) become increasingly powerful, traditional evaluation metrics tend to saturate, making it challenging to distinguish between models based on their performance. We propose a general method to transform existing LLM evaluations into a series of progressively more difficult tasks. These enhanced evaluations emphasize reasoning capabilities and can reveal relative performance differences that are not apparent in the original assessments. To demonstrate the effectiveness of our approach, we create a new multiple-choice test corpus, extend it into a family of evaluations, and assess a collection of LLMs. Our results offer insights into the comparative reasoning abilities of these models, particularly highlighting distinctions between OpenAI's o1-preview and Google's gemini-pro-1.5-002., Comment: 13 pages, 3 figures

We investigate the patterns of incorrect answers produced by large language models (LLMs) during evaluation. These errors exhibit highly non-intuitive behaviors unique to each model. By analyzing these patterns, we measure the similarities between LLMs and construct a taxonomy that categorizes them based on their error correlations. Our findings reveal that the incorrect responses are not randomly distributed but systematically correlated across models, providing new insights into the underlying structures and relationships among LLMs., Comment: 11 pages, 6 figures

Virtual reality is widely adopted for applications such as training, education, and collaboration. The construction industry, known for its complex projects and numerous personnel involved, relies heavily on effective collaboration. Setting up a real-world construction site for experiments can be expensive and time-consuming, whereas conducting experiments in VR is relatively low-cost, scalable, and efficient. We propose Col-Con, a virtual reality simulation testbed for exploring collaborative behaviors in construction. Col-Con is a multi-user testbed that supports users in completing tasks collaboratively. Additionally, Col-Con provides immersive and realistic simulated construction scenes, where real-time voice communication, along with synchronized transformations, animations, sounds, and interactions, enhances the collaborative experience. As a showcase, we implemented a pipe installation construction task based on Col-Con. A user study demonstrated that Col-Con excels in usability, and participants reported a strong sense of immersion and collaboration. We envision that Col-Con will facilitate research on exploring virtual reality-based collaborative behaviors in construction.

There exists an extremely wide array of LLM benchmarking tasks, whereas oftentimes a single number is the most actionable for decision-making, especially by non-experts. No such aggregation schema exists that is not Elo-based, which could be costly or time-consuming. Here we propose a method to aggregate performance across a general space of benchmarks, nicknamed Project "MPG," dubbed Model Performance and Goodness, additionally referencing a metric widely understood to be an important yet inaccurate and crude measure of car performance. Here, we create two numbers: a "Goodness" number (answer accuracy) and a "Fastness" number (cost or QPS). We compare models against each other and present a ranking according to our general metric as well as subdomains. We find significant agreement between the raw Pearson correlation of our scores and those of Chatbot Arena, even improving on the correlation of the MMLU leaderboard to Chatbot Arena.

Interrupted X-ray computed tomography (X-CT) has been the common way to observe the deformation of materials during an experiment. While this approach is effective for quasi-static experiments, it has never been possible to reconstruct a full 3d tomography during a dynamic experiment which cannot be interrupted. In this work, we propose that neural rendering tools can be used to drive the paradigm shift to enable 3d reconstruction during dynamic events. First, we derive theoretical results to support the selection of projections angles. Via a combination of synthetic and experimental data, we demonstrate that neural radiance fields can reconstruct data modalities of interest more efficiently than conventional reconstruction methods. Finally, we develop a spatio-temporal model with spline-based deformation field and demonstrate that such model can reconstruct the spatio-temporal deformation of lattice samples in real-world experiments., Comment: 24 pages, 14 figures. Submitted to NeurIPS 2024 ML4PS. For associated visualizations, see https://neural-xray.github.io/nerfxray

Optical-based terahertz sources are important for many burgeoning scientific and technological applications. Among such applications is precision spectroscopy of molecules, which exhibit rotational transitions at terahertz frequencies. Stemming from precision spectroscopy is frequency discrimination and stabilization of terahertz sources. Because many molecular species exist in the gas phase at room temperature, their transitions are prime candidates for practical terahertz frequency references. We demonstrate the stabilization of a low phase-noise, dual-wavelength Brillouin laser (DWBL) terahertz oscillator to a rotational transition of carbonyl sulfide (\ce{OCS}). We achieve an instability of $1.2\times10^{-12}/\sqrt{\tau}$, where $\tau$ is the averaging time in seconds. The signal-to-noise ratio and intermodulation limitations of the experiment are also discussed. We thus demonstrate a highly stable and spectrally pure terahertz frequency source. Our presented architecture will likely benefit metrology, spectroscopy, precision terahertz studies, and beyond.

Language model pre-training benefits from diverse data to enhance performance across domains and languages. However, training on such heterogeneous corpora requires extensive and costly efforts. Since these data sources vary lexically, syntactically, and semantically, they cause negative interference or the ``curse of multilinguality''. We propose a novel pre-training framework to alleviate this curse. Our method, DEPT, decouples embeddings from the transformer body while simultaneously training the latter in multiple contexts. DEPT enables training without a shared global vocabulary and: (1) can train robustly and effectively under significant data heterogeneity, (2) reduces token embedding parameters by up to 80% and the communication costs by 675x for billion-scale models, (3) enhances model generalization and plasticity in adapting to new languages and domains, and (4) permits training with custom optimized vocabularies per data source. We demonstrate DEPT's potential via the first vocabulary-agnostic federated multilingual pre-training of a 1.3 billion-parameter model, limiting its embedding size to 102.4 million instead of 512 million.

This work is dedicated to debias the Near-Earth Objects (NEO) population based on observations from the Asteroid Terrestrial-impact Last Alert System (ATLAS) telescopes. We have applied similar methods used to develop the recently released NEO model generator (NEOMOD), once debiasing the NEO population using data from Catalina Sky Survey (CSS) G96 telescope. ATLAS is composed of four different telescopes. We first analyzed observational data from each of all four telescopes separately and later combined them. Our results highlight main differences between CSS and ATLAS, e.g., sky coverage and survey power at debiasing the NEO population. ATLAS has a much larger sky coverage than CSS, allowing it to find bright NEOs that would be constantly "hiding" from CSS. Consequently, ATLAS is more powerful than CSS at debiasing the NEO population for H $\lesssim$ 19. With its intrinsically greater sensitivity and emphasis on observing near opposition, CSS excels in the debiasing of smaller objects. ATLAS, as an all sky survey designed to find imminent hazardous objects, necessarily spends a significant fraction of time looking at places on the sky where objects do not appear, reducing its power for debiasing the population of small objects. We estimate a NEO population completeness of $\approx$ 88%$^{+3\%}_{-2\%}$ for H $<$ 17.75 and $\approx$ 36%$^{+1\%}_{-1\%}$ for H $<$ 22.25. Those numbers are similar to previous estimates (within error bars for H $<$ 17.75) from CSS, yet, around 3% and 8% smaller at their face values, respectively. We also confirm previous finding that the $\nu_6$ secular resonance is the main source of small and faint NEOs at H = 28, whereas the 3:1 mean motion resonance with Jupiter dominates for larger and brighter NEOs at H = 15., Comment: Accepted for publication in Icarus, 26 pages, 13 figures, 2 tables

Many natural phenomena are quantified by counts of observable events, from the annihilation of quasiparticles in a lattice to predator-prey encounters on a landscape to spikes in a neural network. These events are triggered at random intervals when an underlying dynamical system occupies a set of reactive states in its phase space. We derive a general expression for the distribution of times between events in such counting processes assuming the underlying triggering dynamics is a stochastic process that converges to a stationary distribution. Our results contribute to resolving a long-standing dichotomy in the study of reaction-diffusion processes, showing the inter-reaction point process interpolates between a reaction- and a diffusion-limited regime. At low reaction rates, the inter-reaction process is Poisson with a rate depending on stationary properties of the event-triggering stochastic process. At high reaction rates, inter-reaction times are dominated by the hitting times to the reactive states. To further illustrate the power of this approach we apply our framework to obtain the counting statistics of two counting processes appearing in several biophysical scenarios. First, we study the common situation of estimating an animal's activity level by how often it crosses a detector, showing that the mean number of crossing events can decrease monotonically with the hitting rate, a seemingly 'paradoxical' result that could possibly lead to misinterpretation of experimental count data. Second, we derive the ensemble statistics for the detection of many particles, recovering and generalizing known results in the biophysics of chemosensation. Overall, we develop a unifying theoretical framework to quantify inter-event time distributions in reaction-diffusion systems that clarifies existing debates in the literature and provide examples of application to real-world scenarios., Comment: 14 pages, 3 figures

Anomalous metallic behavior is often observed near superconductor-insulator quantum critical points and, if persistent to zero temperature, poses a challenge to current theories of metals. One explanation for this behavior could be incomplete thermal equilibrium between the sample and the cryostat. However, despite decades of study, the temperature of an anomalous metal has not yet been measured. We introduce a new experimental probe by capturing microwave radiation emitted from the anomalous metal, using a two-dimensional array of superconductor-semiconductor hybrid Josephson junctions as a model system. The total emitted radiation exceeds the limits of thermodynamic equilibrium, and is well described by an elevated sample temperature. The extracted sample temperature matches the onset of anomalous metallic behavior. Additionally, we discover scaling behavior of radiative noise in the presence of an applied bias. Elements of our noise-scaling observations were predicted based on nonlinear critical field theories and gauge-gravity duality between current noise and Hawking radiation from a black hole. This work shows that, in a prominent model system, anomalous metallic behavior is a non-equilibrium effect, and opens a new frontier in the study of universal, non-equilibrium phenomena near quantum criticality., Comment: 7+8 pages, 4+11 figures

We obtained new spectra of Kepler-34 and Kepler-35 with Keck-HIRES, nearly a decade after these systems were originally characterized with this spectrograph and other instruments, to search for RV trends from a potential third stellar-mass companion at long periods. For Kepler-34, we rule out coplanar stellar masses as low as $0.12 M_\odot$ at an orbital period of $\lesssim 52$ years. For Kepler-35, we rule out stellar masses of $0.13 M_\odot$ at orbital periods of $\lesssim 55$ years. Highly stable, extreme precision RV instruments, as well as improved methodologies in characterizing double-lined spectroscopic binaries that come with these new instruments, will provide an opportunity to push these mass limits lower in the future., Comment: Submitted to ApJ. Comments are welcome

Majorana bound states (MBSs) are quasiparticles which are their own antiparticles. They are predicted to emerge as zero-energy modes localized at the boundary of a topological superconductor. No intrinsic topological superconductor is known to date. However, by interfacing conventional superconductors and semiconductors with strong spin--orbit coupling it is possible to create a system hosting topological states. Hence epitaxial superconductors and semiconductors have emerged as an attractive materials system with atomically sharp interfaces and broad flexibility in device fabrications incorporating Josephson junctions. We discuss the basics of topological superconductivity and provide insight on how to go beyond current state-of-the-art experiments. We argue that the ultimate success in realizing MBS physics requires the observation of non-Abelian braiding and fusion experiments.

The signature of a $p$-weakly geometric rough path summarises a path up to a generalised notion of reparameterisation. The quotient space of equivalence classes on which the signature is constant yields unparameterised path space. The study of topologies on unparameterised path space, initiated in [CT24b] for paths of bounded variation, has practical bearing on the use of signature based methods in a variety applications. This note extends the majority of results from [CT24b] to unparameterised weakly geometric rough path space. We study three classes of topologies: metrisable topologies for which the quotient map is continuous; the quotient topology derived from the underlying path space; and an explicit metric between the tree-reduced representatives of each equivalence class. We prove that topologies of the first type (under an additional assumption) are separable and Lusin, but not locally compact or completely metrisable. The quotient topology is Hausdorff but not metrisable, while the metric generating the third topology is not complete and its topology is not locally compact. We also show that the third topology is Polish when $p=1$.

The asteroid (142) Polana is classified as a B-type asteroid located in the inner Main Belt. This asteroid is the parent of the New Polana family, which has been proposed to be the likely source of primitive near-Earth asteroids such as the B-type asteroid (101955) Bennu. To investigate the compositional correlation between Polana and Bennu at the 3-micron band and their aqueous alteration histories, we analyzed the spectra of Polana in the ~2.0-4.0-micron spectral range using the NASA Infrared Telescope Facility in Hawaii. Our findings indicate that Polana does not exhibit discernable 3-micron hydrated mineral absorption (within 2 sigma), which is in contrast to asteroid Bennu. Bennu displayed a significant 3-micron absorption feature similar to CM- and CI-type carbonaceous chondrites. This suggests two possibilities: either Bennu did not originate from the New Polana family parented by asteroid Polana or the interior of Bennu's parent body was not homogenous, with diverse levels of aqueous alteration. Several explanations support the latter possibility, including heating due to shock waves and pressure, which could have caused the current dehydrated state of Bennu's parent body.

We present High-Density Visual Particle Dynamics (HD-VPD), a learned world model that can emulate the physical dynamics of real scenes by processing massive latent point clouds containing 100K+ particles. To enable efficiency at this scale, we introduce a novel family of Point Cloud Transformers (PCTs) called Interlacers leveraging intertwined linear-attention Performer layers and graph-based neighbour attention layers. We demonstrate the capabilities of HD-VPD by modeling the dynamics of high degree-of-freedom bi-manual robots with two RGB-D cameras. Compared to the previous graph neural network approach, our Interlacer dynamics is twice as fast with the same prediction quality, and can achieve higher quality using 4x as many particles. We illustrate how HD-VPD can evaluate motion plan quality with robotic box pushing and can grasping tasks. See videos and particle dynamics rendered by HD-VPD at https://sites.google.com/view/hd-vpd.

Recently, machine unlearning, which seeks to erase specific data stored in the pre-trained or fine-tuned models, has emerged as a crucial protective measure for LLMs. However, unlearning approaches for LLMs that have been considered thus far have focused on the removal of independent data points and have not taken into account that the stored facts are logically connected to one another and form an implicit knowledge graph. To facilitate the development of structural unlearning methods, which are essential for the practical application of unlearning, we propose PISTOL, a pipeline for compiling multi-scenario datasets for benchmarking structural LLM unlearning. Additionally, leveraging sample datasets synthesized using PISTOL, we conducted benchmarks with four distinct unlearning methods on both Llama2-7B and Mistral-7B models. This analysis helps to illustrate the prevailing challenges in effectively and robustly removing highly inter-connected data, batched data, or data skewed towards a specific domain. It also highlights the choice of pre-trained model can impact unlearning performance. This work not only advances our understandings on the limitation of current LLMs unlearning methods and proposes future research directions, but also provides a replicable framework for ongoing exploration and validation in the field.

As the AI supply chain grows more complex, AI systems and models are increasingly likely to incorporate multiple internally- or externally-sourced components such as datasets and (pre-trained) models. In such cases, determining whether or not the aggregate AI system or model complies with the EU AI Act (AIA) requires a multi-step process in which compliance-related information about both the AI system or model and all its component parts is: (1) gathered, potentially from multiple arms-length sources; (2) harmonized, if necessary; (3) inputted into an analysis that looks across all of it to render a compliance prediction. Because this process is so complex and time-consuming, it threatens to overburden the limited compliance resources of the AI providers (i.e., developers) who bear much of the responsibility for complying with the AIA. It also renders rapid or real-time compliance analyses infeasible in many AI development scenarios where they would be beneficial to providers. To address these shortcomings, we introduce a complete system for automating provider-side AIA compliance analyses amidst a complex AI supply chain. This system has two key elements. First is an interlocking set of computational, multi-stakeholder transparency artifacts that capture AIA-specific metadata about both: (1) the provider's overall AI system or model; and (2) the datasets and pre-trained models it incorporates as components. Second is an algorithm that operates across all those artifacts to render a real-time prediction about whether or not the aggregate AI system or model complies with the AIA. All told, this system promises to dramatically facilitate and democratize provider-side AIA compliance analyses (and, perhaps by extension, provider-side AIA compliance).

We present characterization results and performance of a prototype Multiple-Amplifier Sensing (MAS) silicon charge-coupled device (CCD) sensor with 16 channels potentially suitable for faint object astronomical spectroscopy and low-signal, photon-limited imaging. The MAS CCD is designed to reach sub-electron readout noise by repeatedly measuring charge through a line of amplifiers during the serial transfer shifts. Using synchronized readout electronics based on the DESI CCD controller, we report a read noise of 1.03 e$^-$ rms/pix at a speed of 26 $\mu$s/pix with a single-sample readout scheme where charge in a pixel is measured only once for each output stage. At these operating parameters, we find the amplifier-to-amplifier charge transfer efficiency (ACTE) to be $>0.9995$ at low counts for all amplifiers but one for which the ACTE is 0.997. This charge transfer efficiency falls above 50,000 electrons for the read-noise optimized voltage configuration we chose for the serial clocks and gates. The amplifier linearity across a broad dynamic range from $\sim$300 to 35,000 e$^-$ was also measured to be $\pm 2.5\%$. We describe key operating parameters to optimize on these characteristics and describe the specific applications for which the MAS CCD may be a suitable detector candidate., Comment: 21 pages, 18 figures, accepted to PASP

The reliance of language model training on massive amounts of computation and vast datasets scraped from potentially low-quality, copyrighted, or sensitive data has come into question practically, legally, and ethically. Federated learning provides a plausible alternative by enabling previously untapped data to be voluntarily gathered from collaborating organizations. However, when scaled globally, federated learning requires collaboration across heterogeneous legal, security, and privacy regimes while accounting for the inherent locality of language data; this further exacerbates the established challenge of federated statistical heterogeneity. We propose a Worldwide Federated Language Model Training~(WorldLM) system based on federations of federations, where each federation has the autonomy to account for factors such as its industry, operating jurisdiction, or competitive environment. WorldLM enables such autonomy in the presence of statistical heterogeneity via partial model localization by allowing sub-federations to attentively aggregate key layers from their constituents. Furthermore, it can adaptively share information across federations via residual layer embeddings. Evaluations of language modeling on naturally heterogeneous datasets show that WorldLM outperforms standard federations by up to $1.91\times$, approaches the personalized performance of fully local models, and maintains these advantages under privacy-enhancing techniques., Comment: 19 pages, 8 figures, Under Review

Simulating large scenes with many rigid objects is crucial for a variety of applications, such as robotics, engineering, film and video games. Rigid interactions are notoriously hard to model: small changes to the initial state or the simulation parameters can lead to large changes in the final state. Recently, learned simulators based on graph networks (GNNs) were developed as an alternative to hand-designed simulators like MuJoCo and PyBullet. They are able to accurately capture dynamics of real objects directly from real-world observations. However, current state-of-the-art learned simulators operate on meshes and scale poorly to scenes with many objects or detailed shapes. Here we present SDF-Sim, the first learned rigid-body simulator designed for scale. We use learned signed-distance functions (SDFs) to represent the object shapes and to speed up distance computation. We design the simulator to leverage SDFs and avoid the fundamental bottleneck of the previous simulators associated with collision detection. For the first time in literature, we demonstrate that we can scale the GNN-based simulators to scenes with hundreds of objects and up to 1.1 million nodes, where mesh-based approaches run out of memory. Finally, we show that SDF-Sim can be applied to real world scenes by extracting SDFs from multi-view images.

Generative pre-trained large language models (LLMs) have demonstrated impressive performance over a wide range of tasks, thanks to the unprecedented amount of data they have been trained on. As established scaling laws indicate, LLMs' future performance improvement depends on the amount of computing and data sources they can leverage for pre-training. Federated learning (FL) has the potential to unleash the majority of the planet's data and computational resources, which are underutilized by the data-center-focused training methodology of current LLM practice. Our work presents a robust, flexible, reproducible FL approach that enables large-scale collaboration across institutions to train LLMs. We propose a scalable deployment system called Photon to enable the investigation and development of this new training paradigm for LLM pre-training. We show that Photon can be used by organizations interested in collaborating with their private data sources and computational resources for pre-training LLMs with billions of parameters. This paradigm would mobilize more computational and data resources while matching or potentially exceeding centralized performance. We further show the effectiveness of the federated training scales with model size and present our approach for training billion-scale federated LLMs using limited resources. Thus far, we have used Photon to train LLM models to the size of 7B parameters and anticipate larger models being completed in the near future. Finally, we show that LLM training is highly resilient to the classical challenges of federated statistical and hardware heterogeneity. Furthermore, we show that convergence is robust to partial participation, opening the avenue for compute-efficient collaborative training. Photon will help data-rich actors to become the protagonists of LLMs pre-training instead of leaving the stage to compute-rich actors alone., Comment: 24 pages, 15 figures, pre-print

Our previous model (NEOMOD2) for the orbital and absolute magnitude distribution of Near Earth Objects (NEOs) was calibrated on the Catalina Sky Survey observations between 2013 and 2022. Here we extend NEOMOD2 to include visible albedo information from the Wide-Field Infrared Survey Explorer. The debiased albedo distribution of NEOs can be approximated by the sum of two Rayleigh distributions with the scale parameters p_V,dark=0.03 and p_V,bright=0.17. We find evidence for smaller NEOs having (on average) higher albedos than larger NEOs; this is likely a consequence of the size-dependent sampling of different main belt sources. These inferences and the absolute magnitude distribution from NEOMOD2 are used to construct the debiased size distribution of NEOs. We estimate 830+/-60 NEOs with diameters D>1 km and 20,000+/-2,000 NEOs with D>140 m. The new model, NEOMOD3, is available via the NEOMOD Simulator -- an easy-to-operate code that can be used to generate user-defined samples (orbits, sizes and albedos) from the model., Comment: Icarus, in press

The study of small ($<$300 m) near-Earth objects (NEOs) is important because they are more closely related than larger objects to the precursors of meteorites that fall on Earth. Collisions of these bodies with Earth are also more frequent. Although such collisions cannot produce massive extinction events, they can still produce significant local damage. Here we present the results of a photometric and spectroscopic survey of small NEOs, which include near-infrared (NIR) spectra of 84 objects with a mean diameter of 126 m and photometric data of 59 objects with a mean diameter of 87 m. We found that S-complex asteroids are the most abundant among the NEOs, comprising $\sim$66\% of the sample. Most asteroids in the S-complex were found to have compositions consistent with LL-chondrites. Our study revealed the existence of NEOs with spectral characteristics similar to those in the S-complex, but that could be hidden within the C- or X-complex due to their weak absorption bands. We suggest that the presence of metal or shock-darkening could be responsible for the attenuation of the absorption bands. These objects have been grouped into a new subclass within the S-complex called Sx-types. The dynamical modeling showed that 83\% of the NEOs escaped from the $\nu_{6}$ resonance, 16\% from the 3:1 and just 1\% from the 5:2 resonance. Lightcurves and rotational periods were derived from the photometric data. No clear trend between the axis ratio and the absolute magnitude or rotational period of the NEOs was found., Comment: 61 pages, 43 figures, 5 tables. Accepted for publication in the Planetary Science Journal

Planar Josephson junctions (JJs) have emerged as a promising platform for the realization of topological superconductivity and Majorana zero modes. To obtain robust quasi one-dimensional (1D) topological superconducting states using planar JJs, limiting the number of 1D Andreev bound states' subbands that can be present, and increasing the size of the topological superconducting gap, are two fundamental challenges. It has been suggested that both problems can be addressed by properly designing the interfaces between the JJ's normal region and the superconducting leads. We fabricated Josephson junctions with periodic hole structures on the superconducting contact leads on InAs heterostructures with epitaxial superconducting Al. By depleting the chemical potential inside the hole region with a top gate, we observed an enhancement of the supercurrent across the junction. Such an enhancement is reproduced in theoretical simulations. The theoretical analysis shows that the enhancement of the JJ's critical current is achieved when the hole depletion is such to optimize the matching of quasiparticles' wave-function at the normal/superconductor interface. These results show how the combination of carefully designed patterns for the Al coverage, and external gates, can be successfully used to tune the density and wave functions' profiles in the normal region of the JJ, and therefore open a new avenue to tune some of the critical properties, such as number of subbands and size of the topological gap, that must be optimized to obtain robust quasi 1D superconducting states supporting Majorana bound states.

The main asteroid belt (MAB) is known to be primarily composed of objects from two distinct taxonomic classes, generically defined here as S- and C-complex. The former probably originated from the inner solar system (interior to Jupiter's orbit), while the latter probably from the outer solar system. Following this definition, (4) Vesta, a V-type residing in the inner MAB (a < 2.5 au), is the sole D > 500 km object akin to S-complex that potentially formed in-situ. This provides a useful constraint on the number of D > 500 km bodies that could have formed, or grown, within the primordial MAB. In this work we numerically simulate the accretion of objects in the MAB region during the time when gas in the protoplanetary disk still existed, while assuming different MAB primordial masses. We then accounted for the depletion of that population happening after gas disk dispersal. In our analysis, we subdivided the MAB into five sub-regions and showed that the depletion factor varies throughout the MAB. This results in uneven radial- and size-dependent depletion of the MAB. We show that the MAB primordial mass has to be $\lesssim$ 2.14$\times$10$^{-3}$ Earth masses. Larger primordial masses would lead to the accretion of tens-to-thousands of S-complex objects with D > 500 km in the MAB. Such large objects would survive depletion even in the outer sub-regions (a > 2.5 au), thus being inconsistent with observations. Our results also indicate that S-complex objects with D > 200-300 km, including (4) Vesta, are likely to be terrestrial planetesimals implanted into the MAB rather than formed in-situ., Comment: Accepted for publication in PSJ, 23 pages, 7 figures, 4 tables

Recent years have seen the Kirkwood-Dirac (KD) distribution come to the forefront as a powerful quasi-probability distribution for analysing quantum mechanics. The KD distribution allows tools from statistics and probability theory to be applied to problems in quantum-information processing. A notable difference to the Wigner function is that the KD distribution can represent a quantum state in terms of arbitrary observables. This paper reviews the KD distribution, in three parts. First, we present definitions and basic properties of the KD distribution and its generalisations. Second, we summarise the KD distribution's extensive usage in the study or development of measurement disturbance; quantum metrology; weak values; direct measurements of quantum states; quantum thermodynamics; quantum scrambling and out-of-time-ordered correlators; and the foundations of quantum mechanics, including Leggett-Garg inequalities, the consistent-histories interpretation and contextuality. We emphasise connections between operational quantum advantages and negative or non-real KD quasi-probabilities. Third, we delve into the KD distribution's mathematical structure. We summarise the current knowledge regarding the geometry of KD-positive states (the states for which the KD distribution is a classical probability distribution), describe how to witness and quantify KD non-positivity, and outline relationships between KD non-positivity, coherence and observables' incompatibility., Comment: 42 pages, 14 figures; as published in NJP

Life is an exergonic chemical reaction. Many individual reactions in metabolism entail slightly endergonic processes that are coupled to free energy release, typically as ATP hydrolysis, in order to go forward. ATP is almost always supplied by the rotor-stator ATP synthetase (the ATPase), which harnesses chemiosmotic ion gradients. Because the ATPase is a protein, it arose after the ribosome did. Here we address two questions using comparative physiology: What was the energy currency of metabolism before the origin of the ATPase? How (and why) did ATP come to be the universal energy currency? About 27 percent of a cell's energy budget is consumed as GTP during translation. The universality of GTP-dependence in ribosome function indicates that GTP was the ancestral energy currency of protein synthesis. The use of GTP in translation and ATP in small molecule synthesis are conserved across all lineages, representing energetic compartments that arose in the last universal common ancestor, LUCA., Comment: 27 pages, 5 figures, 2 tables

Autoencoders are powerful machine learning models used to compress information from multiple data sources. However, autoencoders, like all artificial neural networks, are often unidentifiable and uninterpretable. This research focuses on creating an identifiable and interpretable autoencoder that can be used to meld and combine climate data products. The proposed autoencoder utilizes a Bayesian statistical framework, allowing for probabilistic interpretations while also varying spatially to capture useful spatial patterns across the various data products. Constraints are placed on the autoencoder as it learns patterns in the data, creating an interpretable consensus that includes the important features from each input. We demonstrate the utility of the autoencoder by combining information from multiple precipitation products in High Mountain Asia., Comment: 13 pages, 7 figures

Random developments of a path into a matrix Lie group $G_N$ have recently been used to construct signature-based kernels on path space. Two examples include developments into GL$(N;\mathbb{R})$ and $U(N;\mathbb{C})$, the general linear and unitary groups of dimension $N$. For the former, [MLS23] showed that the signature kernel is obtained via a scaling limit of developments with Gaussian vector fields. The second instance was used in [LLN23] to construct a metric between probability measures on path space. We present a unified treatment to obtaining large $N$ limits by leveraging the tools of free probability theory. An important conclusion is that the limiting kernels, while dependent on the choice of Lie group, are nonetheless universal limits with respect to how the development map is randomised. For unitary developments, the limiting kernel is given by the contraction of a signature against the monomials of freely independent semicircular random variables. Using the Schwinger-Dyson equations, we show that this kernel can be obtained by solving a novel quadratic functional equation. We provide a convergent numerical scheme for this equation, together with rates, which does not require computation of signatures themselves.

Jupiter Trojans (JTs) librate about the Lagrangian stationary centers L4 and L5 associated with this planet on a typically small-eccentricity and moderate-inclination heliocentric orbits. The physical and orbital properties of JTs provide important clues about the dynamical evolution of the giant planets in the early Solar System, as well as populations of planetesimals in their source regions. Here we use decade long observations from the Catalina Sky Survey (station G96) to determine the bias-corrected orbital and magnitude distributions of JTs. We distinguish the background JT population, filling smoothly the long-term stable orbital zone about L4 and L5 points, and collisional families. We find that the cumulative magnitude distribution of JTs (the background population in our case) has a steep slope for $H\leq 9$, followed with a moderately shallow slope till $H\simeq 14.5$, beyond which the distribution becomes even shallower. At $H=15$ we find a local power-law exponent $0.38\pm 0.01$. We confirm the asymmetry between the magnitude limited background populations in L4 and L5 clouds characterized by a ratio $1.45\pm 0.05$ for $H<15$. Our analysis suggests an asymmetry in the inclination distribution of JTs, with the L4 population being tighter and the L5 population being broader. We also provide a new catalog of the synthetic proper elements for JTs with an updated identification of statistically robust families (9 at L4, and 4 at L5). The previously known Ennomos family is found to consist of two, overlapping Deiphobus and Ennomos families., Comment: 46 pages, 36 figures, 4 tables, accepted for publication in The Astronomical Journal

Accurately simulating real world object dynamics is essential for various applications such as robotics, engineering, graphics, and design. To better capture complex real dynamics such as contact and friction, learned simulators based on graph networks have recently shown great promise. However, applying these learned simulators to real scenes comes with two major challenges: first, scaling learned simulators to handle the complexity of real world scenes which can involve hundreds of objects each with complicated 3D shapes, and second, handling inputs from perception rather than 3D state information. Here we introduce a method which substantially reduces the memory required to run graph-based learned simulators. Based on this memory-efficient simulation model, we then present a perceptual interface in the form of editable NeRFs which can convert real-world scenes into a structured representation that can be processed by graph network simulator. We show that our method uses substantially less memory than previous graph-based simulators while retaining their accuracy, and that the simulators learned in synthetic environments can be applied to real world scenes captured from multiple camera angles. This paves the way for expanding the application of learned simulators to settings where only perceptual information is available at inference time., Comment: 16 pages, 12 figures

Catalina Sky Survey (CSS) is a major survey of Near-Earth Objects (NEOs). In a recent work, we used CSS observations from 2005-2012 to develop a new population model of NEOs (NEOMOD). CSS's G96 telescope was upgraded in 2016 and detected over 10,000 unique NEOs since then. Here we characterize the NEO detection efficiency of G96 and use G96's NEO detections from 2013-2022 to update NEOMOD. This resolves previous model inconsistencies related to the population of large NEOs. We estimate there are 936+/-29 NEOs with absolute magnitude H<17.75 (diameter D>1 km for the reference albedo p_V=0.14). The slope of the NEO size distribution for H=25-28 is found to be relatively shallow (cumulative index 2.6) and the number of H<28 NEOs (D>9 m) is determined to be (1.20+/-0.04)x10^7. Small NEOs have a different orbital distribution and higher impact probabilities than large NEOs. We estimate 0.034+/-0.002 impacts of H<28 NEOs on the Earth per year, which is near the low end of the impact flux range inferred from atmospheric bolide observations. Relative to a model where all NEOs are delivered directly from the main belt, the population of small NEOs detected by G96 shows an excess of low-eccentricity orbits with a=1--1.6 au that appears to increase with H. We suggest that the population of very small NEOs is boosted by tidal disruption of large NEOs during close encounters to the terrestrial planets. When the effect of tidal disruption is (approximately) accounted for in the model, we estimate 0.06+/-0.01 impacts of H<28 NEOs on the Earth per year, which is more in line with the bolide data., Comment: Icarus, in press. arXiv admin note: text overlap with arXiv:2306.09521

Realistic simulation is critical for applications ranging from robotics to animation. Traditional analytic simulators sometimes struggle to capture sufficiently realistic simulation which can lead to problems including the well known "sim-to-real" gap in robotics. Learned simulators have emerged as an alternative for better capturing real-world physical dynamics, but require access to privileged ground truth physics information such as precise object geometry or particle tracks. Here we propose a method for learning simulators directly from observations. Visual Particle Dynamics (VPD) jointly learns a latent particle-based representation of 3D scenes, a neural simulator of the latent particle dynamics, and a renderer that can produce images of the scene from arbitrary views. VPD learns end to end from posed RGB-D videos and does not require access to privileged information. Unlike existing 2D video prediction models, we show that VPD's 3D structure enables scene editing and long-term predictions. These results pave the way for downstream applications ranging from video editing to robotic planning.

The origin of the Jupiter Trojan asteroids has long been a mystery. Dynamically, the population, which is considerably smaller than the main asteroid belt, librates around Jupiter's stable L4 and L5 Lagrange points, 60 deg ahead and behind Jupiter. It is thought that these bodies were captured into these orbits early in solar system history, but any capture mechanism must also explain why the Trojans have an excited inclination distribution, with some objects reaching inclinations of 35 deg. The Trojans themselves, individually and in aggregate, also have spectral and physical properties that appear consistent with many small bodies found in the outer solar system (e.g., irregular satellites, Kuiper belt objects). In this review, we assemble what is known about the Trojans and discuss various models for their origin and collisional evolution. It can be argued that the Trojans are unlikely to be captured planetesimals from the giant planet zone, but instead were once denizens of the primordial Kuiper belt, trapped by the events taking place during a giant planet instability. The Lucy mission to the Trojans is therefore well positioned to not only answer questions about these objects, but also about their place in planet formation and solar system evolution studies., Comment: 30 pages, 4 figures

Grid decarbonization for climate change requires dispatchable carbon-free energy like nuclear fusion. The tokamak concept offers a promising path for fusion, but one of the foremost challenges in implementation is the occurrence of energetic plasma disruptions. In this study, we delve into Machine Learning approaches to predict plasma state outcomes. Our contributions are twofold: (1) We present a novel application of Continuous Convolutional Neural Networks for disruption prediction and (2) We examine the advantages and disadvantages of continuous models over discrete models for disruption prediction by comparing our model with the previous, discrete state of the art, and show that continuous models offer significantly better performance (Area Under the Receiver Operating Characteristic Curve = 0.974 v.s. 0.799) with fewer parameters, Comment: Accepted at CCAI NeurIPS 2023