Search

Neutral hydrogen (HI) bubbles and shells are common in the interstellar medium (ISM). Studying their properties provides insight into the characteristics of the local ISM as well as the galaxy in which the bubbles reside. We report the detection of magnetic fields associated with superbubbles in the nearby irregular galaxy, the Small Magellanic Cloud (SMC). Using the Polarisation Sky Survey of the Universe's Magnetism (POSSUM) pilot survey, we obtain a high-density grid ($\approx 25 \,\rm sources\,deg^{-2}$) of Faraday rotation measure (RM) from polarized sources behind the SMC. This provides a sufficiently large number of RM measurements to study the magnetic properties of three of the largest HI shells previously identified in the SMC. The RM profiles as a function of distance from the shell centre show characteristic patterns at angular scales comparable to the shell size. We demonstrate that this can be explained by magneto-hydrodynamic simulation models of bubbles expanding in magnetised environments. From the observations, we estimate the line-of-sight magnetic field strength at the edges of the shells is enhanced by $\sim1\,\rm \mu G$ with respect to their centres. This is an order of magnitude larger than the field strength in the ambient medium ($\sim 0.1\,\rm \mu G$) estimated based on the expansion velocity of the shells. This paper highlights the power of densely mapped RM grids in studying the magnetic properties of galactic substructures beyond the Milky Way., Comment: 15 pages, 7 figures, Accepted for publication in MNRAS

This paper presents a novel approach for robust global localisation and 6DoF pose estimation of ground robots in forest environments by leveraging cross-view factor graph optimisation and deep-learned re-localisation. The proposed method addresses the challenges of aligning aerial and ground data for pose estimation, which is crucial for accurate point-to-point navigation in GPS-denied environments. By integrating information from both perspectives into a factor graph framework, our approach effectively estimates the robot's global position and orientation. We validate the performance of our method through extensive experiments in diverse forest scenarios, demonstrating its superiority over existing baselines in terms of accuracy and robustness in these challenging environments. Experimental results show that our proposed localisation system can achieve drift-free localisation with bounded positioning errors, ensuring reliable and safe robot navigation under canopies., Comment: 7 pages, 4 figures, Submitted to ICRA2025

In this work, we quantify the irregularity of a given cylindrical L\'evy process $L$ in $L^2({\mathbb R}^d)$ by determining the range of weighted Besov spaces $B$ in which $L$ has a regularised version $Y$, that is a stochastic process $Y$ in the classical sense with values in $B$. Our approach is based on characterising L\'evy measures on Besov spaces. As a by-product, we determine those Besov spaces $B$ for which the embedding of $L^2({\mathbb R}^d)$ into $B$ is $0$-Radonifying and $p$-Radonifying for $p>1$.

Marginal structural models (MSMs) are widely used in observational studies to estimate the causal effect of time-varying treatments. Despite its popularity, limited attention has been paid to summarizing the treatment history in the outcome model, which proves particularly challenging when individuals' treatment trajectories exhibit complex patterns over time. Commonly used metrics such as the average treatment level fail to adequately capture the treatment history, hindering causal interpretation. For scenarios where treatment histories exhibit distinct temporal patterns, we develop a new approach to parameterize the outcome model. We apply latent growth curve analysis to identify representative treatment trajectories from the observed data and use the posterior probability of latent class membership to summarize the different treatment trajectories. We demonstrate its use in parameterizing the MSMs, which facilitates the interpretations of the results. We apply the method to analyze data from an existing cohort of lung transplant recipients to estimate the effect of Tacrolimus concentrations on the risk of incident chronic kidney disease., Comment: We have discovered that the core idea of our paper overlaps with a previously published work. In light of this, we need to conduct a more thorough update and revision of our research before proceeding further

Automation transformed various aspects of our human civilization, revolutionizing industries and streamlining processes. In the domain of scientific inquiry, automated approaches emerged as powerful tools, holding promise for accelerating discovery, enhancing reproducibility, and overcoming the traditional impediments to scientific progress. This article evaluates the scope of automation within scientific practice and assesses recent approaches. Furthermore, it discusses different perspectives to the following questions: Where do the greatest opportunities lie for automation in scientific practice?; What are the current bottlenecks of automating scientific practice?; and What are significant ethical and practical consequences of automating scientific practice? By discussing the motivations behind automated science, analyzing the hurdles encountered, and examining its implications, this article invites researchers, policymakers, and stakeholders to navigate the rapidly evolving frontier of automated scientific practice.

The concept of personalised medicine in cancer therapy is becoming increasingly important. There already exist drugs administered specifically for patients with tumours presenting well-defined mutations. However, the field is still in its infancy, and personalised treatments are far from being standard of care. Personalised medicine is often associated with the utilisation of omics data. Yet, implementation of multi-omics data has proven difficult, due to the variety and scale of the information within the data, as well as the complexity behind the myriad of interactions taking place within the cell. An alternative approach to precision medicine is to employ a function-based profile of the cell. This involves screening a range of drugs against patient derived cells. Here we demonstrate a proof-of-concept, where a collection of drug screens against a highly diverse set of patient-derived cell lines, are leveraged to identify putative treatment options for a 'new patient'. We show that this methodology is highly efficient in ranking the drugs according to their activity towards the target cells. We argue that this approach offers great potential, as activities can be efficiently imputed from various subsets of the drug treated cell lines that do not necessarily originate from the same tissue type., Comment: 3 figures and 5 tables

Existing local Explainable AI (XAI) methods, such as LIME, select a region of the input space in the vicinity of a given input instance, for which they approximate the behaviour of a model using a simpler and more interpretable surrogate model. The size of this region is often controlled by a user-defined locality hyperparameter. In this paper, we demonstrate the difficulties associated with defining a suitable locality size to capture impactful model behaviour, as well as the inadequacy of using a single locality size to explain all predictions. We propose a novel method, MASALA, for generating explanations, which automatically determines the appropriate local region of impactful model behaviour for each individual instance being explained. MASALA approximates the local behaviour used by a complex model to make a prediction by fitting a linear surrogate model to a set of points which experience similar model behaviour. These points are found by clustering the input space into regions of linear behavioural trends exhibited by the model. We compare the fidelity and consistency of explanations generated by our method with existing local XAI methods, namely LIME and CHILLI. Experiments on the PHM08 and MIDAS datasets show that our method produces more faithful and consistent explanations than existing methods, without the need to define any sensitive locality hyperparameters.

Clinical question answering systems have the potential to provide clinicians with relevant and timely answers to their questions. Nonetheless, despite the advances that have been made, adoption of these systems in clinical settings has been slow. One issue is a lack of question-answering datasets which reflect the real-world needs of health professionals. In this work, we present RealMedQA, a dataset of realistic clinical questions generated by humans and an LLM. We describe the process for generating and verifying the QA pairs and assess several QA models on BioASQ and RealMedQA to assess the relative difficulty of matching answers to questions. We show that the LLM is more cost-efficient for generating "ideal" QA pairs. Additionally, we achieve a lower lexical similarity between questions and answers than BioASQ which provides an additional challenge to the top two QA models, as per the results. We release our code and our dataset publicly to encourage further research., Comment: Accepted at AMIA Annual Symposium 2024

We have fabricated a device which includes two lateral p-n junctions on an n-type GaAs/Al$_{0.33}$Ga$_{0.67}$As heterostructure. A section of the n-type material has been converted to p-type by removing dopants and applying a voltage to a gate placed in this region. Controlled electroluminescence from both of the p-n junctions has been demonstrated by varying the applied bias voltages. An emission peak with a width of ~8 nm is observed around 812 nm. The electroluminescence seen from both junctions is considered to originate from the GaAs quantum well layer in the device. The lithographic techniques that we have developed are compatible for further integration of gated quantum devices such as single-electron pumps to build on-demand single-photon sources.

Understanding how people behave in strategic settings--where they make decisions based on their expectations about the behavior of others--is a long-standing problem in the behavioral sciences. We conduct the largest study to date of strategic decision-making in the context of initial play in two-player matrix games, analyzing over 90,000 human decisions across more than 2,400 procedurally generated games that span a much wider space than previous datasets. We show that a deep neural network trained on these data predicts people's choices better than leading theories of strategic behavior, indicating that there is systematic variation that is not explained by those theories. We then modify the network to produce a new, interpretable behavioral model, revealing what the original network learned about people: their ability to optimally respond and their capacity to reason about others are dependent on the complexity of individual games. This context-dependence is critical in explaining deviations from the rational Nash equilibrium, response times, and uncertainty in strategic decisions. More broadly, our results demonstrate how machine learning can be applied beyond prediction to further help generate novel explanations of complex human behavior.

The First Large Absorption Survey in HI (FLASH) is a large-area radio survey for neutral hydrogen in the redshift range 0.41$, and appear to be a mixture of intervening and associated systems. The overall detection rate for HI absorption lines in the Pilot Surveys (0.3 to 0.5 lines per ASKAP field) is a factor of two below the expected value. There are several possible reasons for this, but one likely factor is the presence of a range of spectral-line artefacts in the Pilot Survey data that have now been mitigated and are not expected to recur in the full FLASH survey. A future paper will discuss the host galaxies of the HI absorption systems identified here., Comment: 46 pages, 25 figures, 10 tables. Submitted to PASA

Stars with initial mass above roughly 8 solar masses will evolve to form a core made of iron group elements at which point no further exothermic nuclear reactions between charged nuclei may prevent the core collapse. Electron captures, neutrino losses, and the photo-disintegration of heavy nuclei trigger the collapse of these stars. Models at the brink of core collapse are produced using stellar evolution codes and these pre-collapse models may be used in the study of the subsequent dynamical evolution (including their explosion as supernovae and the formation of compact remnants such as neutron stars or black holes). We upgrade the physical ingredients employed by the GENeva stellar Evolution Code, GENEC, so that it may cover the regime of high temperatures and high densities required to produce progenitors of core-collapse. We have improved GENEC in three directions, equation of state, the nuclear reaction network and the radiative and conductive opacities adapted for the computation of the advanced phases of evolution. We produce a small grid of pre-supernova models of stars with zero-age main sequence masses of 15, 20 and 25 solar masses at solar and less than half solar metallicities. The results are compared with analogous models produced with the MESA code. The global properties of our new models, particularly of their inner cores, are comparable to models computed with MESA and pre-existing progenitors in the literature. Between codes the exact shell structure varies impacting explosion predictions. Using GENEC with state-of-the-art physics, we have produced massive stellar progenitors prior to collapse. These progenitors are suitable for follow-up studies, including the dynamical collapse and supernova phases. Larger grids of supernova progenitors are now feasible, with potential for further dynamical evolution., Comment: 19 pages, 14 figures

An I-surface $X$ is a surface of general type with $K_X^2 =1$ and $p_g(X) =2$. This paper studies the asymptotic behavior of the period map for I-surfaces acquiring simple elliptic singularities. First we describe the relationship between the deformation theory of such surfaces and their $d$-semistable models. Next we analyze the mixed Hodge structures on the $d$-semistable models, the corresponding limiting mixed Hodge structures, and the monodromy. There are $6$ possible boundary strata for which the relevant limiting mixed Hodge structures satisfy: $\dim W_1 = 4$, and hence $W_2/W_1$ is of pure type $(1,1)$. We show that, in each case, the nilpotent orbit of limiting mixed Hodge structures determines the boundary stratum and prove a global Torelli theorem for one such stratum., Comment: 40 pages. v.2: typos corrected and other minor revisions

We present initial results from the Polarisation Sky Survey of the Universe's Magnetism (POSSUM), analysing 22,817 Faraday Rotation Measures (RMs) with median uncertainties of 1.2 rad m^-2 across 1,520 square degrees to study magnetised gas associated with 55 nearby galaxy groups (z less than 0.025) with halo masses between 10^12.5 and 10^14.0 M_sun. We identify two distinct gas phases: the Intragroup Medium (IGrM) within 0-2 splashback radii and the Warm-Hot Intergalactic Medium (WHIM) extending from 2 to 7 splashback radii. These phases enhance the standard deviation of residual (i.e., Galactic foreground RM-subtracted) RMs by 6.9 +/- 1.8 rad m^-2 and 4.2 +/- 1.2 rad m^-2, respectively. Estimated magnetic field strengths are several microGauss within the IGrM and 0.1-1 microGauss in the WHIM. We estimate the plasma beta in both phases and show that magnetic pressure might be more dynamically important than in the ICM of more massive clusters or sparse cosmic web filaments. Our findings indicate that "missing baryons" in the WHIM likely extend beyond the gravitational radii of group-mass halos to Mpc scales, consistent with large-scale, outflow-driven "magnetised bubbles" seen in cosmological simulations. We demonstrate that RM grids are an effective method for detecting magnetised thermal gas at galaxy group interfaces and within the cosmic web. This approach complements X-ray and Sunyaev-Zel'dovich effect methods, and when combined with Fast Radio Burst Dispersion Measures, data from the full POSSUM survey, comprising approximately a million RMs, will allow direct magnetic field measurements to further our understanding of baryon circulation in these environments and the magnetised universe., Comment: 15 pages, 7 figures, 3 tables. Accepted for publication in MNRAS

Initial value problems -- a system of ordinary differential equations and corresponding initial conditions -- can be used to describe many physical phenomena including those arise in classical mechanics. We have developed a novel approach to solve physics-based initial value problems using unsupervised machine learning. We propose a deep learning framework that models the dynamics of a variety of mechanical systems through neural networks. Our framework is flexible, allowing us to solve non-linear, coupled, and chaotic dynamical systems. We demonstrate the effectiveness of our approach on systems including a free particle, a particle in a gravitational field, a classical pendulum, and the H\'enon--Heiles system (a pair of coupled harmonic oscillators with a non-linear perturbation, used in celestial mechanics). Our results show that deep neural networks can successfully approximate solutions to these problems, producing trajectories which conserve physical properties such as energy and those with stationary action. We note that probabilistic activation functions, as defined in this paper, are required to learn any solutions of initial value problems in their strictest sense, and we introduce coupled neural networks to learn solutions of coupled systems.

What do we want from machine intelligence? We envision machines that are not just tools for thought, but partners in thought: reasonable, insightful, knowledgeable, reliable, and trustworthy systems that think with us. Current artificial intelligence (AI) systems satisfy some of these criteria, some of the time. In this Perspective, we show how the science of collaborative cognition can be put to work to engineer systems that really can be called ``thought partners,'' systems built to meet our expectations and complement our limitations. We lay out several modes of collaborative thought in which humans and AI thought partners can engage and propose desiderata for human-compatible thought partnerships. Drawing on motifs from computational cognitive science, we motivate an alternative scaling path for the design of thought partners and ecosystems around their use through a Bayesian lens, whereby the partners we construct actively build and reason over models of the human and world.

We present results from a search for X-ray/gamma-ray counterparts of gravitational-wave (GW) candidates from the third observing run (O3) of the LIGO-Virgo-KAGRA (LVK) network using the Swift Burst Alert Telescope (Swift-BAT). The search includes 636 GW candidates received in low latency, 86 of which have been confirmed by the offline analysis and included in the third cumulative Gravitational-Wave Transient Catalogs (GWTC-3). Targeted searches were carried out on the entire GW sample using the maximum--likelihood NITRATES pipeline on the BAT data made available via the GUANO infrastructure. We do not detect any significant electromagnetic emission that is temporally and spatially coincident with any of the GW candidates. We report flux upper limits in the 15-350 keV band as a function of sky position for all the catalog candidates. For GW candidates where the Swift-BAT false alarm rate is less than 10$^{-3}$ Hz, we compute the GW--BAT joint false alarm rate. Finally, the derived Swift-BAT upper limits are used to infer constraints on the putative electromagnetic emission associated with binary black hole mergers., Comment: 50 pages, 10 figures, 4 tables

The trustworthiness of Machine Learning (ML) models can be difficult to assess, but is critical in high-risk or ethically sensitive applications. Many models are treated as a `black-box' where the reasoning or criteria for a final decision is opaque to the user. To address this, some existing Explainable AI (XAI) approaches approximate model behaviour using perturbed data. However, such methods have been criticised for ignoring feature dependencies, with explanations being based on potentially unrealistic data. We propose a novel framework, CHILLI, for incorporating data context into XAI by generating contextually aware perturbations, which are faithful to the training data of the base model being explained. This is shown to improve both the soundness and accuracy of the explanations.

Increasing electric current from a single-electron source is a main challenge in an effort to establish the standard of the ampere defined by the fixed value of the elementary charge $e$ and operation frequency $f$. While the current scales with $f$, due to an operation frequency limit for maintaining accurate single-electron transfer, parallelisation of singleelectron sources is expected to be a more practical solution to increase the generated electric current $I = Nef$, where $N$ is a number of parallelised devices. One way to parallelise single-electron sources without increasing the complexity in device operation is to use a common gate. Such a scheme will require each device to have the same operation parameters for single-electron transfer. In order to investigate this possibility, we study the statistics for operation gate voltages using single-electron sources embedded in a multiplexer circuit. The multiplexer circuit allows us to measure 64 single-electron sources individually in a single cooldown. We also demonstrate the parallelisation of three single-electron sources and observe the generated current enhanced by a factor of three., Comment: 5 pages, 3 figures

Chain-of-Thought (CoT) prompting has been shown to enhance the multi-step reasoning capabilities of Large Language Models (LLMs). However, debates persist about whether LLMs exhibit abstract generalization or rely on shallow heuristics when given CoT prompts. To understand the factors influencing CoT reasoning we provide a detailed case study of the symbolic reasoning task of decoding shift ciphers, where letters are shifted forward some number of steps in the alphabet. GPT-4 achieves zero accuracy on most shift ciphers with standard prompting, but with CoT its accuracy improves to an average of 32%. By focusing on a single relatively simple task, we are able to identify three factors that systematically affect CoT performance: the probability of the task's expected output (probability), what the model has implicitly learned during pre-training (memorization), and the number of intermediate operations involved in reasoning (noisy reasoning). We show that these factors can drastically influence the task accuracy; e.g., varying the output's probability of occurrence can shift accuracy from 26% to 70%. We also demonstrate that it is essential for the model to explicitly produce intermediate steps as output that can be conditioned on to increase the probability of the correct answer. Our experiments indicate that as long as the model does so, the validity of the demonstrations in the prompt does not matter. Overall, we conclude that CoT prompting performance reflects both memorization and a probabilistic version of genuine reasoning., Comment: 9 pages plus references and appendices

The classic Johnson Kendall Roberts (JKR) theory describes the short-ranged adhesive contact of elastic bodies, but is only valid for axisymmetric contact. A theory for non-axisymmetric contact, which relies on approximating the contact region as an ellipse, was proposed by Johnson and Greenwood (JG). The theory includes the effects of adhesion via Griffith's criterion applied only at the semi-major and semi-minor axes of the contact ellipse. Although JG's work is in good agreement with numerical and experimental results for quasi-circular contacts, the agreement worsens as the eccentricity of the contact region increases. In this paper, we show that including the effects of adhesion by instead minimizing the sum of elastic and surface energy yields results that are in excellent agreement with previous numerical simulations over the full range of contact eccentricities.

We introduce the notion of infinitesimal variations of mixed Hodge structures and invariants associated to them. We describe these invariants in the case of a pair $(X,Y)$ with $X$ a Fano 3-fold and $Y$ a smooth anticanonical K3 surface and in more detail in the case when $X$ is a cubic threefold. In this last setting, we obtain a generic global Torelli theorem for pairs.

In order for AI systems to communicate effectively with people, they must understand how we make decisions. However, people's decisions are not always rational, so the implicit internal models of human decision-making in Large Language Models (LLMs) must account for this. Previous empirical evidence seems to suggest that these implicit models are accurate -- LLMs offer believable proxies of human behavior, acting how we expect humans would in everyday interactions. However, by comparing LLM behavior and predictions to a large dataset of human decisions, we find that this is actually not the case: when both simulating and predicting people's choices, a suite of cutting-edge LLMs (GPT-4o & 4-Turbo, Llama-3-8B & 70B, Claude 3 Opus) assume that people are more rational than we really are. Specifically, these models deviate from human behavior and align more closely with a classic model of rational choice -- expected value theory. Interestingly, people also tend to assume that other people are rational when interpreting their behavior. As a consequence, when we compare the inferences that LLMs and people draw from the decisions of others using another psychological dataset, we find that these inferences are highly correlated. Thus, the implicit decision-making models of LLMs appear to be aligned with the human expectation that other people will act rationally, rather than with how people actually act.

Using both single-index measures and stochastic dominance concepts, we show how Bayesian inference can be used to make multivariate welfare comparisons. A four-dimensional distribution for the well-being attributes income, mental health, education, and happiness are estimated via Bayesian Markov chain Monte Carlo using unit-record data taken from the Household, Income and Labour Dynamics in Australia survey. Marginal distributions of beta and gamma mixtures and discrete ordinal distributions are combined using a copula. Improvements in both well-being generally and poverty magnitude are assessed using posterior means of single-index measures and posterior probabilities of stochastic dominance. The conditions for stochastic dominance depend on the class of utility functions that is assumed to define a social welfare function and the number of attributes in the utility function. Three classes of utility functions are considered, and posterior probabilities of dominance are computed for one, two, and four-attribute utility functions for three time intervals within the period 2001 to 2019.

Current multimodal and multitask foundation models like 4M or UnifiedIO show promising results, but in practice their out-of-the-box abilities to accept diverse inputs and perform diverse tasks are limited by the (usually rather small) number of modalities and tasks they are trained on. In this paper, we expand upon the capabilities of them by training a single model on tens of highly diverse modalities and by performing co-training on large-scale multimodal datasets and text corpora. This includes training on several semantic and geometric modalities, feature maps from recent state of the art models like DINOv2 and ImageBind, pseudo labels of specialist models like SAM and 4DHumans, and a range of new modalities that allow for novel ways to interact with the model and steer the generation, for example image metadata or color palettes. A crucial step in this process is performing discrete tokenization on various modalities, whether they are image-like, neural network feature maps, vectors, structured data like instance segmentation or human poses, or data that can be represented as text. Through this, we expand on the out-of-the-box capabilities of multimodal models and specifically show the possibility of training one model to solve at least 3x more tasks/modalities than existing ones and doing so without a loss in performance. This enables more fine-grained and controllable multimodal generation capabilities and allows us to study the distillation of models trained on diverse data and objectives into a unified model. We successfully scale the training to a three billion parameter model using tens of modalities and different datasets. The resulting models and training code are open sourced at 4m.epfl.ch., Comment: Project page at 4m.epfl.ch

`scores` is a Python package containing mathematical functions for the verification, evaluation and optimisation of forecasts, predictions or models. It supports labelled n-dimensional (multidimensional) data, which is used in many scientific fields and in machine learning. At present, `scores` primarily supports the geoscience communities; in particular, the meteorological, climatological and oceanographic communities. `scores` not only includes common scores (e.g., Mean Absolute Error), it also includes novel scores not commonly found elsewhere (e.g., FIxed Risk Multicategorical (FIRM) score, Flip-Flop Index), complex scores (e.g., threshold-weighted continuous ranked probability score), and statistical tests (such as the Diebold Mariano test). It also contains isotonic regression which is becoming an increasingly important tool in forecast verification and can be used to generate stable reliability diagrams. Additionally, it provides pre-processing tools for preparing data for scores in a variety of formats including cumulative distribution functions (CDF). At the time of writing, `scores` includes over 50 metrics, statistical techniques and data processing tools. All of the scores and statistical techniques in this package have undergone a thorough scientific and software review. Every score has a companion Jupyter Notebook tutorial that demonstrates its use in practice. `scores` supports `xarray` datatypes, allowing it to work with Earth system data in a range of formats including NetCDF4, HDF5, Zarr and GRIB among others. `scores` uses Dask for scaling and performance. Support for `pandas` is being introduced. The `scores` software repository can be found at https://github.com/nci/scores/, Comment: Minor revisions to text and table. Updated title. 6 pages, 1 table. Software repository at https://github.com/nci/scores/

We demonstrate a method to generate photons by injecting hot electrons into a {\it pn} junction within a \ce{GaAs/AlGaAs} heterostructure. Hot electrons are generated by biasing across a mesoscopic potential in {\it n}-type region and travel toward {\it p}-type region through quantum Hall edge channel in the presence of magnetic field perpendicular to the substrate. The {\it p}-type region is created several microns away from the hot electron emitter by inducing interfacial charges using a surface gate. The energy relaxation of the hot electrons is suppressed by separating the orbitals before and after longitudinal-optical (LO) phonon emission. This technique enables the hot electrons to reach the {\it p}-type region and to recombine with induced holes followed by photon emissions. Hot electron-induced hole recombination is confirmed by a peak around \qty{810}{nm} in an optical spectrum that corresponds to excitonic recombination in a \ce{GaAs} quantum well. An asymmetric structure observed in the optical spectrum as a function of the magnetic field originates from the chiral transport of the hot electrons in the Hall edge channel. We propose the combination of our technology and on-demand single-electron source would enable the development of an on-demand single photon source that is an essential building block to drive an optical quantum circuit and to transfer quantum information for a long distance., Comment: 8 pages, 4 figures

A good teacher should not only be knowledgeable; but should be able to communicate in a way that the student understands -- to share the student's representation of the world. In this work, we integrate insights from machine teaching and pragmatic communication with the burgeoning literature on representational alignment to characterize a utility curve defining a relationship between representational alignment and teacher capability for promoting student learning. To explore the characteristics of this utility curve, we design a supervised learning environment that disentangles representational alignment from teacher accuracy. We conduct extensive computational experiments with machines teaching machines, complemented by a series of experiments in which machines teach humans. Drawing on our findings that improved representational alignment with a student improves student learning outcomes (i.e., task accuracy), we design a classroom matching procedure that assigns students to teachers based on the utility curve. If we are to design effective machine teachers, it is not enough to build teachers that are accurate -- we want teachers that can align, representationally, to their students too., Comment: Preprint

Autoregressive language models have demonstrated a remarkable ability to extract latent structure from text. The embeddings from large language models have been shown to capture aspects of the syntax and semantics of language. But what {\em should} embeddings represent? We connect the autoregressive prediction objective to the idea of constructing predictive sufficient statistics to summarize the information contained in a sequence of observations, and use this connection to identify three settings where the optimal content of embeddings can be identified: independent identically distributed data, where the embedding should capture the sufficient statistics of the data; latent state models, where the embedding should encode the posterior distribution over states given the data; and discrete hypothesis spaces, where the embedding should reflect the posterior distribution over hypotheses given the data. We then conduct empirical probing studies to show that transformers encode these three kinds of latent generating distributions, and that they perform well in out-of-distribution cases and without token memorization in these settings., Comment: 15 pages, 8 figures

Categories can be represented at different levels of abstraction, from prototypes focused on the most typical members to remembering all observed exemplars of the category. These representations have been explored in the context of supervised learning, where stimuli are presented with known category labels. We examine the benefits of prototype-based representations in a less-studied domain: semi-supervised learning, where agents must form unsupervised representations of stimuli before receiving category labels. We study this problem in a Bayesian unsupervised learning model called a variational auto-encoder, and we draw on recent advances in machine learning to implement a prior that encourages the model to use abstract prototypes to represent data. We apply this approach to image datasets and show that forming prototypes can improve semi-supervised category learning. Additionally, we study the latent embeddings of the models and show that these prototypes allow the models to form clustered representations without supervision, contributing to their success in downstream categorization performance., Comment: 7 pages, 3 figures

As Large Language Models (LLMs) are increasingly deployed in real-world settings, understanding the knowledge they implicitly use when making decisions is critical. One way to capture this knowledge is in the form of Bayesian prior distributions. We develop a prompt-based workflow for eliciting prior distributions from LLMs. Our approach is based on iterated learning, a Markov chain Monte Carlo method in which successive inferences are chained in a way that supports sampling from the prior distribution. We validated our method in settings where iterated learning has previously been used to estimate the priors of human participants -- causal learning, proportion estimation, and predicting everyday quantities. We found that priors elicited from GPT-4 qualitatively align with human priors in these settings. We then used the same method to elicit priors from GPT-4 for a variety of speculative events, such as the timing of the development of superhuman AI.

We present an efficient machine learning based automated framework for the fast tuning of single-electron pump devices into current quantization regimes. It uses a sparse measurement approach based on an iterative active learning algorithm to take targeted measurements in the gate voltage parameter space. When compared to conventional parameter scans, our automated framework allows us to decrease the number of measurement points by about an order of magnitude. This corresponds to an eight-fold decrease in the time required to determine quantization errors, which are estimated via an exponential extrapolation of the first current plateau embedded into the algorithm. We show the robustness of the framework by characterizing 28 individual devices arranged in a GaAs/AlGaAs multiplexer array, which we use to identify a subset of devices suitable for parallel operation at communal gate voltages. The method opens up the possibility to efficiently scale the characterization of such multiplexed devices to a large number of pumps., Comment: 6 pages, 3 figures

The article `Bell Nonlocality in Classical Systems Coexisting with Other System Types' by Chiribella et al. defines `classical' in a quantum context in a way that ignores noncommuting quantum projectors, and is hence inconsistent with Hilbert-space quantum theory., Comment: 3 pages

Humans rely on strong inductive biases to learn from few examples and abstract useful information from sensory data. Instilling such biases in machine learning models has been shown to improve their performance on various benchmarks including few-shot learning, robustness, and alignment. However, finding effective training procedures to achieve that goal can be challenging as psychologically-rich training data such as human similarity judgments are expensive to scale, and Bayesian models of human inductive biases are often intractable for complex, realistic domains. Here, we address this challenge by introducing a Bayesian notion of generative similarity whereby two datapoints are considered similar if they are likely to have been sampled from the same distribution. This measure can be applied to complex generative processes, including probabilistic programs. We show that generative similarity can be used to define a contrastive learning objective even when its exact form is intractable, enabling learning of spatial embeddings that express specific inductive biases. We demonstrate the utility of our approach by showing how it can be used to capture human inductive biases for geometric shapes, and to better distinguish different abstract drawing styles that are parameterized by probabilistic programs.

The observed similarities in the behavior of humans and Large Language Models (LLMs) have prompted researchers to consider the potential of using LLMs as models of human cognition. However, several significant challenges must be addressed before LLMs can be legitimately regarded as cognitive models. For instance, LLMs are trained on far more data than humans typically encounter, and may have been directly trained on human data in specific cognitive tasks or aligned with human preferences. Consequently, the origins of these behavioral similarities are not well understood. In this paper, we propose a novel way to enhance the utility of LLMs as cognitive models. This approach involves (i) leveraging computationally equivalent tasks that both an LLM and a rational agent need to master for solving a cognitive problem and (ii) examining the specific task distributions required for an LLM to exhibit human-like behaviors. We apply this approach to decision-making -- specifically risky and intertemporal choice -- where the key computationally equivalent task is the arithmetic of expected value calculations. We show that an LLM pretrained on an ecologically valid arithmetic dataset, which we call Arithmetic-GPT, predicts human behavior better than many traditional cognitive models. Pretraining LLMs on ecologically valid arithmetic datasets is sufficient to produce a strong correspondence between these models and human decision-making. Our results also suggest that LLMs used as cognitive models should be carefully investigated via ablation studies of the pretraining data.

The Galactic center lobe (GCL) is a $\sim 1^\circ$ object located north of the Galactic center. In the mid-infrared (MIR), the GCL appears as two 8.0-micron filaments that roughly define an ellipse. There is strong 24-micron and radio continuum emission in the interior of the ellipse. Due to its morphology and location in the sky, previous authors have argued that the GCL is created by outflows from star formation in the central molecular zone or by activity of the central black hole Sgr~A$^*$. We present images of the GCL from the GaLactic and Extragalactic All-sky Murchison Widefield Array survey in radio continuum that show thermal absorption against the Galactic center, incompatible with an interpretation of synchrotron self-absorption. Estimates of the cosmic ray emissivity in this direction allow us to place a distance constraint on the GCL. To be consistent with standard emissivity assumptions, the GCL would be located 2kpc away. At a distance of 8kpc, the synchrotron background emissivity is enhanced by $\sim75$% in the direction of the GCL. We also present radio recombination line data from the Green Bank Telescope that constrains the electron temperature and line widths in this region, which are also more explicable if the GCL lies relatively close., Comment: 6 figures, accepted to ApJ

The Galactic center lobe (GCL) is an object ~1{\deg} across that is located north of the Galactic center. In the mid-infrared (MIR) the GCL appears as two 8.0${\mu}$m filaments between which is strong 24${\mu}$m and radio continuum emission. Due to its morphology and location in the sky, previous authors have argued that the GCL is located in the Galactic center region, created by outflows from star formation or by activity of the central black hole Sagittarius A*. In an associated paper (Hurley-Walker et al., 2024, in press), low-frequency radio emission indicates that the GCL must instead lie foreground to the Galactic center. If the GCL is foreground to the Galactic center, it is likely to be a type of object common throughout the Galactic disk; we here investigate whether its properties are similar to those of Galactic HII regions. We find that the GCL's MIR morphology, MIR flux densities, dust temperatures, and radio recombination line (RRL) properties as traced by the GBT Diffuse Ionized Gas Survey (GDIGS) are consistent with those of known Galactic HII regions, although the derived electron temperature is low. We search for the ionizing source(s) of the possible HII region and identify a stellar cluster candidate (Camargo #1092/Ryu & Lee #532) and a cluster of young stellar objects (SPICY G359.3+0.3) whose members have Gaia parallaxes distances of 1.7${\pm}$0.4kpc. Taken together, the results of our companion paper and those shown here suggest that the GCL has properties consistent with those of an HII region located ~2kpc from the Sun., Comment: Accepted to ApJ

The COVID-19 pandemic led to a large global effort to sequence SARS-CoV-2 genomes from patient samples to track viral evolution and inform public health response. Millions of SARS-CoV-2 genome sequences have been deposited in global public repositories. The Canadian COVID-19 Genomics Network (CanCOGeN - VirusSeq), a consortium tasked with coordinating expanded sequencing of SARS-CoV-2 genomes across Canada early in the pandemic, created the Canadian VirusSeq Data Portal, with associated data pipelines and procedures, to support these efforts. The goal of VirusSeq was to allow open access to Canadian SARS-CoV-2 genomic sequences and enhanced, standardized contextual data that were unavailable in other repositories and that meet FAIR standards (Findable, Accessible, Interoperable and Reusable). The Portal data submission pipeline contains data quality checking procedures and appropriate acknowledgement of data generators that encourages collaboration. Here we also highlight Duotang, a web platform that presents genomic epidemiology and modeling analyses on circulating and emerging SARS-CoV-2 variants in Canada. Duotang presents dynamic changes in variant composition of SARS-CoV-2 in Canada and by province, estimates variant growth, and displays complementary interactive visualizations, with a text overview of the current situation. The VirusSeq Data Portal and Duotang resources, alongside additional analyses and resources computed from the Portal (COVID-MVP, CoVizu), are all open-source and freely available. Together, they provide an updated picture of SARS-CoV-2 evolution to spur scientific discussions, inform public discourse, and support communication with and within public health authorities. They also serve as a framework for other jurisdictions interested in open, collaborative sequence data sharing and analyses.

We prove global well-posedness for the cubic nonlinear Schr\"odinger equation with nonlinearity concentrated on a homogeneous Poisson process., Comment: 23 pages