Search

We measure the association between generative AI (GAI) tool adoption and security operations center productivity. We find that GAI adoption is associated with a 30.13% reduction in security incident mean time to resolution. This result is robust to several modeling decisions. While unobserved confounders inhibit causal identification, this result is among the first to use observational data from live operations to investigate the relationship between GAI adoption and security worker productivity.

As generative AI (GAI) tools become increasingly integrated into workplace environments, it is essential to measure their impact on productivity across specific domains. This study evaluates the effects of Microsoft's Security Copilot ("Copilot") on information technology administrators ("IT admins") through randomized controlled trials. Participants were divided into treatment and control groups, with the former granted access to Copilot within Microsoft's Entra and Intune admin centers. Across three IT admin scenarios - sign-in troubleshooting, device policy management, and device troubleshooting - Copilot users demonstrated substantial improvements in both accuracy and speed. Across all scenarios and tasks, Copilot subjects experienced a 34.53% improvement in overall accuracy and a 29.79% reduction in task completion time. We also find that the productivity benefits vary by task type, with more complex tasks showing greater improvement. In free response tasks, Copilot users identified 146.07% more relevant facts and reduced task completion time by 61.14%. Subject satisfaction with Copilot was high, with participants reporting reduced effort and a strong preference for using the tool in future tasks. These findings suggest that GAI tools like Copilot can significantly enhance the productivity and efficiency of IT admins, especially in scenarios requiring information synthesis and complex decision-making., Comment: Speed results were off by a few decimal points. Qualitative results are unchanged

In the theory of matrix-valued orthogonal polynomials, there exists a longstanding problem known as the Matrix Bochner Problem: the classification of all $N \times N$ weight matrices $W(x)$ such that the associated orthogonal polynomials are eigenfunctions of a second-order differential operator. In [4], Casper and Yakimov made an important breakthrough in this area, proving that, under certain hypotheses, every solution to this problem can be obtained as a bispectral Darboux transformation of a direct sum of classical scalar weights. In the present paper, we construct three families of weight matrices $W(x)$ of size $N \times N$, associated with Hermite, Laguerre, and Jacobi weights, which can be considered 'singular' solutions to the Matrix Bochner Problem because they cannot be obtained as a Darboux transformation of classical scalar weights., Comment: 16 pages

Large Language Models (LLMs) and search engines have the potential to perpetuate biases and stereotypes by amplifying existing prejudices in their training data and algorithmic processes, thereby influencing public perception and decision-making. While most work has focused on Western-centric AI technologies, we study Chinese-based tools by investigating social biases embedded in the major Chinese search engine, Baidu, and two leading LLMs, Ernie and Qwen. Leveraging a dataset of 240 social groups across 13 categories describing Chinese society, we collect over 30k views encoded in the aforementioned tools by prompting them for candidate words describing such groups. We find that language models exhibit a larger variety of embedded views compared to the search engine, although Baidu and Qwen generate negative content more often than Ernie. We also find a moderate prevalence of stereotypes embedded in the language models, many of which potentially promote offensive and derogatory views. Our work highlights the importance of promoting fairness and inclusivity in AI technologies with a global perspective.

Nonlinear dynamics have long been exploited in order to damp vibrations in solid mechanics. The phenomenon of irreversible energy transfer from a linear primary system to a nonlinear absorber has driven great attention to the optimal design of vibration absorbers both for stationary and transient regimes. Recently, the same principle has also been targeted in acoustics for the absorption of sound waves at high excitation amplitudes. Meanwhile, acoustic absorption by electro-active means has found great success for noise reduction in the linear regime. This study uses a method allowing the design of nonlinear resonators at amplitudes that typically induce linear behaviors. This research proposes an analytical study of the implementation of the duffing equation as a nonlinear electroacoustic resonator coupled to an acoustic mode of a tube. Experiments are carried out and compared to the analytical results. The experimental implementation is done using a real-time-based algorithm retrieving the measured pressure from a microphone and giving the electrical current to send to a loudspeaker as an output thanks to a Runge-Kutta-like algorithm., Comment: Preprint of a submitted article

The new generation of Ultra-High-By-Pass-Ratio (UHBR) turbofan engine while considerably reducing fuel consumption, threatens higher noise levels at low frequencies because of its larger diameter, lower number of blades and rotational speed. This is accompanied by a shorter nacelle, leaving less available space for acoustic treatments. In this context, a progress in the liner technology is highly demanded, prospecting alternative solutions to classic liners. The SALUTE H2020 project has taken up this challenge, proposing electro-active acoustic liners, made up of loudspeakers (actuators) and microphones (sensors). The electro-active means allow to program the surface impedance on the electroacoustic liner, but also to conceive alternative boundary laws. Test-rigs of gradually increasing complexities have allowed to raise the Technology Readiness Level (TRL) up to 3-4. In this second part, the electroacoustic liner is adapted to treat the walls at the intake of a scaled turbofan nacelle. The performance of the electroacoustic liner demonstrate its potentialities for reducing noise radiation from turbofan., Comment: arXiv admin note: text overlap with arXiv:2403.10426

The new generation of Ultra-High-By-Pass-Ratio (UHBR) turbofan engine while considerably reducing fuel consumption, threatens higher noise levels at low frequencies because of its larger diameter, lower number of blades and rotational speed. This is accompanied by a shorter nacelle, leaving less available space for acoustic treatments. In this context, a progress in the liner technology is highly demanded, prospecting alternative solutions to classic liners. The SALUTE H2020 project has taken up this challenge, proposing electro-active acoustic liners, made up of loudspeakers (actuators) and microphones (sensors). The electro-active means allow to program the surface impedance on the electroacoustic liner, but also to conceive alternative boundary laws. Test-rigs of gradually increasing complexities have allowed to raise the Technology Readiness Level (TRL) up to 3-4. In this first part, we describe the control laws employed in the experimental campaigns, and present the scattering performances in rectangular waveguides with monomodal guided propagation. These results have assessed the isolation capabilities, stability and robustness of such programmable boundary technology, allowing to gain confidence for the successive implementation in a scaled turbofan test-rig., Comment: arXiv admin note: substantial text overlap with arXiv:2403.10426

In this work, we give some criteria that allow us to decide when two sequences of matrix-valued orthogonal polynomials are related via a Darboux transformation and to build explicitly such transformation. In particular, they allow us to see when and how any given sequence of polynomials is Darboux related to a diagonal matrix of classic orthogonal polynomials. We also explore the notion of Darboux-irreducibility and study some sequences that are not a Darboux transformation of classical orthogonal polynomials.

The task of sampling efficiently the Gibbs-Boltzmann distribution of disordered systems is important both for the theoretical understanding of these models and for the solution of practical optimization problems. Unfortunately, this task is known to be hard, especially for spin glasses at low temperatures. Recently, many attempts have been made to tackle the problem by mixing classical Monte Carlo schemes with newly devised Neural Networks that learn to propose smart moves. In this article we introduce the Nearest-Neighbours Neural Network (4N) architecture, a physically-interpretable deep architecture whose number of parameters scales linearly with the size of the system and that can be applied to a large variety of topologies. We show that the 4N architecture can accurately learn the Gibbs-Boltzmann distribution for the two-dimensional Edwards-Anderson model, and specifically for some of its most difficult instances. In particular, it captures properties such as the energy, the correlation function and the overlap probability distribution. Finally, we show that the 4N performance increases with the number of layers, in a way that clearly connects to the correlation length of the system, thus providing a simple and interpretable criterion to choose the optimal depth., Comment: 10 pages, 6 figures; SI: 3 pages

We introduce Java-Class-Hijack, a novel software supply chain attack that enables an attacker to inject malicious code by crafting a class that shadows a legitimate class that is in the dependency tree. We describe the attack, provide a proof-of-concept demonstrating its feasibility, and replicate it in the German Corona-Warn-App server application. The proof-of-concept illustrates how a transitive dependency deep within the dependency tree can hijack a class from a direct dependency and entirely alter its behavior, posing a significant security risk to Java applications. The replication on the Corona-Warn-App demonstrates how compromising a small JSON validation library could result in a complete database takeover., Comment: 6 pages, added a mitigation chapter

We present the first application of data-driven techniques for dynamical system analysis based on Koopman theory to variable stars. We focus on light curves of RRLyrae type variables, in the Galactic globular cluster $\omega$ Centauri. Light curves are thus summarized by a handful of complex eigenvalues, corresponding to oscillatory or fading dynamical modes. We find that variable stars of the RRc subclass can be summarized in terms of fewer ($\approx 8$) eigenvalues, while RRab need comparatively more ($\approx 12$). This result can be leveraged for classification and reflects the simpler structure of RRc light curves. We then consider variable stars displaying secular variations due to the Tseraskaya-Blazhko effect and find a change in relevant eigenvalues with time, with possible implications for the physical interpretation of the effect., Comment: Accepted (spotlight) at the Scaling AI for Science workshop at ICML 2024

Continuous-time dynamic graphs (CTDGs) are essential for modeling interconnected, evolving systems. Traditional methods for extracting knowledge from these graphs often depend on feature engineering or deep learning. Feature engineering is limited by the manual and time-intensive nature of crafting features, while deep learning approaches suffer from high inference latency, making them impractical for real-time applications. This paper introduces Deep-Graph-Sprints (DGS), a novel deep learning architecture designed for efficient representation learning on CTDGs with low-latency inference requirements. We benchmark DGS against state-of-the-art (SOTA) feature engineering and graph neural network methods using five diverse datasets. The results indicate that DGS achieves competitive performance while inference speed improves between 4x and 12x compared to other deep learning approaches on our benchmark datasets. Our method effectively bridges the gap between deep representation learning and low-latency application requirements for CTDGs.

The Extremely Large Telescopes (ELTs), thanks to their large apertures and cutting-edge Multi-Conjugate Adaptive Optics (MCAO) systems, promise to deliver sharper and deeper data even than the JWST. SHARP is a concept study for a near-IR (0.95-2.45 $\mu$m) spectrograph conceived to fully exploit the collecting area and the angular resolution of the upcoming generation of ELTs. In particular, SHARP is designed for the 2nd port of MORFEO@ELT. Composed of a Multi-Object Spectrograph, NEXUS, and a multi-Integral Field Unit, VESPER, MORFEO-SHARP will deliver high angular ($\sim$30 mas) and spectral (R$\simeq$300, 2000, 6000, 17000) resolution, outperforming NIRSpec@JWST (100 mas). SHARP will enable studies of the nearby Universe and the early Universe in unprecedented detail. NEXUS is fed by a configurable slit system deploying up to 30 slits with $\sim$2.4 arcsec length and adjustable width, over a field of about 1.2"$\times$1.2" (35 mas/pix). Each slit is fed by an inversion prism able to rotate by an arbitrary angle the field that can be seen by the slit. VESPER is composed of 12 probes of 1.7"$\times$1.5" each (spaxel 31 mas) probing a field 24"$\times$70". SHARP is conceived to exploit the ELTs apertures reaching the faintest flux and the sharpest angular resolution by joining the sensitivity of NEXUS and the high spatial sampling of VESPER to MORFEO capabilities. This article provides an overview of the scientific design drivers, their solutions, and the resulting optical design of the instrument achieving the required optical performance., Comment: Proceedings of the SPIE Astronomical Telescopes and Instrumentation 2024, Volume 13096, Paper No. 130965I, 11 pp, 11 figs

In the ever-evolving landscape of scientific computing, properly supporting the modularity and complexity of modern scientific applications requires new approaches to workflow execution, like seamless interoperability between different workflow systems, distributed-by-design workflow models, and automatic optimisation of data movements. In order to address this need, this article introduces SWIRL, an intermediate representation language for scientific workflows. In contrast with other product-agnostic workflow languages, SWIRL is not designed for human interaction but to serve as a low-level compilation target for distributed workflow execution plans. The main advantages of SWIRL semantics are low-level primitives based on the send/receive programming model and a formal framework ensuring the consistency of the semantics and the specification of translating workflow models represented by Directed Acyclic Graphs (DAGs) into SWIRL workflow descriptions. Additionally, SWIRL offers rewriting rules designed to optimise execution traces, accompanied by corresponding equivalence. An open-source SWIRL compiler toolchain has been developed using the ANTLR Python3 bindings.

How to distribute a set of points uniformly on a spherical surface is a very old problem that still lacks a definite answer. In this work, we introduce a physical measure of uniformity based on the distribution of distances between points, as an alternative to commonly adopted measures based on interaction potentials. We then use this new measure of uniformity to characterize several algorithms available in the literature. We also study the effect of optimizing the position of the points through the minimization of different interaction potentials via a gradient descent procedure. In this way, we can classify different algorithms and interaction potentials to find the one that generates the most uniform distribution of points on the sphere., Comment: 17 pages, 15 figures

Results regarding spin glass models are, to this day, mainly confined to models with Ising spins. Spin glass models with continuous spins exhibit interesting new physical behaviors related to the additional degrees of freedom, but have been primarily studied on fully connected topologies. Only recently some advancements have been made in the study of continuous models on sparse graphs. In this work we partially fill this void by introducing a method to solve numerically the Belief Propagation equations for systems of Heisenberg spins on sparse random graphs via a discretization of the sphere. We introduce techniques to study the finite-temperature, finite-connectivity case as well as novel algorithms to deal with the zero-temperature and large connectivity limits. As an application, we locate the de Almeida-Thouless line for this class of models and the critical field at zero temperature, showing the full consistency of the methods presented. Beyond the specific results reported for Heisenberg models, the approaches presented in this paper have a much broader scope of application and pave the way to the solution of strongly disordered models with continuous variables., Comment: 28 pages, 15 figures; added references

This work is part of VESTALE, a project initiated within the Rubin-LSST Cadence Strategy Optimization Process . Its goal is to explore the potential of Rubin-LSST observations aimed at the Galaxy's bulge (Bulge) for studying RR Lyrae stars (RRL). Observation and analysis of RR Lyrae stars in the Bulge are crucial for tracing the old population of the central part of our galaxy and reconstructing the history of Bulge formation. Based on observations conducted with CTIO/DECam by Saha et al. 2019 towards the Baade Window, our simulations demonstrate that early Rubin-LSST observations will enable the recovery of RR Lyrae light curves at Galactic center distances with sufficient precision. This will allow us to utilize theoretical relations from Marconi et al. 2022 to determine their distances and/or metallicity, following the REDIME algorithm introduced in Bono et al. 2019. We show how reddening and crowding affect our simulations and highlight the importance of considering these effects when deriving pulsation parameters (luminosity amplitudes, mean magnitudes) based on the light curves especially if the goal is to explore the opposite side of the Bulge through the observation of its RRL. The simulations discussed in this investigation were conducted to support the SCOC's decision to observe this important sky region since it has only recently been decided to include part of the Bulge as a target within the LSST main survey., Comment: Accepted for publication in ApJS. 12 pages and 6 figures

We review the phenomenology of classical Cepheids (CCs), Anomalous Cepheids (ACs) and type II Cepheids (TIICs) in the Milky Way (MW) and in the Magellanic Clouds (MCs). We also examine the Hertzsprung progression in different stellar systems by using the shape of I-band light curves (Fourier parameters) and observables based on the difference in magnitude and in phase between the bump and the minimum in luminosity. The distribution of Cepheids in optical and in optical-near infrared (NIR) color--magnitude diagrams is investigated to constrain the topology of the instability strip. The use of Cepheids as tracers of young (CCs), intermediate (ACs) and old (TIICs) stellar populations are brought forward by the comparison between observations (MCs) and cluster isochrones covering a broad range in stellar ages and in chemical compositions. The different diagnostics adopted to estimate individual distances (period--luminosity, period--Wesenheit, period--luminosity--color relations) are reviewed together with pros and cons in the use of fundamental and overtones, optical and NIR photometric bands, and reddening free pseudo magnitudes (Wesenheit). We also discuss the use of CCs as stellar tracers and the radial gradients among the different groups of elements (iron, alpha, neutron-capture) together with their age-dependence. Finally, we briefly outline the role that near-future space and ground-based facilities will play in the astrophysical and cosmological use of Cepheids., Comment: Published in The Astronomy and Astrophysics Review Volume 32, article number 4

Stellar mergers and accretion events have been crucial in shaping the evolution of the Milky Way (MW). These events have been dynamically identified and chemically characterised using red giants and main-sequence stars. RR Lyrae (RRL) variables can play a crucial role in tracing the early formation of the MW since they are ubiquitous, old (t$\ge$10 Gyr) low-mass stars and accurate distance indicators. We exploited Data Release 3 of the GALAH survey to identify 78 field RRLs suitable for chemical analysis. Using synthetic spectra calculations, we determined atmospheric parameters and abundances of Fe, Mg, Ca, Y, and Ba. Most of our stars exhibit halo-like chemical compositions, with an iron peak around [Fe/H]$\approx -$1.40, and enhanced Ca and Mg content. Notably, we discovered a metal-rich tail, with [Fe/H] values ranging from $-$1 to approximately solar metallicity. This sub-group includes almost ~1/4 of the sample, it is characterised by thin disc kinematics and displays sub-solar $\alpha$-element abundances, marginally consistent with the majority of the MW stars. Surprisingly, they differ distinctly from typical MW disc stars in terms of the s-process elements Y and Ba. We took advantage of similar data available in the literature and built a total sample of 535 field RRLs for which we estimated kinematical and dynamical properties. We found that metal-rich RRLs (1/3 of the sample) likely represent an old component of the MW thin disc. We also detected RRLs with retrograde orbits and provided preliminary associations with the Gaia-Sausage-Enceladus, Helmi, Sequoia, Sagittarius, and Thamnos stellar streams., Comment: Accepted for publication in MNRAS. 29 pages, 20 figures

Benchmarks have emerged as the central approach for evaluating Large Language Models (LLMs). The research community often relies on a model's average performance across the test prompts of a benchmark to evaluate the model's performance. This is consistent with the assumption that the test prompts within a benchmark represent a random sample from a real-world distribution of interest. We note that this is generally not the case; instead, we hold that the distribution of interest varies according to the specific use case. We find that (1) the correlation in model performance across test prompts is non-random, (2) accounting for correlations across test prompts can change model rankings on major benchmarks, (3) explanatory factors for these correlations include semantic similarity and common LLM failure points.

The increasing use of polymer composites in industry asks for the creation of better, faster and cost-effective methods to detect the damage state of such materials. This work presents the investigation of the phase decay , $\Delta{\Phi}$, as a new parameter to characterise crack growth in composites materials utilising an experimental framework of High Frequency Fatigue Testing (HFFT), a framework where the excitation occurs at vibration resonance. The proposed methodology empirically relates the crack growth measurements, from interrupted testing, with the structural phase decay response, distinctive of material strength degradation, Comment: 13 pages, 10 figures, to be published in the International Journal of Fatigue

This paper presents a new software framework for HRI experimentation with the sixth version of the common NAO robot produced by the United Robotics Group. Embracing the common demand of researchers for better performance and new features for NAO, the authors took advantage of the ability to run ROS2 onboard on the NAO to develop a framework independent of the APIs provided by the manufacturer. Such a system provides NAO with not only the basic skills of a humanoid robot such as walking and reproducing movements of interest but also features often used in HRI such as: speech recognition/synthesis, face and object detention, and the use of Generative Pre-trained Transformer (GPT) models for conversation. The developed code is therefore configured as a ready-to-use but also highly expandable and improvable tool thanks to the possibilities provided by the ROS community., Comment: 7 pages, 3 figures

Binary Cepheids with giant companions are crucial for studying the physical properties of Cepheid variables, providing the best means to measure their masses. Systems composed of two Cepheids are even more important but to date, only one such system in the Large Magellanic Cloud (LMC) was known. Our current aim is to increase the number of these systems tenfold and provide their basic characteristics. The final goal is to obtain the physical properties of the component Cepheids, including their masses and radii, and to learn about their evolution in the multiple systems, also revealing their origin. We started a spectroscopic monitoring of nine unresolved pairs of Cepheids from the OGLE catalog, to check if they are gravitationally bound. Two of these so-called double Cepheids are located in the LMC, five in the Small Magellanic Cloud (SMC), and two in the Milky Way (MW). We report the spectroscopic detection of binarity of all 9 of these double Cepheids with orbital periods from 2 to 18 years. This increases the number of known binary double (BIND) Cepheids from 1 to 10 and triples the number of all confirmed double-lined binary (SB2) Cepheids. For five BIND Cepheids disentangled pulsational light curves of the components show anti-correlated phase shifts due to orbital motion. We show the first empirical evidence that typical period-luminosity relations (PLRs) are rather binary Cepheid PLRs that include the companion's light. The statistics of pulsation period ratios of BIND Cepheids do not agree with those expected for pairs of the same-age Cepheids. These ratios together with the mass ratios far from unity suggest merger-origin of at least one component for about half of the systems. The SMC and MW objects are the first found in SB2 systems composed of giants in their host galaxies. The Milky Way BIND Cepheids are also the closest such systems, being located at about 11 and 26 kpc., Comment: 13 pages, 7 figures, 4 tables, accepted for publication in Astronomy & Astrophysics

Sound attenuation along a waveguide is intensively studied for applications ranging from heating and air-conditioning ventilation systems, to aircraft turbofan engines. In particular, the new generation of Ultra-High-By-Pass-Ratio turbofan requires higher attenuation at low frequencies, in less space for liner treatment. This demands to go beyond the classical acoustic liner concepts and overcome their limitations. In this paper, we discuss an unconventional boundary operator, called Advection Boundary Law, which can be artificially synthesized by electroactive means, such as Electroacoustic Resonators. This boundary condition entails nonreciprocal propagation, meanwhile enhancing noise transmission attenuation with respect to purely locally-reacting boundaries, along one sense of propagation. Because of its artificial nature though, its acoustical passivity limits are yet to be defined. In this paper, we provide a thorough numerical study to assess the performances of the Advection Boundary Law, in absence of mean flow. An experimental test-bench validates the numerical outcomes in terms of passivity limits, non-reciprocal propagation and enhanced isolation with respect to local impedance operators. This work provides the guidelines to properly implement the Advection Boundary Law for optimal noise transmission attenuation. Moreover, the tools and criteria provided here can also be employed for the design and characterization of other innovative liners.

The Wide-field Spectroscopic Telescope (WST) is proposed as a new facility dedicated to the efficient delivery of spectroscopic surveys. This white paper summarises the initial concept as well as the corresponding science cases. WST will feature simultaneous operation of a large field-of-view (3 sq. degree), a high multiplex (20,000) multi-object spectrograph (MOS) and a giant 3x3 sq. arcmin integral field spectrograph (IFS). In scientific capability these requirements place WST far ahead of existing and planned facilities. Given the current investment in deep imaging surveys and noting the diagnostic power of spectroscopy, WST will fill a crucial gap in astronomical capability and work synergistically with future ground and space-based facilities. This white paper shows that WST can address outstanding scientific questions in the areas of cosmology; galaxy assembly, evolution, and enrichment, including our own Milky Way; origin of stars and planets; time domain and multi-messenger astrophysics. WST's uniquely rich dataset will deliver unforeseen discoveries in many of these areas. The WST Science Team (already including more than 500 scientists worldwide) is open to the all astronomical community. To register in the WST Science Team please visit https://www.wstelescope.com/for-scientists/participate, Comment: 194 pages, 66 figures. Comments are welcome (wstelescope@gmail.com)

The Matrix Bochner Problem aims to classify weight matrices $W$ such that its algebra $\mathcal D(W)$, of all differential operators that have a sequence of these matrix orthogonal polynomials as eigenfunctions, contains a second-order differential operator. In [5] it is proven that, under certain assumptions, the solutions to the Matrix Bochner Problem can be obtained through a noncommutative bispectral Darboux transformation of some classical scalar weights. The main aim of this paper is to introduce the concept of strong Darboux transformation among weight matrices and explore the relationship between the algebras $\mathcal{D}(W)$ and $\mathcal{D}(\widetilde{W})$ when $\widetilde{W}$ is a strong Darboux transformation of $W$. Starting from a direct sum of classical scalar weights $\widetilde W$, and leveraging our complete knowledge of the algebra of $\mathcal D(\widetilde W)$, we can easily determine the algebra $\mathcal D(W)$ of a weight $W$ that is a strong Darboux transformation of $\widetilde W$., Comment: 23 pages. arXiv admin note: text overlap with arXiv:2311.16325

We present details on a new measurement of the muon magnetic anomaly, $a_\mu = (g_\mu -2)/2$. The result is based on positive muon data taken at Fermilab's Muon Campus during the 2019 and 2020 accelerator runs. The measurement uses $3.1$ GeV$/c$ polarized muons stored in a $7.1$-m-radius storage ring with a $1.45$ T uniform magnetic field. The value of $ a_{\mu}$ is determined from the measured difference between the muon spin precession frequency and its cyclotron frequency. This difference is normalized to the strength of the magnetic field, measured using Nuclear Magnetic Resonance (NMR). The ratio is then corrected for small contributions from beam motion, beam dispersion, and transient magnetic fields. We measure $a_\mu = 116 592 057 (25) \times 10^{-11}$ (0.21 ppm). This is the world's most precise measurement of this quantity and represents a factor of $2.2$ improvement over our previous result based on the 2018 dataset. In combination, the two datasets yield $a_\mu(\text{FNAL}) = 116 592 055 (24) \times 10^{-11}$ (0.20 ppm). Combining this with the measurements from Brookhaven National Laboratory for both positive and negative muons, the new world average is $a_\mu$(exp) $ = 116 592 059 (22) \times 10^{-11}$ (0.19 ppm)., Comment: 48 pages, 29 figures; 4 pages of Supplement Material; version accepted for publication in Physical Review D

Solutions to Markov Decision Processes (MDP) are often very sensitive to state transition probabilities. As the estimation of these probabilities is often inaccurate in practice, it is important to understand when and how Reinforcement Learning (RL) agents generalize when transition probabilities change. Here we present a new methodology to evaluate such generalization of RL agents under small shifts in the transition probabilities. Specifically, we evaluate agents in new environments (MDPs) in the vicinity of the training MDP created by adding quantifiable, parametric noise into the transition function of the training MDP. We refer to this process as Noise Injection, and the resulting environments as $\delta$-environments. This process allows us to create controlled variations of the same environment with the level of the noise serving as a metric of distance between environments. Conventional wisdom suggests that training and testing on the same MDP should yield the best results. However, we report several cases of the opposite -- when targeting a specific environment, training the agent in an alternative noise setting can yield superior outcomes. We showcase this phenomenon across $60$ different variations of ATARI games, including PacMan, Pong, and Breakout.

In the context of autonomous navigation of terrestrial robots, the creation of realistic models for agent dynamics and sensing is a widespread habit in the robotics literature and in commercial applications, where they are used for model based control and/or for localization and mapping. The more recent Embodied AI literature, on the other hand, focuses on modular or end-to-end agents trained in simulators like Habitat or AI-Thor, where the emphasis is put on photo-realistic rendering and scene diversity, but high-fidelity robot motion is assigned a less privileged role. The resulting sim2real gap significantly impacts transfer of the trained models to real robotic platforms. In this work we explore end-to-end training of agents in simulation in settings which minimize the sim2real gap both, in sensing and in actuation. Our agent directly predicts (discretized) velocity commands, which are maintained through closed-loop control in the real robot. The behavior of the real robot (including the underlying low-level controller) is identified and simulated in a modified Habitat simulator. Noise models for odometry and localization further contribute in lowering the sim2real gap. We evaluate on real navigation scenarios, explore different localization and point goal calculation methods and report significant gains in performance and robustness compared to prior work.

Navigation has been classically solved in robotics through the combination of SLAM and planning. More recently, beyond waypoint planning, problems involving significant components of (visual) high-level reasoning have been explored in simulated environments, mostly addressed with large-scale machine learning, in particular RL, offline-RL or imitation learning. These methods require the agent to learn various skills like local planning, mapping objects and querying the learned spatial representations. In contrast to simpler tasks like waypoint planning (PointGoal), for these more complex tasks the current state-of-the-art models have been thoroughly evaluated in simulation but, to our best knowledge, not yet in real environments. In this work we focus on sim2real transfer. We target the challenging Multi-Object Navigation (Multi-ON) task and port it to a physical environment containing real replicas of the originally virtual Multi-ON objects. We introduce a hybrid navigation method, which decomposes the problem into two different skills: (1) waypoint navigation is addressed with classical SLAM combined with a symbolic planner, whereas (2) exploration, semantic mapping and goal retrieval are dealt with deep neural networks trained with a combination of supervised learning and RL. We show the advantages of this approach compared to end-to-end methods both in simulation and a real environment and outperform the SOTA for this task.

Model interpretability plays a central role in human-AI decision-making systems. Ideally, explanations should be expressed using human-interpretable semantic concepts. Moreover, the causal relations between these concepts should be captured by the explainer to allow for reasoning about the explanations. Lastly, explanation methods should be efficient and not compromise the performance of the predictive task. Despite the rapid advances in AI explainability in recent years, as far as we know to date, no method fulfills these three properties. Indeed, mainstream methods for local concept explainability do not produce causal explanations and incur a trade-off between explainability and prediction performance. We present DiConStruct, an explanation method that is both concept-based and causal, with the goal of creating more interpretable local explanations in the form of structural causal models and concept attributions. Our explainer works as a distillation model to any black-box machine learning model by approximating its predictions while producing the respective explanations. Because of this, DiConStruct generates explanations efficiently while not impacting the black-box prediction task. We validate our method on an image dataset and a tabular dataset, showing that DiConStruct approximates the black-box models with higher fidelity than other concept explainability baselines, while providing explanations that include the causal relations between the concepts., Comment: Accepted at Conference on Causal Learning and Reasoning (CLeaR 2024, https://www.cclear.cc/2024). To be published at Proceedings of Machine Learning Research (PMLR)

It is known that a player in a noncooperative game can benefit by publicly restricting his possible moves before play begins. We show that, more generally, a player may benefit by publicly committing to pay an external party an amount that is contingent on the game's outcome. We explore what happens when external parties -- who we call ``game miners'' -- discover this fact and seek to profit from it by entering an outcome-contingent contract with the players. We analyze various structured bargaining games between miners and players for determining such an outcome-contingent contract. These bargaining games include playing the players against one another, as well as allowing the players to pay the miner(s) for exclusivity and first-mover advantage. We establish restrictions on the strategic settings in which a game miner can profit and bounds on the game miner's profit. We also find that game miners can lead to both efficient and inefficient equilibria., Comment: 25 pages, 1 figure