Search

Simulation-based testing is widely used to assess the reliability of Autonomous Driving Systems (ADS), but its effectiveness is limited by the operational design domain (ODD) conditions available in such simulators. To address this limitation, in this work, we explore the integration of generative artificial intelligence techniques with physics-based simulators to enhance ADS system-level testing. Our study evaluates the effectiveness and computational overhead of three generative strategies based on diffusion models, namely instruction-editing, inpainting, and inpainting with refinement. Specifically, we assess these techniques' capabilities to produce augmented simulator-generated images of driving scenarios representing new ODDs. We employ a novel automated detector for invalid inputs based on semantic segmentation to ensure semantic preservation and realism of the neural generated images. We then perform system-level testing to evaluate the ADS's generalization ability to newly synthesized ODDs. Our findings show that diffusion models help increase the ODD coverage for system-level testing of ADS. Our automated semantic validator achieved a percentage of false positives as low as 3\%, retaining the correctness and quality of the generated images for testing. Our approach successfully identified new ADS system failures before real-world testing., Comment: 10 pages + 2 for references

Massive stars may form in or be captured into AGN disks. Recent 1D studies employing stellar-evolution codes have demonstrated the potential for rapid growth of such stars through accretion up to a few hundred $M_\odot$. We perform 3D radiation hydrodynamic simulations of moderately massive stars' envelopes, in order to determine the rate and critical radius $R_{\rm crit}$ of their accretion process in an isotropic gas-rich environment in the absence of luminosity-driven mass loss. We find that in the ``fast-diffusion" regime where characteristic radiative diffusion speed $c/\tau$ is faster than the gas sound speed $c_s$, the accretion rate is suppressed by feedback from gravitational and radiative advection energy flux, in addition to the stellar luminosity. Alternatively, in the ``slow-diffusion" regime where $c/\tau

Migration commonly occurs during the epoch of planet formation. For emerging gas giant planets, it proceeds concurrently with their growth through the accretion of gas from their natal protoplanetary disks. Similar migration process should also be applied to the stellar-mass black holes embedded in active galactic nucleus disks. In this work, we perform high resolution 3D and 2D numerical hydrodynamical simulations to study the migration dynamics for accreting embedded objects over the disk viscous timescales in a self-consistent manner. We find that an accreting planet embedded in a predominantly viscous disk has a tendency to migrate outward, in contrast to the inward orbital decay of non-accreting planets. 3D and 2D simulations find the consistent outward migration results for the accreting planets. Under this circumstance, the accreting planet's outward migration is mainly due to the asymmetric spiral arms feeding from the global disk into the Hill radius. This is analogous to the unsaturated corotation torque although the imbalance is due to material accretion within the libration timescale rather than diffusion onto the inner disk. In a disk with a relatively small viscosity, the accreting planets clear deep gaps near their orbits. The tendency of inward migration is recovered, albeit with suppressed rates. By performing a parameter survey with a range of disks' viscosity, we find that the transition from outward to inward migration occurs with the effective viscous efficiency factor $\alpha\sim 0.003$ for Jupiter-mass planets., Comment: 17 pages, 9 figures, Accepted by ApJ

We introduce APPLE, a novel planetary evolution code designed specifically for the study of giant exoplanet and Jovian planet evolution in the era of Galileo, Juno, and Cassini. With APPLE, state-of-the-art equations of state for hydrogen, helium, ice, and rock are integrated with advanced features to treat ice/rock cores and metals in the gaseous envelope; models for helium rain and hydrogen/helium immiscibility; detailed atmosphere boundary tables that also provide self-consistent albedos and spectra; and options to address envelope metal gradients and stably-stratified regions. Our hope is that these purpose-built features of APPLE will help catalyze the development of the next generation of giant exoplanet and Jovian planet evolutionary models., Comment: Accepted for publication in ApJ

Gravitational instability in the outskirts of Active Galactic Nuclei (AGN) disks lead to disk fragmentation and formation of super-massive (several 10^2Msun) stars with potentially long lifetimes. Alternatively, stars can be captured ex-situ and grow from gas accretion in the AGN disk. However, the number density distribution throughout the disk is limited by thermal feedback as their luminosities provide the dominant heating source. We derive equilibrium stellar surface density profiles under two limiting contexts: in the case where the stellar lifetimes are prolonged due to recycling of hydrogen rich disk gas, only the fraction of gas converted into heat is removed from the disk accretion flow. Alternatively, if stellar composition recycling is inefficient and stars can evolve off the main sequence, the disk accretion rate is quenched towards smaller radii resembling a classical star-burst disk, albeit the effective removal rate depends not only on the stellar lifetime, but also the mass of stellar remnants. For AGNs with central Supermassive Black Hole (SMBH) masses of \sim 10^6 to 10^8Msun accreting at \sim 0.1 Eddington efficiency, we estimate a total number of 10^3 to 10^5 coexisting massive stars and the rate of stellar mergers to be 10^-3 to 1 per year. We motivate the detailed study of interaction between a swarm of massive stars through hydro and N body simulations to provide better prescriptions of dynamical processes in AGN disks, and to constrain more accurate estimates of the stellar population., Comment: Published in ApJ

Constitutive models are essential for describing the mechanical behavior of materials under high temperatures and pressures, among which the Steinberg-Guinan (SG) model is widely adopted. Recent work has discovered a peculiar dual anomaly of compression-induced softening and heating-induced hardening in the elasticity of compressed vanadium [Phys. Rev. B 104, 134102 (2021)], which is beyond the capability of the SG model to describe. In this work, a modified SG constitutive model is proposed to embody such an anomalous behavior. Elemental vanadium is considered as an example to demonstrate the effectiveness of this improved model in describing the dual anomalies of mechanical elasticity. This new variant of the SG model can also be applied to other materials that present an irregular variation in the mechanical elasticity, and is important to faithfully model and simulate the mechanical response of materials under extreme conditions., Comment: 14 pages, 4 figures

Dust trapping in the global pressure bump induced by magnetospheric truncation offers a promising formation mechanism for close-in super-Earths/sub-Neptunes. These planets likely form in evolved protoplanetary discs, where the gas temperature at the expanding truncation radius become amiable to refractory solids. However, dust accumulation may alter the disc opacity such that thermal evolution is inevitable. To better understand how thermodynamics affects this planet formation pathway, we conduct a suite of local dust evolution simulations in an idealized inner disc model. Our calculations take into account self-consistent opacity-dependent temperature changes as well as dust evaporation and vapour condensation. We find that disc thermal evolution regulates dust growth and evolution, discouraging any accumulation of small particles that drives the increase of opacity and temperature. Significant retention of dust mass takes place when the disc environments allow runaway growth of large solids beyond the fragmentation barrier, where small particles are then swept up and preserved. Our results further validate dust accumulation near disc truncation as a promising mechanism to form close-in planets., Comment: Accepted by MNRAS and in production; 11 pages, 8 figures

We perform a series of 3D simulations to study the accretion of giant planet embedded in protoplanetary discs (PPDs) over gap-opening timescales. We find that the accretion mass flux mainly comes from the intermediate latitude above the disc midplane. The circumplanetary disc (CPD) for a super-thermal planet is rotation-supported up to $\sim$20-30\% of the planet Hill radius. While both mass inflow and outflow exists in the CPD midplane, the overall trend is an outflow that forms a meridional circulation with high-latitude inflows. We confirm the absence of accretion outburst from disc eccentricity excited by massive planets in our 3D simulations, contrary to the consensus of previous 2D simulations. This suggests the necessity of 3D simulations of accretion even for super-Jupiters. The accretion rates of planets measured in steady-state can be decomposed into the ``geometric" and ``density depletion" factors. Through extensive parameter survey, we identify a power-law scaling for the geometric factor $\propto q_{\rm th}^{2/3}$ for super-thermal planets ($q_{\rm th}$ being the thermal mass ratio), which transforms to $\propto q_{\rm th}^{2}$ for less massive cases. The density depletion factor is limited by the disc accretion rate for mildly super-thermal planets, and by gap-opening for highly super-thermal ones. Moderate planetary eccentricities can enhance the accretion rates by a factor of $2-3$ through making the gap shallower, but does not impact the flow geometry. We have applied our simulations results to accreting protoplanet system PDS 70 and can satisfactorily explain the accretion rate and CPD size in observations., Comment: 19 pages, 17 figures, Accepted for publication in MNRAS

We present updated atmospheric tables suitable for calculating the post-formation evolution and cooling of Jupiter and Jupiter-like exoplanets. These tables are generated using a 1D radiative transfer modeling code that incorporates the latest opacities and realistic prescriptions for stellar irradiation and ammonia clouds. To ensure the accuracy of our model parameters, we calibrate them against the measured temperature structure and geometric albedo spectrum of Jupiter, its effective temperature, and its inferred internal temperature. As a test case, we calculate the cooling history of Jupiter using an adiabatic and homogeneous interior and compare with extant models now used to evolve Jupiter and the giant planets. We find that our model reasonably matches Jupiter after evolving a hot-start initial condition to the present age of the solar system, with a discrepancy in brightness temperature/radius within two per cent. Our algorithm allows us to customize for different cloud, irradiation, and metallicity parameters. This class of boundary conditions can be used to study the evolution of solar-system giant planets and exoplanets with more complicated interior structures and non-adiabatic, inhomogeneous internal profiles., Comment: 11 pages, 5 figures. Accepted to ApJ

The traditional paradigm of viscosity-dominated evolution of protoplanetary discs has been recently challenged by magnetized disc winds. However, distinguishing wind-driven and turbulence-driven accretion through observations has been difficult. In this study, we present a novel approach to identifying their separate contribution to angular momentum transport by studying the gap and ring morphology of planet-forming discs in the ALMA continuum. We model the gap-opening process of planets in discs with both viscous evolution and wind-driven accretion by 2D multi-fluid hydrodynamical simulations. Our results show that gap-opening planets in wind-driven accreting discs generate characteristic substructures that differ from those in purely viscous discs. Specifically, we demonstrate that discs, where wind-driven accretion dominates the production of substructures, exhibit significant asymmetries. Based on the diverse outputs of mock images in the ALMA continuum, we roughly divide the planet-induced features into four regimes (moderate-viscosity dominated, moderate-wind dominated, strong-wind dominated, inviscid). The classification of these regimes sets up a potential method to constrain the strength of magnetized disc wind and viscosity based on the observed gap and ring morphology. We discuss the asymmetry feature in our mock images and its potential manifestation in ALMA observations., Comment: 11 pages, 6 figures, accepted by MNRAS. Welcome any comments and suggestions!

Accretion discs around super-massive black holes (SMBH) not only power active galactic nuclei (AGNs), but also host single and binary embedded stellar-mass black holes (EBHs) that grow rapidly from gas accretion. The merger of these EBHs provides a promising mechanism for the excitation of some gravitational wave events observed by LIGO-Virgo, especially those with source masses considerably larger than isolated stellar-mass black hole binaries. In addition to their mass and mass-ratio distribution, their hitherto enigmatic small spin-parameters chi_effective carry important clues and stringent constraints on their formation channels and evolutionary pathways. Here we show that, between each coalescence, the typical rapid spin of the merged EBHs is suppressed by their subsequent accretion of gas from a turbulent environment, due to its ability to randomize the flow's spin orientation with respect to that of the EBHs on an eddy-turnover timescale. This theory provides supporting evidence for the prolificacy of EBH mergers and suggests that their mass growth may be dominated by gas accretion rather than their coalescence in AGN discs., Comment: accepted by MNRAS, 11 pages, 8 figures

We perform 3D radiation hydrodynamic local shearing box simulations to study the outcome of gravitational instability (GI) in optically thick Active Galactic Nuclei (AGN) accretion disks. GI develops when the Toomre parameter QT \leq 1, and may lead to turbulent heating that balances radiative cooling. However, when radiative cooling is too efficient, the disk may undergo runaway gravitational fragmentation. In the fully gas-pressure-dominated case, we confirm the classical result that such a thermal balance holds when the Shakura-Sunyaev viscosity parameter (alpha) due to the gravitationally-driven turbulence is \sim 0.2, corresponding to dimensionless cooling times Omega tcool \sim 5. As the fraction of support by radiation pressure increases, the disk becomes more prone to fragmentation, with a reduced (increased) critical value of alpha (omega tcool). The effect is already significant when the radiation pressure exceeds 10% of the gas pressure, while fully radiation-pressure-dominated disks fragment at Omega tcool <50 . The latter translates to a maximum turbulence level alpha<0.02, comparable to that generated by Magnetorotational Instability (MRI). Our results suggest that gravitationally unstable (QT \sim 1) outer regions of AGN disks with significant radiation pressure (likely for high/near- Eddington accretion rates) should always fragment into stars, and perhaps black holes., Comment: 26 pages, 19 figures, ApJ in Press

In autonomous robot exploration tasks, a mobile robot needs to actively explore and map an unknown environment as fast as possible. Since the environment is being revealed during exploration, the robot needs to frequently re-plan its path online, as new information is acquired by onboard sensors and used to update its partial map. While state-of-the-art exploration planners are frontier- and sampling-based, encouraged by the recent development in deep reinforcement learning (DRL), we propose ARiADNE, an attention-based neural approach to obtain real-time, non-myopic path planning for autonomous exploration. ARiADNE is able to learn dependencies at multiple spatial scales between areas of the agent's partial map, and implicitly predict potential gains associated with exploring those areas. This allows the agent to sequence movement actions that balance the natural trade-off between exploitation/refinement of the map in known areas and exploration of new areas. We experimentally demonstrate that our method outperforms both learning and non-learning state-of-the-art baselines in terms of average trajectory length to complete exploration in hundreds of simplified 2D indoor scenarios. We further validate our approach in high-fidelity Robot Operating System (ROS) simulations, where we consider a real sensor model and a realistic low-level motion controller, toward deployment on real robots., Comment: \c{opyright} 20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

We apply 3D hydrodynamical simulations to study the rotational aspect of gas flow patterns around eccentric companions embedded in an accretion disk around its primary host. We sample a wide range of companion mass ratio q and disk aspect ratio h, and confirm a generic transition from prograde (steady tidal interaction dominated) to retrograde (background Keplerian shear dominated) circum-companion flow when orbital eccentricity exceeds a critical value et. We find et \sim h for sub-thermal companions while et \sim (q/h)^1/3 for super-thermal companions, and propose an empirical formula to unify the two scenarios. Our results also suggest that et is insensitive to modest levels of turbulence, modeled in the form of a kinematic viscosity term. In the context of stellar-mass Black Holes (sBHs) embedded in AGN accretion disks, the bifurcation of their circum-stellar disk (CSD) rotation suggest the formation of a population of nearly anti-aligned sBHs, whose relevance to low spin gravitational waves (GW) events can be probed in more details with future population models of sBH evolution in AGN disks, making use of our quantitative scaling for et; In the context of circum-planetary disks (CPDs), our results suggest the possibility of forming retrograde satellites in-situ in retrograde CPDs around eccentric planets., Comment: Accepted to ApJL, 10 pages, 5 figures

We study the long-term radial transport of micron to mm-size grain in protostellar disks (PSDs) based on diffusion and viscosity coefficients measured from 3D global stratified-disk simulations with a Lagrangian hydrodynamic method. While gas-drag tend to transport dust species radially inwards, stochastic diffusion can spread a considerable fraction of dust radially outwards (upstream) depending on the nature of turbulence. In gravitationally unstable disks, we measure a high radial diffusion coefficient Dr with little dependence on altitude. This leads to strong and vertically homogeneous upstream diffusion in early PSDs. In the solar nebula, the robust upstream diffusion of micron to mm size grains not only efficiently transports highly refractory mocron-size grains (such as those identified in the samples of comet 81P/Wild 2) from their regions of formation inside the snow line out to the Kuiper Belt, but can also spread mm-size CAI formed in the stellar proximity to distances where they can be assimilated into chondritic meteorites. In disks dominated by magnetorotational instability (MRI), the upstream diffusion effect is generally milder, with a separating feature due to diffusion being stronger in the surface layer than the midplane. This variation becomes much more pronounced if we additionally consider a quiescent midplane with lower turbulence and larger characteristic dust size due to non-ideal MHD effects. This segregation scenario helps to account for dichotomy of two dust populations' spatial distribution as observed in scattered light and ALMA images., Comment: ApJ published, 14 pages, 12 figures

Studying the physical properties of materials under high pressure and temperature through experiments is difficult. Theoretical simulations can compensate for this deficiency. Currently, large-scale simulations using machine learning force fields are gaining popularity. As an important nuclear energy material, the evolution of the physical properties of uranium under extreme conditions is still unclear. Herein, we trained an accurate machine learning force field on alpha-U and predicted the lattice dynamics and elastic properties at high pressures and temperatures. The force field agrees well with the ab initio molecular dynamics (AIMD) and experimental results, and it exhibits higher accuracy than classical potentials. Based on the high-temperature lattice dynamics study, we first present the temperature-pressure range in which the Kohn anomalous behavior of the ${\Sigma}$4 optical mode exists. Phonon spectral function analysis showed that the phonon anharmonicity of alpha-U is very weak. We predict that the single-crystal elastic constants C44, C55, C66, polycrystalline modulus (E,G), and polycrystalline sound velocity ($C_L$,$C_S$) have strong heating-induced softening. All the elastic moduli exhibited compression-induced hardening behavior. The Poisson's ratio shows that it is difficult to compress alpha-U at high pressures and temperatures. Moreover, we observed that the material becomes substantially more anisotropic at high pressures and temperatures. The accurate predictions of alpha-U demonstrate the reliability of the method. This versatile method facilitates the study of other complex metallic materials., Comment: 21 pages, 9 figures, with Supplementary Material

Nowadays a wide range of applications is constrained by low-latency requirements that cloud infrastructures cannot meet. Multi-access Edge Computing (MEC) has been proposed as the reference architecture for executing applications closer to users and reduce latency, but new challenges arise: edge nodes are resource-constrained, the workload can vary significantly since users are nomadic, and task complexity is increasing (e.g., machine learning inference). To overcome these problems, the paper presents NEPTUNE, a serverless-based framework for managing complex MEC solutions. NEPTUNE i) places functions on edge nodes according to user locations, ii) avoids the saturation of single nodes, iii) exploits GPUs when available, and iv) allocates resources (CPU cores) dynamically to meet foreseen execution times. A prototype, built on top of K3S, was used to evaluate NEPTUNE on a set of experiments that demonstrate a significant reduction in terms of response time, network overhead, and resource consumption compared to three state-of-the-art approaches., Comment: 11 pages + 2 pages of references

Spin evolution of stellar-mass Black Holes (sBHs) embedded in AGN accretion disks is an important process relevant to production of gravitaional waves from binary Black Hole (BBH) merger events through the AGN channel. Since embedded sBHs are surrounded by circum-stellar disks (CSDs), the rotation of CSD gas flows determine the direction of the angular momentum it accretes. In this Letter, we use global 2D hydrodynamic simulations to show that while a disk-embedded sBH on a circular orbit transforms the initial retrograde Keplerian shear of the background accretion disk into a prograde CSD flow, as in the classical picture of companion-disk interaction theory, moderate orbital eccentricity could disrupt the steady-state tidal perturbation and preserve a retrograde CSD flow around the sBH. This switch of CSD orientation occurs at a transition eccentricity that scales nearly proportional with local sound speed. This bifurcation in the CSD flow and thereafter spin-up direction of SBHs leads to formation of a population of nearly anti-aligned sBHs and should be incorporated in future population models of sBH and BBH evolutions., Comment: ApJL published

The prevalence of short-period super-Earths that are independent of host metallicity challenges the theoretical construction of their origin. We propose that dust trapping in the global pressure bump induced by magnetospheric truncation in evolved protoplanetary discs (PPDs) around T Tauri stars offers a promising formation mechanism for super-Earths, where the host metallicity is already established. To better understand this planet forming scenario, we construct a toy inner disc model and focus on the evolution of dust trapped in the bump, taking into account the supply from drifting pebbles and loss due to funnel flows. We develop an implicit coagulation-fragmentation code, $\mathtt{Rubble}$, and perform a suite of simulations to evolve the local dust size distributions. Our study for the first time considers dust feedback effect on turbulent diffusion in this kind of model. We report that efficient dust growth and significant accumulation of dust mass is possible in less turbulent disc with sturdier solids and with faster external supply, laying out a solid foundation for further growth towards planetesimals and planetary embryos. We further find that, depending on the dominant process, solid mass may predominantly accumulate in cm-sized grains or particles in runaway growth, indicating different ways of forming planetesimals. Furthermore, these various outcomes show different efficiencies in saving dust from funnel flows, suggesting that they may be distinguishable by constraining the opacity of funnel flows. Also, these diverse dust behaviours may help explain the observed dipper stars and rapidly varying shadows in PPDs., Comment: Accepted by MNRAS and in production; 20 pages, 14 figures

Objectives. The aim of this study was to investigate whether a deep convolutional neural network (CNN) with an attention module can detect osteoporosis on panoramic radiographs. Study Design. A dataset of 70 panoramic radiographs (PRs) from 70 different subjects of age between 49 to 60 was used, including 49 subjects with osteoporosis and 21 normal subjects. We utilized the leave-one-out cross-validation approach to generate 70 training and test splits. Specifically, for each split, one image was used for testing and the remaining 69 images were used for training. A deep convolutional neural network (CNN) using the Siamese architecture was implemented through a fine-tuning process to classify an PR image using patches extracted from eight representative trabecula bone areas (Figure 1). In order to automatically learn the importance of different PR patches, an attention module was integrated into the deep CNN. Three metrics, including osteoporosis accuracy (OPA), non-osteoporosis accuracy (NOPA) and overall accuracy (OA), were utilized for performance evaluation. Results. The proposed baseline CNN approach achieved the OPA, NOPA and OA scores of 0.667, 0.878 and 0.814, respectively. With the help of the attention module, the OPA, NOPA and OA scores were further improved to 0.714, 0.939 and 0.871, respectively. Conclusions. The proposed method obtained promising results using deep CNN with an attention module, which might be applied to osteoporosis prescreening., Comment: 9 pages

The tidal perturbation of embedded protoplanets on their natal disks has been widely attributed to be the cause of gap-ring structures in sub-mm images of protoplanetary disks around T Tauri stars. Numerical simulations of this process have been used to propose scalings of characteristic dust gap width/gap-ring distance with respect to planet mass. Applying such scalings to analyze observed gap samples yields a continuous mass distribution for a rich population of hypothetical planets in the range of several Earth to Jupiter masses. In contrast, the conventional core-accretion scenario of planet formation predicts a bi-modal mass function due to 1) the onset of runaway gas accretion above \sim20 Earth masses and 2) suppression of accretion induced by gap opening. Here we examine the dust disk response to the tidal perturbation of eccentric planets as a possible resolution of this paradox. Based on simulated gas and dust distributions, we show the gap-ring separation of Neptune-mass planets with small eccentricities might become comparable to that induced by Saturn-mass planets on circular orbits. This degeneracy may obliterate the discrepancy between the theoretical bi-modal mass distribution and the observed continuous gap width distribution. Despite damping due to planet-disk interaction, modest eccentricity may be sustained either in the outer regions of relatively thick disks or through resonant excitation among multiple super Earths. Moreover, the ring-like dust distribution induced by planets with small eccentricities is axisymmetric even in low viscosity environments, consistent with the paucity of vortices in ALMA images., Comment: 13 pages, 7 figures, ApJ in press