623509 research outputs found
Sort by
Reinforcement Learning-Based Model Matching to Reduce the Sim-Real Gap in COBRA
No full textThis paper employs a reinforcement learning-based model identification method aimed at enhancing the accuracy of the dynamics for our snake robot, called COBRA. Leveraging gradient information and iterative optimization, the proposed approach refines the parameters of COBRA\u27s dynamical model such as coefficient of friction and actuator parameters using experimental and simulated data. Experimental validation on the hardware platform demonstrates the efficacy of the proposed approach, highlighting its potential to address sim-to-real gap in robot implementation
Kuramoto oscillators in random networks
No full textBy means of numerical analysis conducted with the aid of the computer, the collective synchronization of coupled phase oscillators in the Kuramoto model in the connected regime of random networks of various sizes is studied. The oscillators synchronize and achieve phase coherence, and this process is not significantly affected by the level of connectivity of the network. If the probability that two oscillators are coupled is around the network connectivity threshold synchronization still occurs, although in a more attenuated way. If the size of the network is sufficiently large the oscillators have a phase transition.9 pages, 4 figure
Normalizer Quotients of Symmetric Groups and Inner Holomorphs
No full textWe show that every finite group is isomorphic to a normalizer quotient for some and a subgroup . We show that this holds for all large enough and also with replaced by . The two main ingredients in the proof are a recent construction due to Cornulier and Sambale of a finite group with (for any given finite group ) and the determination of the normalizer in of the inner holomorph for any centerless indecomposable finite group , which may be of independent interest
Taming Generative Diffusion Prior for Universal Blind Image Restoration
No full textDiffusion models have been widely utilized for image restoration. However, previous blind image restoration methods still need to assume the type of degradation model while leaving the parameters to be optimized, limiting their real-world applications. Therefore, we aim to tame generative diffusion prior for universal blind image restoration dubbed BIR-D, which utilizes an optimizable convolutional kernel to simulate the degradation model and dynamically update the parameters of the kernel in the diffusion steps, enabling it to achieve blind image restoration results even in various complex situations. Besides, based on mathematical reasoning, we have provided an empirical formula for the chosen of adaptive guidance scale, eliminating the need for a grid search for the optimal parameter. Experimentally, Our BIR-D has demonstrated superior practicality and versatility than off-the-shelf unsupervised methods across various tasks both on real-world and synthetic datasets, qualitatively and quantitatively. BIR-D is able to fulfill multi-guidance blind image restoration. Moreover, BIR-D can also restore images that undergo multiple and complicated degradations, demonstrating the practical applications.15 pages, 12 figures, 8 table
Gravitational lensing effect of black holes in effective quantum gravity
No full textIn the present work, we investigate the gravitational lensing effects of two quantum-modified black hole models recently proposed in effective quantum gravity. The light deflection angles are calculated for both the weak-field and strong-field limits. Furthermore, using the data for the supermassive black holes SgrA* and M87*, we calculate the lensing observables in the strong-field limit. We find that the quantum parameter plays a role analogous to the electric charge in weak gravitational lensing. In the strong-field limit, in contrast, the effects of the quantum parameter on the deflection angle, the angular separation, and the relative magnification are opposite to those of the electric charge, the scalar charge, and the quantum parameters in some gravity theories. The results indicate the crucial difference between the classical black holes and the two quantum-modified black hole models that depend on the quantum correction, making them a valuable tool for distinguishing these black hole models
The companion mass distribution of post common envelope hot subdwarf binaries: evidence for boosted and disrupted magnetic braking?
No full textWe measure the mass distribution of main-sequence (MS) companions to hot subdwarf B stars (sdBs) in post-common envelope binaries (PCEBs). We carried out a spectroscopic survey of 14 eclipsing systems ( HW Vir binaries ) with orbital periods of , with only two systems hosting companions above the fully-convective limit. There is no correlation between and within the sample. A similar drop-off in the companion mass distribution of white dwarf (WD) + MS PCEBs has been attributed to disrupted magnetic braking (MB) below the fully-convective limit. We compare the sdB companion mass distribution to predictions of binary evolution simulations with a range of MB laws. Because sdBs have short lifetimes compared to WDs, explaining the lack of higher-mass MS companions to sdBs with disrupted MB requires MB to be boosted by a factor of 20-100 relative to MB laws inferred from the rotation evolution of single stars. We speculate that such boosting may be a result of irradiation-driven enhancement of the MS stars\u27 winds. An alternative possibility is that common envelope evolution favors low-mass companions in short-period orbits, but the existence of massive WD companions to sdBs with similar periods disfavors this scenario.20 pages, 12 figures, accepted for publication in PAS
A Systematic Survey of Moon-Forming Giant Impacts. II. Rotating bodies
No full textIn the leading theory of lunar formation, known as the giant impact hypothesis, a collision between two planet-size objects resulted in a young Earth surrounded by a circumplanetary debris disk from which the Moon later accreted. The range of giant impacts that could conceivably explain the Earth-Moon system is limited by the set of known physical and geochemical constraints. However, while several distinct Moon-forming impact scenarios have been proposed -- from small, high-velocity impactors to low-velocity mergers between equal-mass objects -- none of these scenarios have been successful at explaining the full set of known constraints, especially without invoking one or more controversial post-impact processes. Allowing for pre-impact rotation of the colliding bodies has been suggested as an avenue which may produce more promising collision outcomes. However, to date, only limited studies of pre-impact rotation have been conducted. Therefore, in the second paper of this series, we focus on pairwise impacts between rotating bodies. Using non-rotating collisions as a baseline, we systematically study the effects of rotation on collision outcomes. We consider nine distinct rotation configurations and a range of rotation rates up to the rotational stability limit. Notably, we identify a population of collisions that can produce low post-impact angular momentum budgets and massive, iron-poor protolunar disks. Furthermore, even when pre-impact rotation is included, we demonstrate that the canonical Moon-forming impact can only generate sufficiently massive protolunar disks in the presence of excessive post-impact angular momentum budgets; this casts doubt on the canonical impact scenario.30 pages, 18 figures, 2 tables, accepteed for publication in Ap
TFG: Unified Training-Free Guidance for Diffusion Models
No full textGiven an unconditional diffusion model and a predictor for a target property of interest (e.g., a classifier), the goal of training-free guidance is to generate samples with desirable target properties without additional training. Existing methods, though effective in various individual applications, often lack theoretical grounding and rigorous testing on extensive benchmarks. As a result, they could even fail on simple tasks, and applying them to a new problem becomes unavoidably difficult. This paper introduces a novel algorithmic framework encompassing existing methods as special cases, unifying the study of training-free guidance into the analysis of an algorithm-agnostic design space. Via theoretical and empirical investigation, we propose an efficient and effective hyper-parameter searching strategy that can be readily applied to any downstream task. We systematically benchmark across 7 diffusion models on 16 tasks with 40 targets, and improve performance by 8.5% on average. Our framework and benchmark offer a solid foundation for conditional generation in a training-free manner
Diversity-Driven Synthesis: Enhancing Dataset Distillation through Directed Weight Adjustment
No full textThe sharp increase in data-related expenses has motivated research into condensing datasets while retaining the most informative features. Dataset distillation has thus recently come to the fore. This paradigm generates synthetic datasets that are representative enough to replace the original dataset in training a neural network. To avoid redundancy in these synthetic datasets, it is crucial that each element contains unique features and remains diverse from others during the synthesis stage. In this paper, we provide a thorough theoretical and empirical analysis of diversity within synthesized datasets. We argue that enhancing diversity can improve the parallelizable yet isolated synthesizing approach. Specifically, we introduce a novel method that employs dynamic and directed weight adjustment techniques to modulate the synthesis process, thereby maximizing the representativeness and diversity of each synthetic instance. Our method ensures that each batch of synthetic data mirrors the characteristics of a large, varying subset of the original dataset. Extensive experiments across multiple datasets, including CIFAR, Tiny-ImageNet, and ImageNet-1K, demonstrate the superior performance of our method, highlighting its effectiveness in producing diverse and representative synthetic datasets with minimal computational expense. Our code is available at https://github.com/AngusDujw/Diversity-Driven-Synthesis.https://github.com/AngusDujw/Diversity-Driven-Synthesis
Photospheric Hot Spots at Solar Coronal Loop Footpoints Revealed by Hyperspectral Imaging Observations
No full textPoynting flux generated by random shuffling of photospheric magnetic footpoints is transferred through the upper atmosphere of the Sun where the plasma is heated to over 1 MK in the corona. High spatiotemporal resolution observations of the lower atmosphere at the base of coronal magnetic loops are crucial to better understand the nature of the footpoint dynamics and the details of magnetic processes that eventually channel energy into the corona. Here we report high spatial resolution (0.1\arcsec) and cadence (1.33 s) hyperspectral imaging of the solar H line, acquired by the Microlensed Hyperspectral Imager prototype installed at the Swedish 1-m Solar Telescope, that reveal photospheric hot spots at the base of solar coronal loops. These hot spots manifest themselves as H wing enhancements, occurring on small spatial scales of 0.2\arcsec, and timescales of less than 100 s. By assuming that the H wings and the continuum form under the local thermodynamic equilibrium condition, we inverted the H line profiles and found that the hot spots are compatible with a temperature increase of about 1000 K above the ambient quiet-Sun temperature. The H wing integrated Stokes maps indicate that hot spots are related to magnetic patches with field strengths comparable to or even stronger than the surrounding network elements. But they do not show the presence of parasitic polarity magnetic field that would support the interpretation that these hot spots are reconnection-driven Ellerman bombs. Therefore, we interpret these features as proxies of locations where convection-driven magnetic field intensification in the photosphere can lead to energy transfer into higher layers. We suggest that such hot spots at coronal loop footpoints may be indicative of the specific locations and onset of energy flux injection into the upper atmosphere.Published in the Astrophysical Journa