arXiv subjects

Leyi Wang

Publications and source records attributed to Leyi Wang.

4 recordsLinked to original sources

arXiv2026

Magnetically Arrested Discs Powering Jets in a Large Sample of Low-Accretion FR I Radio Galaxies

We study a sample of 289 Fanaroff-Riley type I (FR I) radio galaxies selected from the LOFAR Two-Metre Sky Survey (LoTSS) DR1, identified by their edge-darkened radio morphologies. Using Sloan Digital Sky Survey (SDSS) DR17 optical photometry and spectroscopy, we derive Eddington-scaled accretion rates spanning -6.84 < log $\dot{m}$ < -0.87 (median $\approx$ -2.84). The vast majority of sources lie below $\dot{m}$ = 0.01, indicating that their central engines are well described by advection-dominated accretion flows (ADAFs). However, even for a rapidly spinning black hole with a = 0.95, the maximum jet power predicted by the Blandford-Znajek mechanism in the standard ADAF regime is lower than the observed jet power (estimated from 151 MHz radio luminosity) for approximately 70% of the sample. We demonstrate that the magnetically arrested disc (MAD) scenario, in which large-scale poloidal magnetic flux accumulates near the event horizon, can fully account for the powerful jets observed in these low-accretion systems. Within the MAD framework, the data are consistent with slow-spinning black holes with $a < 0.5$. This large, uniformly selected LoTSS sample extends the MAD requirement previously established for the bright 3CR FR I population, indicating that magnetically arrested discs are common in FR I radio galaxies across a wide range of luminosities.

astro-ph.GA↗

arXiv2025

Generative assimilation and prediction for weather and climate

Machine learning models have shown great success in predicting weather up to two weeks ahead, outperforming process-based benchmarks. However, existing approaches mostly focus on the prediction task, and do not incorporate the necessary data assimilation. Moreover, these models suffer from error accumulation in long roll-outs, limiting their applicability to seasonal predictions or climate projections. Here, we introduce Generative Assimilation and Prediction (GAP), a unified deep generative framework for assimilation and prediction of both weather and climate. By learning to quantify the probabilistic distribution of atmospheric states under observational, predictive, and external forcing constraints, GAP excels in a broad range of weather-climate related tasks, including data assimilation, seamless prediction, and climate simulation. In particular, GAP is competitive with state-of-the-art ensemble assimilation, probabilistic weather forecast and seasonal prediction, yields stable millennial simulations, and reproduces climate variability from daily to decadal time scales.

cs.LG↗

arXiv2020

Deep Filtering

This paper develops a deep learning method for linear and nonlinear filtering. The idea is to start with a nominal dynamic model and generate Monte Carlo sample paths. Then these samples are used to train a deep neutral network. A least square error is used as a loss function for network training. Then the resulting weights are applied to Monte Carlo sampl\ es from an actual dynamic model. The deep filter obtained in such a way compares favorably to the traditional Kalman filter in linear cases and the extended Kalman filter in nonlinear cases. Moreover, a switching model with jumps is studied to show the adaptiveness and power of our deep filtering method. A main advantage of deep filtering is its robustness when the nominal model and actual model differ. Another advantage of deep filtering is that real data can be used directly to train the deep neutral network. Therefore, one does not need to calibrate the model.

math.OC↗

arXiv2013

Exponential Mixing for Retarded Stochastic Differential Equations

In this paper, we discuss exponential mixing property for Markovian semigroups generated by segment processes associated with several class of retarded Stochastic Differential Equations (SDEs) which cover SDEs with constant/variable/distributed time-lags. In particular, we investigate the exponential mixing property for (a) non-autonomous retarded SDEs by the Arzelà--Ascoli tightness characterization of the space $\C$ equipped with the uniform topology (b) neutral SDEs with continuous sample paths by a generalized Razumikhin-type argument and a stability-in-distribution approach and (c) jump-diffusion retarded SDEs by the Kurtz criterion of tightness for the space $\D$ endowed with the Skorohod topology.

math.PR↗