科研成果

https://arxiv.org/abs/2312.01848

Existing diffusion models for low-light image enhancement typically incrementally remove noise introduced during the forward diffusion process using a denoising loss, with the process being conditioned on input low-light images. While these models demonstrate remarkable abilities in generating realistic high-frequency details, they often struggle to restore fine details that are faithful to the input. To address this, we present a novel detail-preserving diffusion model for realistic and faithful low-light image enhancement. Our approach integrates a size-agnostic diffusion process with a reverse process reconstruction loss, significantly enhancing the fidelity of enhanced images to their low-light counterparts and enabling more accurate recovery of fine details. To ensure the preservation of region- and content-aware details, we employ an efficient noise estimation network with a simplified channel-spatial attention mechanism. Additionally, we propose a multiscale ensemble scheme to maintain detail fidelity across diverse illumination regions. Comprehensive experiments on eight benchmark datasets demonstrate that our method achieves state-of-the-art results compared to over twenty existing methods in terms of both perceptual quality (LPIPS) and distortion metrics (PSNR and SSIM). The code is available at: https://github.com/CSYanH/DePDiff.

Surface ozone (O3) pollution affects air quality, human health, and the ecosystem. Understanding the complex non-linear relationship between ozone formation and its precursors, nitrogen oxides (NOx), and volatile organic compounds (VOCs) is critical for policymakers to mitigate the pollution. The Empirical Kinetic Modeling Approach (EKMA) based on classical observation-constrained zero-dimension box model provides the sensitivity of ozone production to precursor concentrations instead of emissions. This makes the box-model EKMA hard to apply in a real emission reduction scenario. Here, we developed an alternative box model approach driven by localized emissions, which are derived from the field-observed concentrations. This model approach reproduced the O3 variations well by capturing the short-term changes of NOx and AVOCs emissions among different phases of pollution control during the 31st World University Games in Chengdu in 2023. The EKMA analysis based on this model approach showed a different O3 response to precursor reductions from the concentration-constrained approach, which overestimated the baseline of O3 concentration. The result from the EKMA analysis demonstrated that the O3 level was most sensitive to NOx due to stringent control strategies during the event and rapidly rebounded to almost VOC-limited regime after the event. The effects of VOCs reduction on O3 control examined by concentration-constrained model approach were less pronounced than those by emission-driven approach due to the lack of consideration of the emission-to-reaction process. Our findings suggest that the emission-driven box model is applicable for developing O3 control strategy in the local scale.

In the context of supply chain digitization and green development in full swing, it is crucial to clarify the impact of the former on green energy innovation. Using exogenous shocks deriving from supply chain innovation and application pilot events, this study examines the impact of supply chain digitization on green energy innovation based on the data of Chinese listed companies from 2012 to 2021. The findings show that supply chain digitization significantly enhances corporate green energy innovation and that receivable asset management is a path mechanism for supply chain digitization to drive green energy innovation. Moreover, there is a significant positive intra-city spillover. Supply chain digitization contributes significantly to corporate green energy innovation in state-controlled manufacturing firms with effective internal controls in the eastern region. This study has important policy implications for promoting green energy innovation and accelerating the development of modern supply chain systems.

Class 1 integrons facilitate horizontal gene transfer, significantly influencing antibiotic resistance gene (ARG) dissemination within microbial communities. Wastewater treatment plants (WWTPs) are critical reservoirs of ARGs and integrons, yet the integron-mediated dynamics of ARG transfer across different WWTP types remain poorly understood. Here we show distinct ARG profiles associated with class 1 integrons in municipal and industrial WWTPs using a novel approach combining nested-like high-throughput qPCR and PacBio sequencing. Although industrial WWTPs contained higher absolute integron abundances, their relative ARG content was lower (1.27 × 107–9.59 × 107 copies/ng integron) compared to municipal WWTPs (3.72 × 107–1.98 × 108 copies/ng integron). Of the 132,084 coding sequences detected from integrons, 56.8 % encoded antibiotic resistance, with industrial plants showing lower ARG proportions, reduced ARG array diversity, and greater incorporation of non-ARG sequences. These findings suggest industrial WWTP integrons integrate a broader array of exogenous genes, reflecting adaptation to complex wastewater compositions. This work enhances our understanding of integron-driven ARG dynamics in wastewater and offers a robust strategy for environmental integron analysis.

Numerical resolution of moderately high-dimensional nonlinear PDEs remains a huge challenge due to the curse of dimensionality for the classical numerical methods including finite difference, finite element and spectral methods. Starting from the weak formulation of the Lawson-Euler scheme, this paper proposes a stochastic particle method (SPM) by tracking the deterministic motion, random jump, resampling and reweighting of particles. Real-valued weighted particles are adopted by SPM to approximate the high-dimensional solution, which automatically adjusts the point distribution to intimate the relevant feature of the solution. A piecewise constant reconstruction with virtual uniform grid is employed to evaluate the nonlinear terms, which fully exploits the intrinsic adaptive characteristic of SPM. Combining both, SPM can achieve the goal of adaptive sampling in time. Numerical experiments on the 6-D Allen-Cahn equation and the 7- D Hamiltonian-Jacobi-Bellman equation demonstrate the potential of SPM in solving moderately high-dimensional nonlinear PDEs efficiently while maintaining an acceptable accuracy