EVENT DOSSIER
The researchers proposed a physical-guided diffusion model approach to reconstruct the surface mass density of galaxy clusters using the DarkClusters-15k dataset (containing 15,000 simulated clusters). This method trains a pluggable diffusion prior to learn the statistical relationship between mass and light, and combines weak and strong lensing observations to generate posterior samples. This enables automated reconstruction with high accuracy without expert tuning, requiring only a few minutes of operation. This achievement reached an expert level in the reconstruction of the MACS 1206 cluster, aiming to provide development support and benchmark tests for future wide-field cosmology surveys.
研究人员提出一种基于物理引导的扩散模型方法，利用暗物质团簇数据集 DarkClusters-15k（含 15,000 个模拟团簇）重建星系团表面质量密度。该方法通过训练可插拔扩散先验学习质量与光的统计关系，并结合弱透镜和强透镜观测约束生成后验样本，实现了无需专家调参、运行仅需数分钟且精度更高的自动化重建。该成果在 MACS 1206 团簇的重建中达到专家级水平，旨在为未来广域宇宙学巡天提供开发支持与基准测试。