CAMELS Multifield Dataset (CMD)

CAMELS Multifield Dataset (CMD)是由普林斯顿大学天体物理系的研究团队开发的大型宇宙学和天体物理数据集，包含从2000个不同模拟宇宙中提取的数十万个2D地图和3D网格。数据集涵盖了宇宙气体、暗物质和恒星的多种属性，数据量超过70TB，是同类数据集中最大的公开数据集。CMD数据集通过先进的流体动力学和仅重力N体模拟生成，旨在训练机器学习模型，以解决宇宙学参数推断等复杂任务。该数据集的应用领域包括参数推断、生成模型、N体到流体动力学的映射、超分辨率处理和时间演化分析，旨在推动宇宙学和天体物理学的研究进展。

CAMELS Multifield Dataset (CMD) was developed by a research team from the Department of Astrophysical Sciences at Princeton University. It is a large-scale cosmological and astrophysical dataset containing hundreds of thousands of 2D maps and 3D grids extracted from 2000 distinct simulated universes. The dataset covers multiple properties of cosmic gas, dark matter and stars, with a total data volume exceeding 70 TB, making it the largest publicly available dataset in its category. Generated using state-of-the-art hydrodynamic and gravity-only N-body simulations, CMD is intended to train machine learning models to solve complex tasks such as cosmological parameter inference. Its application domains include parameter inference, generative models, N-body to hydrodynamic mapping, super-resolution processing and temporal evolution analysis, with the goal of advancing research in cosmology and astrophysics.