graphs-datasets/MD17-uracil
收藏资源简介:
`uracil`数据集是一个分子动力学(MD)数据集,包含使用PBE+vdW-TS电子结构方法计算的总能量和力标签。所有几何结构以埃为单位,能量和力分别以kcal/mol和kcal/mol/A为单位。数据集包含133769个图,平均每个图有12个节点和128.89条边。每个图的数据字段包括节点特征、边索引、边属性和标签等。数据集未进行分割,建议使用交叉验证。
The `uracil` dataset is a molecular dynamics (MD) dataset containing total energy and force labels calculated via the PBE+vdW-TS electronic structure method. All geometric structures are reported in angstroms, with energy and force units being kcal/mol and kcal/mol/Å respectively. The dataset consists of 133,769 graphs, with an average of 12 nodes and 128.89 edges per graph. Data fields for each graph include node features, edge indices, edge attributes, labels and other related contents. This dataset has not been pre-split, and cross-validation is recommended for subsequent analysis.
数据集概述
数据集描述
- 数据集名称: uracil
- 数据集类型: 分子动力学(MD)数据集
- 计算方法: 使用PBE+vdW-TS电子结构方法计算总能量和力标签
- 单位: 几何结构以Angstrom为单位,能量和力分别以kcal/mol和kcal/mol/A为单位
数据集总结
- 任务类型: 有机分子属性预测,回归任务
- 评估指标: 能量预测的平均绝对误差(meV)
数据集结构
数据属性
- 规模: 大
- 图数量: 133769
- 平均节点数: 12.0
- 平均边数: 128.88676085818943
数据字段
node_feat(列表: #nodes x #node-features): 节点特征edge_index(列表: 2 x #edges): 构成边的节点对edge_attr(列表: #edges x #edge-features): 边特征y(列表: #labels): 可用于预测的标签数量num_nodes(整数): 图的节点数
数据分割
- 分割方式: 未分割,建议使用交叉验证
附加信息
许可信息
- 许可类型: 未知
引用信息
@inproceedings{Morris+2020, title={TUDataset: A collection of benchmark datasets for learning with graphs}, author={Christopher Morris and Nils M. Kriege and Franka Bause and Kristian Kersting and Petra Mutzel and Marion Neumann}, booktitle={ICML 2020 Workshop on Graph Representation Learning and Beyond (GRL+ 2020)}, archivePrefix={arXiv}, eprint={2007.08663}, url={www.graphlearning.io}, year={2020} }
@article{Chmiela_2017, doi = {10.1126/sciadv.1603015}, url = {https://doi.org/10.1126%2Fsciadv.1603015}, year = 2017, month = {may}, publisher = {American Association for the Advancement of Science ({AAAS})}, volume = {3}, number = {5}, author = {Stefan Chmiela and Alexandre Tkatchenko and Huziel E. Sauceda and Igor Poltavsky and Kristof T. Schütt and Klaus-Robert Müller}, title = {Machine learning of accurate energy-conserving molecular force fields}, journal = {Science Advances} }




