The Simplified Line-Input Crystal-Encoding System (SLICES) is the first invertible and invariant crystal representation tool. This software supports encoding and decoding crystal structures, reconstructing them, and generating new materials with desired properties using generative deep learning.
Related Publications and Resources:
- Nature Communications: Paper
- MatterGPT 图形界面介绍: Bilibili
- SLICES 晶体语言介绍: Bilibili
- SLICES 101: Bilibili
- Data and Results: Figshare
- MatterGPT Paper: arXiv
- MatterGPT Demo: Huggingface
- SLICES-PLUS Paper: arXiv
- Encode crystal structures into SLICES strings
- Reconstruct original crystal structures (Text2Crystal)
- Inverse design of solid-state materials with desired properties using MatterGPT
- Inverse design of solid-state materials with desired properties and crystal systems using MatterGPT (SLICES-PLUS)
We provide a huggingface space to allow one-click conversion of CIF to SLICES and SLICES to CIF online.
- Installation
- Examples
- Tutorials
- Documentation
- Reproducing Benchmarks
- Citation
- Acknowledgements
- Contact and Support
可以选择 1.本地安装 或者 2.docker安装。
本地安装指南(Ubuntu或者win11的Ubuntu子系统)
如果你已经安装了 Miniconda 并配置好了中国镜像源，可以跳过这一步。
- 首先，更新 Ubuntu 系统的包索引并安装
wget工具，用于下载文件。 - 然后，你通过
wget下载 Miniconda 安装脚本并执行安装，将 Miniconda 安装到~/miniconda3目录。 - 在安装完成后，删除安装脚本，并配置了
pip使用清华大学的镜像源，以提高下载速度。
sudo apt-get update
sudo apt-get install wget
mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm ~/miniconda3/miniconda.sh
source ~/miniconda3/bin/activate
conda init --all
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --set show_channel_urls yes
pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple
python -m pip install --upgrade pip- 下载 SLICES 项目的 GitHub 文件夹，并将其解压。
- 进入解压后的目录，使用
conda创建一个新环境，并安装所需的依赖（环境配置文件是environments.yml）。 - 然后，激活
slices环境，并使用pip install slices安装 SLICES 包。
wget https://github.com/xiaohang007/SLICES/archive/refs/heads/main.zip -O slices_repo.zip
unzip slices_repo.zip
cd SLICES-main
conda env create --name slices --file=environments.yml
conda activate slices
pip install gradio==4.44.1 slices==2.0.8安装完成！
cd MatterGPT
python app.py按住 CTRL (或 Mac 上的 Command)，然后点击terminal中的 http://localhost:7860链接来打开MatterGPT的图形界面。
如果想用Docker来跑SLICES的环境，可以选择下面的安装方法：
请按照以下步骤使用 Docker 安装 SLICES:
-
下载仓库，并解压文件。
-
在
slurm.conf中配置 CPU 线程数量。 -
执行 Docker 集成命令：
# 从 Docker Hub 下载已经构建的 SLICES Docker 镜像
docker pull xiaohang07/slices:v10
# 如果 docker pull 不管用，您可以在 https://figshare.com/s/260701a1accd0192de20 下载压缩的 docker 镜像 v10。
# 然后，使用以下命令4线程并行解压缩以及加载此 docker 镜像：
xz -T4 -dc dockerv10.tar.xz | docker load
# 您也可以使用此仓库中的 Dockerfile 构建自己的 docker 镜像。感谢 Haidi Wang 教授 (https://haidi-ustc.github.io/about/) 提供的 Dockerfile。
# 使脚本具有执行权限
sudo chmod +x entrypoint_set_cpus_gradio.sh entrypoint_set_cpus.sh ./slices/xtb_noring_nooutput_nostdout_noCN
# 运行 Docker (将 [] 替换为您的绝对路径)
docker run -it -p 7860:7860 -h workq --shm-size=0.5gb --gpus all -v /[]:/crystal xiaohang07/slices:v10 /crystal/entrypoint_set_cpus_gradio.sh-
访问图形界面:
- 按住
CTRL(或 Mac 上的Command)，然后点击terminal中的http://localhost:7860链接来打开MatterGPT的图形界面。
- 按住
-
Windows 11：使用完全支持 GPU 的 Docker 环境。
-
Ubuntu：使用本地安装更好，因为 nvidia-docker 比较难安装。
Convert a crystal structure to its SLICES string and reconstruct it.
from slices.core import SLICES
from pymatgen.core.structure import Structure
# Load crystal structure from file
original_structure = Structure.from_file(filename='NdSiRu.cif')
backend = SLICES()
# Convert to SLICES string
slices_NdSiRu = backend.structure2SLICES(original_structure)
# Reconstruct crystal and get predicted energy
reconstructed_structure, final_energy_per_atom = backend.SLICES2structure(slices_NdSiRu)
print('SLICES string of NdSiRu is:', slices_NdSiRu)
print('Reconstructed structure:', reconstructed_structure)
print('Final energy per atom:', final_energy_per_atom, 'eV/atom')Generate augmented SLICES strings and reduce them to a canonical form.
from slices.core import SLICES
from pymatgen.core.structure import Structure
# Load crystal structure
original_structure = Structure.from_file(filename='Sr3Ru2O7.cif')
backend = SLICES(graph_method='econnn')
# Generate augmented SLICES
slices_list = backend.structure2SLICESAug_atom_order(structure=original_structure, num=50)
canonical_slices = list(set(backend.get_canonical_SLICES(s) for s in slices_list))
print('Unique Canonical SLICES:', len(canonical_slices))The official documentation is available at Read the Docs.
Refer to the Benchmarks Guide for detailed instructions.
If you use SLICES, MatterGPT or SLICES-PLUS, please cite the following works:
@article{xiao2023invertible,
title={An invertible, invariant crystal representation for inverse design of solid-state materials using generative deep learning},
author={Xiao, Hang and Li, Rong and Shi, Xiaoyang and Chen, Yan and Zhu, Liangliang and Chen, Xi and Wang, Lei},
journal={Nature Communications},
volume={14},
number={1},
pages={7027},
year={2023},
publisher={Nature Publishing Group UK London}
}
@misc{chen2024mattergptgenerativetransformermultiproperty,
title={MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials},
author={Yan Chen and Xueru Wang and Xiaobin Deng and Yilun Liu and Xi Chen and Yunwei Zhang and Lei Wang and Hang Xiao},
year={2024},
eprint={2408.07608},
archivePrefix={arXiv},
primaryClass={cond-mat.mtrl-sci},
url={https://arxiv.org/abs/2408.07608}
}
@misc{wang2024slicespluscrystalrepresentationleveraging,
title={SLICES-PLUS: A Crystal Representation Leveraging Spatial Symmetry},
author={Baoning Wang and Zhiyuan Xu and Zhiyu Han and Qiwen Nie and Hang Xiao and Gang Yan},
year={2024},
eprint={2410.22828},
archivePrefix={arXiv},
primaryClass={physics.comp-ph},
url={https://arxiv.org/abs/2410.22828},
}Special thanks to the open-source projects and developers that inspired this work:
- Email: hangxiao@ln.edu.hk
- ResearchGate: Hang Xiao
- Start a Discussion: GitHub Discussions