Policy-guided Monte Carlo on general state spaces: Application to glass-forming mixtures

This dataset is associated with "Policy-guided Monte Carlo on general state spaces: Application to glass-forming mixtures" by L. Galliano, R. Rende and D. Coslovich. It includes data and workflow to reproduce the figures of the manuscript. Requirements A working installation of Julia (tested with version 1.10.4) is the only software required for this workflow. You can install Julia by following the instructions on the Julia website. Dataset Some of the simulations for this project ran for more than a week on a cluster. The Trajectories folder contains results from these simulations, which are necessary to reproduce the figures in the paper. Each folder includes a summary file detailing the simulation parameters and other files containing measured quantities such as potential energy, average acceptance, and parameter values. These folders also contain results from postprocessing the original simulation trajectories carried out with atooms-postprocessing. Instructions To fully reproduce the analysis from postprocessed data, use the convenience script make at the root of the data set: ./make all This comprises two main steps: ./make setup will set up the software environment by checking the Julia installation and installing the required packages in a virtual environment. ./make workflow will execute the scripts to perform the workflow from the preprocessed trajectory files in the Trajectories folder. There is also a ./make clean option, which uninstalls the virtual environment and removes the results of the workflow. Notes Figure 3: This figure is the result of a two-dimensional simulation that requires the complete Policy-guided Monte Carlo code. Consequently, it is not reproduced by this workflow. Figure formats: The Julia scripts generate figures in SVG format. After producing the figures, you can use the ./make pdf option to convert them to PDF files. This option requires a working installation of Inkscape, which can be downloaded from Inkscape website Changelog 1.0.0: initial submission本数据集关联L. Galliano、R. Rende与D. Coslovich发表的《通用状态空间下的策略引导蒙特卡洛（Policy-guided Monte Carlo）方法：在玻璃形成混合物中的应用》研究，包含用于复现该论文所有配图的数据与工作流。 ## 运行要求 本工作流仅需正常安装的Julia编程语言（已在1.10.4版本完成测试）。可通过Julia官方网站的指引完成安装。 ## 数据集说明 本项目部分模拟任务在集群上运行时长超过一周。Trajectories文件夹存储了这些模拟的结果，是复现论文配图的必要数据。每个子文件夹均包含一份详述模拟参数的汇总文件，以及存储势能、平均接受率、参数值等可观测物理量的相关文件。这些文件夹同时包含使用atooms-postprocessing工具对原始模拟轨迹进行后处理得到的结果。 ## 使用说明 若要基于后处理数据完整复现分析流程，可使用数据集根目录下的便捷脚本make： ./make all 该命令包含两个核心步骤： 1. ./make setup：检查Julia安装环境，并在虚拟环境中安装所需依赖包，以此完成软件环境配置。 2. ./make workflow：运行脚本，基于Trajectories文件夹中的预处理轨迹文件执行完整工作流。 此外还提供./make clean选项，用于卸载虚拟环境并清除本次工作流生成的结果。 ## 注意事项 - 图3：该配图基于二维模拟生成，需完整的策略引导蒙特卡洛（Policy-guided Monte Carlo）代码支持，因此本工作流无法复现该图。 - 图像格式：Julia脚本默认生成SVG格式的配图。生成SVG图像后，可使用./make pdf命令将其转换为PDF格式文件。该转换步骤需提前安装正常运行的Inkscape软件，可从Inkscape官方网站下载安装。 ## 更新日志 1.0.0：首次提交