In this two lectures I will give in Fall 2023 MatSE 580 (Computational Thermodynamics) at Penn State, which together should provide students with some basic skills in:
- Manipulating and analyzing materials - using pymatgen
- Setting up a small NoSQL database on the cloud to synchronize decentralized processing - using MongoDB Atlas Free Tier
- Interacting with the database and visualizing the results - using pymongo library and MongoDB Charts service
- Installing and running machine learning (ML) tools to predict stability of materials - using pySIPFENN and, if time allows, ALIGNN
- Using ML featurization and dimensionality reduction to embed materials in feature space - using pySIPFENN with MongoDB Charts visualization
The Lecture1.ipynb Jupyter notebook covers points 1, 3 (interaction part), 4 (install part). Points 2 and 3 (visualization part) will be done outside of this environment by student during the lecture.
The Lecture2.ipynb Jupyter notebook covers points 4 and 5, with some final visualization done outside of this environment.
Ideally, you should follow all instructions on your personal computer, so that afterwards you have a neat setup ready to tackle future problems of your choosing or use it in your MatSE 580 final project. I strongly suggest to use VS Code IDE for consistency in the class and with alternative setups (see below), but you are welcome to use anyhing of your choosing.
If for any reason use of personal computer is not possible, you can use GitHub Codespaces which are development containers running in your browser. On the free tier (120 CPU-h/month), you should be able to get 30h of work done on a moderately powerful (4 core / 16GB RAM) machine which should be plenty for this lectures and class project. To start a codespace, you simply go to green Code<> button above and then follow Codespaces -> *** -> New with options..., then make sure main, US East, and 4-core are selected, and finally Create a Codespace. Wait a moment and you should see a nice VS Code environment right in your browser!
See Lecture1.ipynb
Your local changes will be persisted in both local and Codespace environments.
It is good practice to also Commit your progress along the way. If you are using VS Code environment, either locally or through Codespaces, you can go it efortlessly by (1) going to Source Control (3rd icon in the left-side menu), (2) typing a short message, and (3) clicking Commit. If you have forked this repository, you can also Push to "upload" it to GitHub.