Project 3:Density Functional Theory and Machine Learning
MSE 215: Introduction to Computational Materials Science, Spring 2019
University of California, Berkeley
Instructor: Matthew Sherburne
Graduate Student Instructor: John Dagdelen
Run this repo in an online Binder by clicking the button below:
NOTICE TO BINDER USERS: YOUR NOTEBOOK PROGRESS WILL NOT BE SAVED IF YOU CLOSE THIS WINDOW OR LEAVE IT INACTIVE FOR TOO LONG.
PLEASE DOWNLOAD YOUR NOTEBOOKS AND FILES REGULARLY OR DOWNLOAD THIS REPO AND RUN OFFLINE ON YOUR MACHINE. See how to run offline for more info.
Problem Walkthroughs
The notebooks in this repository will walk you through typical workflows that computational materials researchers today would use to solve these problems today. A lot of the details of running DFT calculations have been abstracted away by software in recent years, which has helped make DFT more accessible, faster to use, and robust. We will be using these tools in this project to give you a taste (and some practical training) on how DFT is done today.
However, these abstraction layers can hide some important aspects of DFT that you should be aware of. With this in mind, we will also be asking you to prepare some input files by hand so you have a chance to see these details up close.
Feel free to complete this lab however you would like, but we highly recommend following the jupyter notebooks in this repo and using python to perform the analyses.
Table of Contents
Project Description | MSE 215 Project 3
Problem 1 (Due Tuesday, 04/02/2019):
- Walkthrough Notebook | 01 - Problem 1
Problem 2 (Due Friday, 04/12/2019):
- Walkthrough Notebook | 02 - Problem 2
Problem 3 (Due Friday, 04/12/2019):
- See description in MSE 215 Project 3
Problem 4 (Due Friday, 04/12/2019):
- Walkthrough Notebook | 03 - Problem 4
Project 3 Grading:
Format (this is a professional report) 10pts
Explain Calculations/Introduction 10pts
Output/Input files (use google drive) 10pts
DFT Calculations (and explanation):
(Just showing plots is not sufficient)
Diamond Cubic:
Energy convergence 5pts
Kpts convergence 5pts
Energy volume curve 10pts
Beta-Sn:
Energy convergence 5pts
Kpts convergence 5pts
Energy volume curve 10pts
What is the pressure for transformation 5pts
Compare to experimental value 5pts
Explanation 10pts
Thourough analysis of excerpt from literature 10pts
Bulk Modulus ML:
Linear Regression:
Plot predictions vs DFT 5pts
Report cross validation score 5pts
CV RMSE < 25 GPa 5pts
Discussion of results 5pts
Random Forest:
Plot predictions vs DFT 5pts
Report cross validation score 5pts
CV RMSE < 20 GPa 5pts
Discussion of results 5pts
Compare CPU time for DFT vs ML 10pts