Data_Science_NL_Potential

This is the repository for the final project for the class 'introduction to data science' at the Rutgers University Newark.

Fall semester, 2017

Professor: Patrick Schafto

People: Michele Pavanello, Wenhui Mi, Rupali Chawla, Johannes Tölle, Alina Umberkova

Question: Can the noninteracting Kinetic Energy electron density functional be approximated by data science techniques based on machine learning?

In the following a brief overview about the work, which needs to be done and the motivation is given:

Why it is an interesting question?

By achieving this we would be able to significantly reduce the computational complexity for computing the noninteracting kinetic energy functional - a fundamental ingredient of computational materials science.

This will have an immense impact in the field of chemistry and material science because it would open the door to modeling materials of realistic dimensions and complexity which are currently not approachable by the state-of-the-art methods (such as Kohn Sham Density Functional Theory).

Work, that needs to be done:

• Create the needed data to apply different mashine learning algorithms.

Section for interesting matrial concerning theory:

https://www.youtube.com/watch?v=HC0J_SPm9co (Introduction to Scikit Learn)

Paper:

Review: https://arxiv.org/pdf/1510.07512.pdf