CoDaDri

Computational and Data Driven Physics

Modern physics is characterized by an increasing complexity of systems under investigation, in domains as diverse as condensed matter, astrophysics, biophysics, etc. Establishing adequate models to describe these systems and being able to make quantitative predictions from those models is extremely challenging. The goal of the course is to provide the tools and concepts necessary to tackle those systems.

Course description

We will first cover many algorithms used in many-body problems and complex systems, with special emphasis on Monte Carlo methods, molecular dynamics, and optimization in complex landscapes.

Second, we will provide statistical inference and machine learning tools to harness the growing availability of experimental data to design accurate models of the underlying, complex, strongly non-homogeneous and interacting systems.

Each theoretical lecture will be followed by a tutorial illustrating the concepts with practical applications borrowed from various domains of physics. We will focus on methods and algorithms and physics, not on programming and heavy numerics! You will have to hand in 3 homeworks.

The Team

• Alberto Rosso (Computational physics)
• Rémi Monasson (Data-driven physics)
• Simona Cocco & David Lacoste & Michel Ferrero (Tutorials)

Where and When

• Lectures on Fridays: 14:00-16:00
• Tutorials on Fridays: 16:00-18:00
• ENS, 29 rue D'Ulm, salle Borel + Djebar

Computer Requirements

No previous experience in programming is required.

Programming Language: Python

For practical installation, we recommand either to use Anaconda (See Memento Python) or use google colab.
The Collaboratory platform from Google is quite good way to use powerful computer without buying one: It requires no specific hardware or software, and even allows you to use GPU computing for free, all by writting a jupyter notebook that you can then share.

Grading

Computational Physics:

• Homework 1: 5 points
• Homework 2: 5 points
• Multiple Choice Questions on November: 10 points

Data Driven Physics:

• final exam in January: 20 points

Schedule

Friday, September 3, 2021

• Lecture 1 Introduction to Monte Carlo

• Tutorial 1 Markov Matrix

Friday, September 10, 2021

• Tutorial 2 Markov matrix problems

• Tutorial 3 Thumb rule problems

Homework:

Friday, September 17, 2021

• Lecture 2 Basic Sampling

• Lecture 3: Errors and Precision

Friday, September 24, 2021

• Lecture 4: Ising model and phase transitions

• Tutorial 4: Ising model and phase transitions problems

Friday, October 1, 2021

• Lecture 6: Importance sampling

• Tutorial 6: Faster than the clock algorithms problems

Homework 2:

Friday, October 8, 2021

• Lecture 7: Optimization & Dijkstra algorithm

• Tutorial 7: Simulated annealing problems

Friday, October 15, 2021

• Lecture 8: Introduction to Bayesian inference

• Tutorial 8:

Friday, October 22, 2021

• Lecture 9: Asymptotic inference and information

• Tutorial 9:

Friday, October 29, 2021

• Lecture 10: High-dimensional inference and Principal Component Analysis

• Tutorial 10:

Friday, November 26, 2021

• Lecture 11: Priors, regularisation, sparsity

• Tutorial 11:

Friday, December 3, 2021

• Lecture 12: Network inference

• Tutorial 12:

Friday, December 10, 2021

• Lecture 13: Supervised learning and phase transitions

• Tutorial 13:

Friday, December 17, 2021

• Lecture 14: Unsupervised learning and representations

• Tutorial 14: