This course will introduce modern machine learning techniques for studying classical and quantum many-body problems encountered in condensed matter, quantum information, and related fields of physics. Lectures will emphasize relations between statistical physics and machine learning, while tutorials will include hands-on experience in programming with applications.

Topics to be covered include lattice models for statistical physics, Monte Carlo methods, supervised and unsupervised learning, neural networks, Boltzmann machines, and deep learning. It would be useful if participants had basic knowledge of programming in any language. Tutorials will be given in Python and TensorFlow.

There is no registration fee and limited funds are available for travel and local expenses.

Lecturers: Juan Felipe Carrasquilla (D-Wave Systems Inc., Canada) & Roger Melko (University of Waterloo & Perimeter Institute , Canada)

Program:

Day 1: Statistical mechanics, Monte Carlo
- Lecture 1: Ising model, Gauge theories
- Lecture 2: Monte Carlo simulations
- LAB: Monte Carlo in Python

Day 2: General introduction to Machine Learning
- Lecture 1: Linear Fitting, Regression, Supervised learning
- Lecture 2: Supervised Learning for Ising systems and Backpropagation
- LAB: Feedforward Neural Network

Day 3: Supervised and Unsupervised Learning
- Lecture 1: Convolutional Neural Networks (CNNs)
- Lecture 2: Introduction to Unsupervised Learning, PCA
- LAB: CNN for Ising gauge theory

Day 4: Restricted Boltzmann Machines (RBMs)
- Lecture 1: RBMs for classical systems
- Lecture 2: RBMs for quantum systems
- LAB: An RBM for the Ising model

Day 5: Research Frontiers
- Lecture 1: Quantum State Tomography
- Lecture 2: Quantum Machine Learning
- LAB: Quantum tomograpy of the W state

Organizers:

• Nathan Berkovits (ICTP-SAIFR & IFT-UNESP)
• Alexandre Reily Rocha (IFT-UNESP)
• Pedro Vieira (ICTP-SAIFR & IFT-UNESP & Perimeter Institute)