Date and time: Wednesday and Friday 13:00 -- 15:00
First class: August 20
Location: PUC-Rio, Leme building, Sala L754
Join Whats App group https://chat.whatsapp.com/GrgDTlkiW5w3HbzuDxGyla
Tentative Schedule
20.08 13-15 Lecture 1.
22.08 13-15 Lecture 2.
27.08 13-15 Lecture 3.
29.08 13-15 Lecture 4.
03.09 13-15 Lecture 5.
05.09 13-15 Lecture 6.
10. 09 13-15 Lecture 7.
12.09 13-15 Lecture 8.
17.09 No Class. To be rescheduled
19.09 No Class. To be rescheduled
24.09 13-15 Lecture 9.
26.09 13-15 Lecture 10.
01.10 13-15 Lecture 11.
03.10 13-15 Lecture 12.
08.10 13-15 Lecture 13. Online.
10.10 13-15 Lecture 14.
15.10 13-15 Lecture 15.
17.10 13-15 Lecture 16.
22.10 13-15 Lecture 17.
24.10 13-15 Lecture 18.
29.10 13-15 Lecture 19.
31.10 13-15 Lecture 20.
05.11 13-15 Lecture 21.
07.11 13-15 Lecture 22.
12.11 13-15 Lecture 23.
14.11 13-15 Lecture 24.
19.11 13-15 Lecture 25.
21.11 13-15 Lecture 26.
26.11 13-15 Lecture 27.
28.11 13-15 Lecture 28.
03.12 13-15 Consultation.
05.12 13-15 Consultation.
10.12 11-17 Exam.
12.12 11-17 Exam.
Topic: Study basics of applying of machine learning technics for solving ordinarily differential equations and partial differential equations
Description:
In the last years machine learning becomes a major tool in a lot of practical disciplines. In the course we consider applications of such type of techniques for numerical simulation of ordinarily partial differential equations. The main difference with the classical learning is that instead of trainining dataset we use differential equations. In the course we would consider 3 main techniques Residual Neural Networks (ResNets), Physically inspired neural networks (PINNS), and Gaussian-process regression. We would consider theoretical basis as well as positive (and negative) examples of applications.
The topics includes
-- Basics of neural networks
-- Basics of numerical methods for ODE
-- Residual neural networks
-- Solving ODE with residual neural networks
-- Neural networks as universal approximator
-- Physically inspired neural networks
-- elliptic equations
-- parabolic equations
-- Bayesian method
-- Basics of Gaussian Fields
-- Gaussian-process regression
Prerequsits: Ordinarily differential equations, Probability
Literature:
Hennig, Philipp, Michael A. Osborne, and Hans P. Kersting. Probabilistic Numerics: Computation as Machine Learning. Cambridge University Press, 2022.
Mathematical Introduction to Deep Learning: Methods, Implementations, and Theory, Arnulf Jentzen, Benno Kuckuck, Philippe von Wurstemberger, 2025
Gaussian Processes for Machine Learning, C. E. Rasmussen & C. K. I. Williams, MIT Press, 2006
Parallel Computing and Scientific Machine Learning (SciML): Methods and Applications, Chris Rackauckas, https://book.sciml.ai/
Differential Equations for Continuous-Time Deep Learning, Lars Ruthotto, https://arxiv.org/pdf/2401.03965