Year 2022

Textbook

Rubin H Landau, Manuel J Paez, & Cristian Bordeianu: A Survey of Computational Physics -introductory computational science , Princeton University Press, 2008
Annotated copy
YOU have to read the book during the semester self-paced. There will be no detailed “lecturing” at the classes.
Previous year’s lecture recordings w/ discussion of chapters are available online in Teams

Useful reading materials

What Every Computer Scientist Should Know About Floating-Point Arithmetic link
- A detailed description of machine number representation, floating-point numbers, rounding errors

Overview and outlook presentation

Computation- and Data-intensive Approach in Sciences

2022.09.12

Course intro
Check out the project gudelines at the projects page ! Please note the deadlines!
Questions, requests, etc. concerning the class: contact: szamszimmsc(at)gmail.com

2022.09.19

NVIDIA is one of the major suppliers of GPU cards and developer of CUDA that is one of the workhorses of computer simulation and machine learning frameworks. Their conference GTC starts today. With free registration you can follow it online. As always several interesting talks and announcements are expected.
Lex Fridman’s podcast with Ray Kurzweil. Among other topics they talk about the exponentially growing technology.

2022.09.26

Do not forget to submit the Short description for your planned Project 1
A good “ground up” intro to algorithmic differentiation (or backpropagation) that is at the heart of many machine learning methods: https://www.youtube.com/watch?v=VMj-3S1tku0

2022.10.03

Another surprising breakthrough from deep mind: Faster algorithm for matrix multiplication:

Matrix multiplication - where two grids of numbers are multiplied together - forms the basis of many computing tasks, and an improved technique discovered by an artificial intelligence could boost computation speeds by up to 20 per cent … Hussein Fawzi at Deepmind says the results are mathematically sound, but are far from intuitive for humans. “We don’t really know why the system came up with this, essentially,” he says. “Why is it the best way of multiplying matrices? It’s unclear.” “Somehow, the neural networks get an intuition of what looks good and what looks bad. I honestly can’t tell you exactly how that works. I think there is some theoretical work to be done there on how exactly deep learning manages to do these kinds of things,” says Fawzi.

In the news: https://www.newscientist.com/article/2340343-deepmind-ai-finds-new-way-to-multiply-numbers-and-speed-up-computers/ (local copy: https://kooplex-fiek.elte.hu/seafile/f/aab941d3e99c49b199f8/ )
Original article in Nature: Fawzi, A., Balog, M., Huang, A. et al. Discovering faster matrix multiplication algorithms with reinforcement learning. Nature 610, 47–53 (2022). https://doi.org/10.1038/s41586-022-05172-4 https://www.nature.com/articles/s41586-022-05172-4

Previous links:

https://www.quantamagazine.org/mathematicians-discover-the-perfect-way-to-multiply-20190411/
https://hal.archives-ouvertes.fr/hal-02070778/document
Strassen method: https://en.wikipedia.org/wiki/Strassen_algorithm
Coppersmith–Winograd algorithm: https://en.wikipedia.org/wiki/Coppersmith%E2%80%93Winograd_algorithm)