Generative AI can be magical, but the mathematical ideas and intuition underlying the popular 'Stable Diffusion' approach can be opaque and somewhat inaccessible. However, these ideas are interesting so we will approach the topic in a distilled manner - understanding it as a 'Galton Board', which will not demand any prior mathematical or AI background.
Mathematics is usually written using symbols, however, these symbols are not magic, but purposeful inventions. Is there an interesting and useful example of non-symbol-based mathematical notation? There is and we will do an introduction to Roger Penrose's (physics Nobel Laureate) graphical notation, which we will apply to a classical physics problem. Familiarity with vector calculus will be assumed.
The concept of Gambler's Fallacy is well known - it is an erroneous belief that future events are affected by past outcomes in games of luck. It is the feeling that given a bad run one's luck is likely to turn. But the gambler is ready to give you a bet that he is right - would you be ready to take it?
There is more than one way of doing Generative AI, and we will be looking at a model that can be approached using a lightly flavored physics angle of fluid dynamics - the Continuous Normalizing Flows and Flow Matching model. Prior knowledge of calculus and a basic understanding of Generative AI as such will be assumed.
Nobel Laureate E. Wigner remarked that it is a miracle how well mathematics and physics fit. Physics can motivate and inspire mathematics, but can in reverse mathematics point to something before physicists get to it? To discover the answer we will act as explorers of a specific example - the mathematics of spinors.
A simple mathematical game of 'what if' in exploring how 3 dimensions are special in Newton's Law of Gravity for life to exist. Basic familiarity with vector calculus will be assumed.
A story of whales, highways, and bold solutions.
It is no fun to wait, which also applies to waiting on outputs of Large Language Models. There is a need for speed and there is a fun sampling trick to speed up the output generation. Prior knowledge of basic LLM will be assumed.
The 2022 Nobel Prize in physics was awarded for experiments showing that the world is very weird. Weird in what way? That the following two things are not true simultaneously - things have clear existence and things interact only adjacently. An elegant experiment with elegant reasoning, showing that the universe is a strange one.
Pablo Picasso once said that 'Art is the elimination of the unnecessary', so what does computing look like when everything has been eliminated to minimum? We will explore the heart of computing through visualized lambda calculus. No prior knowledge will be assumed.
What does it mean that physics is 'beautiful'? Mathematics is the language of physics, so is part of the story. We will explore this notion of beauty in physics by the example of 'inventing' Special Relativity based on principles of simplicity and elegance.