Over the past month now I’ve been pivoting away from programming towards improving my mathematics skills. After doing a summer internship as a software engineer in avionics, I realized that I don’t just want to be good at system programming and low-level computer things, but that I want to get better at science.
This is why I’ve chosen to begin a masters degree in theoretical computer science and mathematics. For my first semester, I’ve decided to take complexity theory, graph theory, and information theory.
Complexity theory as I’ve always found Scott Aaronson’s guest lectures and some of his blog posts very intriguing. Graph theory only because it sounded interesting to me and because it seemed like a good course to practice writing proofs in, which I have no proper experience with. Come to think of it, I actually wrote my very first “proper” proof about a month ago for the course. And information theory as I did my bachelors thesis in information theoretic machine learning and want to gain a deeper understanding of the topic and become more comfortable with probability and statistics calculations.
Complexity theory has also been helpful so far in learning to write semi-formal proofs as literally every single lecture is just defining some problem and then proving that it either (a) belongs to a certain complexity class or (b) is reducible to some other problem showing that it is complete in that class.
With this blog post I want to sort of mark the beginning of my math and (continuation of my) research journey and have something to look back on if I do end up seeing this through. (I’m at a cross road in my life right now where I’m not sure if I want to continue the academic route and somehow get into research or start working at startups and writing software full time.)
I got a little inspiration recently from this blogs’ statement of purpose.
What are my goals?
Firstly I’d like to learn how to properly write proofs and become comfortable working on problems by simply sitting down and thinking about them. With this I’d also just like to generally improve my thinking and reasoning skills.
Regarding my university courses, ideally I’d like to get high marks in all three of them, but my main goal is to understand every single thing from every single lecture and homework. When I used to do math in high school or during the start of my bachelors degree, I would often skip definitions or lemmas here and there if they didn’t interest me. I also didn’t read to understand all that often and just told myself that simply reading over a set of definitions or rules was enough to continue past it, which often came back to bite me when I was studying for exams or trying to workout homework problems.
You can memorize derivation rules and such, but when you start working on problems that require proofs, deep understanding, and proper explanations, there is only one way forward and that is to literally understand every single word and statement related to that problem.
I’d also like to reach a level of comfort with math, theory things, and research such that I can have an idea of the direction that I want to go in for some sort of research project and/or my masters thesis. Just to carve out a little sub-area of a topic that interests me and diving deeper into it. This may also end up being theoretical deep learning things. (For example I’m very interested in The Platonic Representation Hypothesis right now.) Being better at mathematics in general will give me a much greater chance at understanding and making contributions in a field like this.
Another thing I’ve been very focused on is simply reading as much as possible every single day and becoming a better/more effective reader. I’ve noticed that having a higher reading comprehension level has made learning, studying and research so much more frictionless more efficient. Obviously reading that sentence back, it seems very obvious, but actually getting into the habit of reading with intention and as a pastime everyday has tremendously improved my quality of life from an intellectual point of view.
Things I read everyday: research articles, math textbook notes, lecture notes and materials, blog posts, articles on world economics and finances, fiction novels, and some non-fiction books every once in a while. With regard to the research articles, blog posts, and lecture notes, I want to start writing more notes for the bookmarks that I add to this site as well.
I think in general, this is the right path for me at the moment, even with “AI taking over all intellectual work.” I think it’s mostly a scare tactic or an over reaction. Understanding, knowledge, and reasoning skills are still the most important aspects of intellectual work for humans. LLM’s simply make this work faster to do for humans by automating repetitive work.
I am of the opinion that the truly important, difficult, and novel work is still only solvable by humans (or with partial AI assistance) as opposed to the idea that “robots” are taking over all of our intellectual work. I think with our current paradigm of artificial intelligence (LLM’s), this will always be the case.