It's not really obscure. This is undergrad linear algebra material

There's virtually an infinitude of mathematics at all levels. One could post about any of it randomly, but perhaps it makes sense to have some focus and relevancy to at least - 1%? - of the HN audience. A post on a radically faster new method of factoring large numbers might appear "just mathematical" but it would have extreme relevance to public key cryptography, for example, with enormous consequences overall. That's probably obvious to many - much more obvious than the import of the article commented on here, which is exactly why I wondered and whether it's something of larger importance that I was missing.

Methods from Numerical Linear Algebra are at the heart of a lot of dimensionality reduction techniques.

Moreover, Nick Higham is one of the most prominent numerical analysts today. He spends a lot of time making clear expositions of arcane material (just this blog post is an example - it ends by saying computation is not important in practice, and lists where this method is used.)

He is the editor of the Princeton Companion to Applied Mathematics. He has also written nice books on mathematical writing. He has a recent book on "how to be creative", which is on my to-read list.

So, not really arcane for the HN crowd - at least, not ideally so.

Well, thank you for an informative response - the kind of reply I wanted.

I do have to ask: what do you think the "HN" crowd is? And why would they ideally be highly mathematical? It might in fact be interesting to try to statistically characterize the composition of HN readers in terms of professions and interests.

This is related to minimal realizations of state space models in control theory

Thanks.

Despite all appearances this isn’t (yet) LLM News. There are a lot of math, physics, numerical, quant, etc folks here that are not exclusively interested in LLM articles.

What’s LLM?

Nice downvoting for asking a serious question about non-obvious relevancy. Evidently, asking about relevancy is verboten on Hacker News by those who think that a query can't just be to elicit information about utility rather than some assumed critique.

Seriously, don't be surprised when many would rather interact with GPT-4 than humans.