HNHacker News
TopNewBestAskShowJobs

Chinjut

3,384 karma · joined June 23, 2014

submissionscomments
Chinjut··on From Sets to Categories (2023)
There are many results, but whether you would find them interesting is up to you. (I don't even know what it would mean for a result to not be a "statement" or "observation". Every group-theoretic result is a statement of an observation, and for that matter, the Riemann Hypothesis is a statement too.)

Since you mention Lagrange's theorem, I presume you are looking for introductory results rather than advanced results. E.g., the result that a functor has at most one left adjoint up to isomorphism (a direct generalization of the fact that inverses are unique in groups). Or the result that a functor which has a left adjoint automatically preserves limits (and the near converse to this given by the Adjoint Functor Theorem). Or the theorem that a category has a limit for the entire category if and only if it has an initial object. Or the Yoneda lemma, which is a direct generalization of Cayley's theorem for groups. Or Lambek's theorem that every initial algebra is invertible.

What is a group-theoretic result of interest to you, apart from the aforementioned Lagrange's theorem? It might in particular be good to see a group-theoretic result which you find interesting even outside the context of finite groups (as there is not such an emphasis on considering finiteness in typical category theory, though one could of course say things specifically about finite categories too).

Chinjut··on From Sets to Categories (2023)
The answers to these questions are exactly the same as for group theory or graph theory or linear algebra or any other such thing.
Chinjut··on From Sets to Categories (2023)
Categories are very similar to groups. If you can see how groups come up often and so it can be useful to think about the concept of groups, it's exactly the same thing with categories.

A group models the invertible maps from one structure to itself. A groupoid models the invertible maps between a number of structures. A monoid models the maps from one structure to itself (without presuming invertibility). A category models the maps between a number of structures (without presuming invertibility).

Chinjut··on Evaluating a class of infinite sums in closed form
This is exactly the approach described in the article.
Chinjut··on Math is running out of problems
No, while it may have been fun to be a generalist in Euler's era, that wasn't bothersome to me. To be clear, the issues I found in academia had little to do with math specifically, and affected academia broadly. The usual issues you've likely heard about dwindling ability to make a comfortable career of it without a great deal of luck.
Chinjut··on Math is running out of problems
For what it's worth, I also have a PhD in mathematics and I also ultimately left academia with some disappointment at the gap between what it is and my sense or fantasy of what it once was or could be.
Chinjut··on Math is running out of problems
Again, it's not clear to me that if you asked 1950s mathematicians to look at math journals in the 1950s, they would react very differently. You haven't presented any evidence of this.

It's certainly true that math is a more fragmented specialized field today than it was in the eras of Euler or Gauss, but without some more concrete evidence or objective claim, I don't know what is so bad about math journals today compared to 1950.

Chinjut··on Math is running out of problems
Is the test about ability to understand or about finding it very interesting? Of course it's easier to understand what has already become old hat and popularized in the field.
Chinjut··on Math is running out of problems
The article suggests this test to establish that math is running out of interesting research: "Take a fairly generalist journal, like the Journal of Algebra (take a topic in which you have expertise — my doctoral thesis was in algebra). Look at some of the papers. How many of them are truly interesting to you?".

But it seems to me fairly likely that applying the same test to math journals from 100 or 200 years ago would produce similar results. Most published papers will not be of great interest to any particular one person.

Chinjut··on The Invention of Zero
You can already represent infinitely many numbers with a finite set of symbols without positional notation, in unary. You can write "1 + 1 + 1 + 1" or "This number of stars: ****" or whatever. This idea (make N marks to represent the natural number N) has surely been understood about as long as counting has been understood.

The advantage of positional notation is greater efficiency at representing large numbers.

Chinjut··on Well-known paradox of R-squared is still buggin me
If you think that's bad, the R^4 coefficient is even lower.
Chinjut··on 25 years of video clips gone as Paramount axes Comedy Central wesbite
Let them make a new version if they want, sure, but removing access to the original is what rankles.
Chinjut··on The shape of information
What do you mean? It works the exact same way for 128 bottles. 64, 32, 16, 8, 4, 2, 1.
Chinjut··on The shape of information
This is essentially the same approach as the solution described in the article. You are observing the bits of the poisoned bottle number in binary, bit by bit. The only difference is that you avoid retesting any bottles once they have been ruled out by previous tests, whereas the fixed design in the article does not bother with this.
Chinjut··on Terence Tao on proof checkers and AI programs
The second ChatGPT example proof is mostly fine, except it failed to see that the claimed theorem isn't actually true at d = 2. (It writes "d ≥ 2" for a step in its reasoning where d ≥ 3 is actually required.)
Chinjut··on Terence Tao on proof checkers and AI programs
ChatGPT just spit out nonsense in the first example. Look at the sum in step 6, over τ ≤ τ. Is τ the bound variable of the summation or a free variable? Put another way, how many terms are in this summation over "τ ≤ τ"? And how does this relate to establishing either the LHS or RHS of part 3, to then conclude the other side? Nothing actually coheres. What happened here is that ChatGPT remembered the Möbius function for the partition lattice, regurgitated it* without providing proof, and then spit out some nonsense afterwards that looks superficially reasonable.

But establishing that Möbius function is essentially the whole ballgame! The question being asked is very nearly the same as just being asked to prove that the Möbius function has that form.

[*: The regurgitation also has a slight error, as μ(τ, σ) with τ a refinement of σ is actually (-1)^(|τ| - |σ|) * the product of (|τ restricted to c| - 1)! over each equivalence class c in σ. The formula given by ChatGPT matches this one when |σ| = 1, which is the important case for the proof at hand, but is incorrect more generally.]

----

For what it's worth, the desired fact can be shown just by elementary induction, without explicitly invoking all the Möbius function machinery anyway:

Let N(t, e) be the number of partitions of {1, ..., e} into t many equivalence classes [we will always take e to be a non-negative integer, but allow for t to be an arbitrary integer, with N(t, e) = 0 when t is negative]. The sum in question is of (-1)^t * t! * N(t, e) over all t. Refer to this sum as Sum(e). We wish to show that Sum(e) comes out to (-1)^e. Since Sum(0) = 1 trivially, what we need to show is that Sum(e + 1) = -Sum(e).

There are two kinds of partitions on the set {1, ..., e + 1}: Those which put e + 1 in an equivalence class all on its own and those which put it in the same equivalence class as some value or values in {1, ..., e}. From this, we obtain the recurrence N(t, e + 1) = N(t - 1, e) + t * N(t, e).

Accordingly, Sum(e + 1) = Sum of (-1)^t * t! * N(t, e + 1) over all t = Sum of (-1)^t * t! * N(t - 1, e) over all t, plus sum of (-1)^t * t! * t * N(t, e) over all t. The first of these two sub-sums can be reparametrized as the sum of (-1)^(t + 1) * (t + 1)! * N(t, e) over all t. Now recombining the two sub-sums term-wise, and keeping in mind (-1)^(t + 1) * (t + 1)! = (-t - 1) * (-1)^t * t!, we get Sum(e + 1) = sum of (-1)^t * t! * (-t - 1 + t) * N(t, e) over all t. As (-t - 1 + t) = -1, this comes to -Sum(e) as desired, completing the proof.

Chinjut··on The Backrooms of the Internet Archive
How did people hunting for the origin of this image discover the random niche website preserved by the Internet Archive that this image happened to come from?
Chinjut··on Auto-brewery syndrome in a 50-year-old woman
It took me nearly two decades of problems before I got a SIBO diagnosis, via breath testing. But now I'm not sure how much stock to put in such tests, given the very recent paper https://onlinelibrary.wiley.com/doi/10.1111/nmo.14817 which suggests positive breath test results may usually be an artifact of decreased transit time more than any small bowel overgrowth.

I definitely sympathize with you on doctors generally being unhelpful and uncurious.

Chinjut··on Transformers Can Do Arithmetic with the Right Embeddings
Minor point, but Blaise Pascal was centuries earlier than the 1850s.
Chinjut··on ChatGPT-4o vs. Math
I have a PhD in mathematics and I agree with you.
Chinjut··on OpenAI departures: Why can’t former employees talk?
There is absolutely something wrong with it. Just because a thing is common doesn't make it good.
Chinjut··on ChatGPT-4o vs. Math
That's fine but it's about the same as you'll get from an encyclopedia also, which makes sense as that's just where GPT got it from anyway. Nothing revolutionary in the ability to read encyclopedia articles. We've had that forever.
Chinjut··on Deactivating Facebook for just a few weeks reduces belief in fake news
CNN, MSNBC, Fox News, etc, are no better. They're just more established.
Chinjut··on ChatGPT-4o vs. Math
It's a choke because it failed to get the answer. Saying other true things but not getting the answer is not a success.
Chinjut··on New Foundations is consistent – a difficult mathematical proof proved using Lean
I do not imagine LLMs will be of any use here.
Chinjut··on Can GPT optimize my taxes? An experiment in letting the LLM be the UX
Why would the LLM be any good at determining where the input should be handled?
Chinjut··on Study: 61 UK firms tried a 4-day workweek and after a year, they still love it
Union meeting 1924: "Mr. President, do we really have to work every Saturday?"
Chinjut··on Marilyn vos Savant and the Monty Hall Problem (2015)
If you want the result where there is a lower conditional probability of your originally selected door having the prize versus the remaining unopened door, then it completely does matter to have the setup be "Monty always opens a door with a goat" instead of "Monty opens a door at random, and in this particular case, it happened to have a goat behind it".
Chinjut··on Institutions try to preserve the problem to which they are the solution
This is in fact the first example in the article.
Chinjut··on Sam Altman owns OpenAI's venture capital fund
I am having tremendous difficulty parsing this sentence.
← PreviousPage 7 of 32Next →