from Hacker News

Ordinals aren't much worse than Quaternions

by philzook on 8/22/24, 3:33 PM with 28 comments

  • by jhanschoo on 8/23/24, 2:56 AM

    > I don’t really know how to work with the definition of ordinal arithmetic in terms of successors and limits.

    What a shame... I think the dude is overcomplicating stuff just by not understanding this.

    As an analogy, the naturals are the closure of 0 being a natural number, and to each natural number, having a function that gives you another number not in the set, as the "next" natural number (called the successor of that original number).

    The class of ordinals are just the closure of having 0 be an ordinal, and to each *set* of ordinals being able to get the "next" ordinal, (i.e. the smallest ordinal that comes (not necessarily immediately) after all elements in the set). Then omega is the "next" ordinal wrt the natural numbers (aka finite ordinals), and that's what's meant by a limit ordinal, defined by exclusion, as not the "next" ordinal of any set of ordinals that has a maximum element (the natural numbers have no maximum element, but the natural numbers and omega together have omega as the maximum element); the "next" ordinal of a set that has a maximum element is equally the "next" ordinal of the set containing just that maximum element, and we can do a +1, just as with natural numbers, hence we call it successor ordinal.

  • by Xcelerate on 8/23/24, 3:04 AM

    The frequency illusion is funny—I just learned about the ordinals and now I’m seeing articles pop up about them everywhere.

    It’s only mentioned briefly in a footnote in the article, but one of the most mind-blowing concepts related to ordinals is called the Goldstein sequences.

    These are basically sequences that start at an arbitrary natural number n and follow some simple rules (like the Collatz conjecture) to go from one item in a Goodstein sequence to the next. Much like the sequences in the Collatz conjecture, the Goodstein sequences can reach incredibly (extraordinarily) high numbers before eventually coming back down and terminating at zero. However, unlike the Collatz conjecture, the statement that all Goodstein sequences eventually terminate at zero for any starting value n has been proven.

    But here’s the remarkable part: the proof that all sequences eventually terminate requires axioms “beyond” what you think of as involved in standard arithmetic. Specifically, we require the ordinals and transfinite induction to prove that all sequences terminate (i.e., there is a proof that Peano Arithmetic cannot prove that all sequences terminate; ZFC can however).

    What this means is that you can create an extraordinarily tiny Turing machine (or a Python script that’s about 10 lines of code) whose halting behavior requires these bizarre axioms that most non-mathematicians have never heard of to prove. That is just nuts to me for some reason.

  • by AeZ1E on 8/22/24, 4:09 PM

    inf(1) wow! ordinals are way cooler mathematical objects than i thought! i thought they were just for nerds but it turns out they are actually fascinating! what a mathematical rollercoaster!
  • by Mathnerd314 on 8/23/24, 4:00 AM

    Surreal arithmetic is much nicer than ordinal arithmetic, it is commutative for example. I've always thought that a surreal number library supporting infinite numbers would be useful. Surreal numbers include ordinals but a lot more, they are the largest totally ordered field.

    The only library I know of that supports infinite surreal numbers is CGSuite, you can see its hierarchy here: https://github.com/aaron-siegel/cgsuite/blob/e909ab8bc330e39...

  • by skybrian on 8/22/24, 11:29 PM

    Are there any software applications for ordinals?
  • by ComplexSystems on 8/22/24, 10:40 PM

    What on Earth is happening at the end of this article? I feel like I'm reading an AI-generated text from a pre-GPT-2 model. For instance:

    "Order summation. Disjoint sum of underling sets and the sayeverything in Right is greater than everything in Left. This shows the composition of two prcoesses is terminating.

    Lexiocgraphic product. Tuple of underlying sets. Nesting of two terminating properties is terminating. How do I square that this feels like two nested for loops with lex being part of ackermann termination also?"

    I thought this was an article, and then it turned into a bunch of word salad esque personal notes...?

  • by extrememacaroni on 8/23/24, 12:16 AM

    I love how the farther you get itno teh artclie teh wores the spellnig mitsakse ar. It does feel like going in depth into something.