Robin Piedeleu

🚀 Call for Submissions – SYCO 13! 🚀 Join us in London (April 24-25) to share your work on category theory, logic, & more. ✅ Easy submission (papers, WIP, theses...), no proceedings 📅 Deadline: 14 March 2025 (this Friday!) 🔗 Details here: cl.cam.ac.uk/events/syco/13/

@arntzenius Do you know Dice? dicelang.cs.ucla.edu It's a functional PL for finite *probabilistic* programming that compiles to BDDs. It does not have higher-order functions, so it would be nice to work out how to compile H.-O. functions to BDDs without the probabilistic constructs

paper idea: finite functional programming

@hackerdocc @VictorTaelin No worries and thanks a lot for taking the time to research the answer!

@rwolffoot @VictorTaelin sorry for taking long, had to do some reading so, I'm not completely sure, but as bend doesn't have a type system, it appears this restriction is not enforced, but is instead considered like some sort of undefined behavior

RELEASE DAY After almost 10 years of hard work, tireless research, and a dive deep into the kernels of computer science, I finally realized a dream: running a high-level language on GPUs. And I'm giving it to the world! Bend compiles modern programming features, including: - Lambdas with full closure support - Unrestricted recursion and loops - Fast object allocations of all kinds - Folds, ADTs, continuations and much more To HVM2, a new runtime capable of spreading that workload across 1000's of cores, in a thread-safe, low-overhead fashion. As a result, we finally have a true high-level language that runs natively on GPUs! Here's a quick demo:

@hackerdocc @VictorTaelin Thank you! The paper attached discusses HVM2, which appears to disallow copying functions which copy their arguments. Do you know how Bend achieves this? Is it through disallowing these functions (with a type system, presumably) or via some transformation at compile-time or...?

@rwolffoot @VictorTaelin (I don't think he can see this since your account is private) HVM2 doesn't solve the matching fans, instead embracing the different semantics of the IC, disallowing cloning functions which clone arguments the paper explains this, also an article of his: @maiavictor/the-abstract-calculus-fe8c46bcf39c" target="_blank" rel="nofollow noopener">medium.com/@maiavictor/th…

@_julesh_ @coecke

@fabgenovese @bgavran3 Haha! Also wondered what the difference was...

@bgavran3 Like, they are literally the same thing, and come literally from the same words. Artificial means "made by skill", and, as such, "man-made". Artifact means "a thing made by skill", and a such, something "man-made". More Latin less improv please!

@bgavran3 It is useful to distinguish what is man-made and what is not, though, even if the distinction is culturally conditioned. This implies that the human agency--i.e. who, for what purpose, and with what data--behind ML or "AI" models (and all programs, really) should not be ignored.

Yes! Bloor's *Knowledge and Social Imagery* (which was apparently quite controversial) has a strong focus on logic and mathematics press.uchicago.edu/ucp/books/book…

Zooming out a bit (a lot!), it looks like the so-called "strong programme" of David Bloor, Barry Barnes and others in the sociology of scientific knowledge is what I'm after. Will need to dig to see whether any of them has looked at maths or other deductive sciences specifically

@_julesh_ betterwithout.ai/gradient-disse…

Gradient dissent

@MathisHammel "En un coup d'oeil" plus efficace que le discriminateur du GAN qui a produit ces images... la classe !

Pas de chance pour Richard, j'ai beaucoup travaillé sur les faux visages générés par intelligence artificielle et je sais les reconnaître en un coup d'oeil. L'indicateur le plus évident, c'est la position des yeux : certaines IA vont toujours les placer au même endroit.

THREAD #OSINT - Comment j'ai découvert un réseau d'entreprises fictives sur LinkedIn qui administre des centaines de faux profils. La semaine dernière, j'ai reçu une invitation LinkedIn de ce profil qui m'a tout de suite paru suspect. Je vous explique ⤵️

@prathyvsh Thank you! (I have read 1/, though it felt closer to philosophy, with its focus on an idealised scenario, in which broader social factors are mostly absent. I should still reread it and pay closer attention to the footnotes and appendix, which link the main text to real events)

@rwolffoot Here is a small list all of which have varying overlaps with the topic: 1/ Proofs and Refutations by Lakatos: archive.org/details/proofs… 2/ Against Method by Feyerabend: en.wikipedia.org/wiki/Against_M… 3/ The Nature and Growth of Modern Mathematics by Kramer: archive.org/details/nature…

What are good books/articles about the sociology of mathematical research? As in, the study of the social factors that influence (the formation of) research communities, guide their activity and how they produce mathematical knowledge?

@tangled_zans Very cool! Several people have now suggested we include string diagrams for premonoidal categories, à la Jeffrey. They are quite intuitive for computer scientists, as diagrams representing state-passing operations. I think they'd be a nice addition to the next version!

@rwolffoot this is super cool! wish something like this was around when I first learned about this stuff :D btw, you might find this interesting - ive been experimenting with (pre)monoidal diagrams as inductive types

Just released a small update to this paper, with more examples of diagrammatic reasoning in action. arxiv.org/abs/2305.08768 Thanks for all the suggestions. We could not include all of them yet, but will certainly consider them for future versions.

@Twisol We use the TikZ library with a GUI tool called TikZit (tikzit.github.io). (And that's the one I used to draw the particular diagram you mention. I assure you there's nothing magical about it!

@rwolffoot Hi! This paper is really well-written; I'm enjoying it a lot. Can I ask if there's a particular library you use to render string diagrams? The one on page 10 (definition 2.4) seems like a feat of magic, and I'd love to know how to draw decompositions in the same way.

@SovsDenEtEller Yes, don't hesitate to DM me with them!

Curious about string diagrams and wondering where to start? Fabio Zanasi and I wrote an introduction designed to onboard CS students into our own research and offer a glimpse into current research on the topic. arxiv.org/abs/2305.08768 This is a first draft--feedback welcome!

@sbbls I use tikzit.github.io, but I'm also aware of rewalt.readthedocs.io, which can output tikz.

@rwolffoot this looks very interesting! would you have any tips/package recommendations about how to do these string diagrams easily in tikz/latex? i have no idea where to look at

@cole_comfort And I'll add diagrams for premonoidal categories and (di)operads to the list of suggestions I've already received.

@cole_comfort Yes, we plan to update it; that's why we would like feedback! If you feel that something is missing, and that we misrepresented or did not cite some work, please let me know. Of course, we won't be able to incorporate everything, but we will certainly consider all suggestions!