Jean Abou Samra
banner
jeanas.bsky.social
Jean Abou Samra
@jeanas.bsky.social
PhD student in theoretical computer science at Eötvös Loránd University in Budapest. Mainly here to chat about TCS/math.
(I meant “factor through true : 1 → Ω”.)
November 25, 2025 at 8:57 AM
I think the stupid trick is this:

Assuming p(x) ⊢ q(x), apply the definition to the subobject ι : {x ∈ X | p(x)} ↪ X. Certainly the map p ◦ ι : {x ∈ X | p(x)} ↪ X → Ω will factor through 1. So q ◦ ι : {x ∈ X | p(x)} ↪ X → Ω will also factor through 1 and we can work from there.
November 25, 2025 at 7:28 AM
And this:
November 24, 2025 at 10:23 PM
Oh and by the way, Lambek & Scott disagree with your characterization of the theorem as “nontrivial”. This is all they have to say about the proof:
November 24, 2025 at 10:15 PM
Offhand I don't actually recall seeing a topos theory book written in the past 15 years. I do have some lecture notes like those by Jaap van Oosten. They're quite tough to follow but still more to my taste.
November 24, 2025 at 9:53 PM
Sadly, no.
November 24, 2025 at 9:50 PM
And a contrario, there are things that they make overly minimalistic (perhaps shorter but less natural): using impredicative encodings seems really useless to me (it's like defining groups by axiomatizing division instead of product — is there a single theorem that this helps prove?).
November 24, 2025 at 9:49 PM
No idea why all those old topos theory books, and especially this one, complicate this definition so much.
November 24, 2025 at 9:40 PM
Especially since they've just spent so many pages on cartesian closed categories and satisfaction in a topos can (I'm pretty sure) be defined by just reusing the cartesian closed structure (the topos structure just gives you some extra distinguished objects and morphisms).
November 24, 2025 at 9:40 PM
I can imagine how this characterization might be useful as a theorem, but it strikes me as really odd to choose it as the definition.
November 24, 2025 at 9:37 PM
Oh I think I see why it works. But everything in this book is so weirdly presented…
November 24, 2025 at 6:51 PM
Quand je vois que ceci était la une du « Figaro » hier, je me demande si ce moment n'est pas déjà arrivé, en fait.
November 23, 2025 at 9:54 PM
Enjoy 🤓

(yes, @gro-tsen.bsky.social, I know you're going to react with some combination of 🙄, 😫 and 😱 by emoji kitchen 😉)
November 21, 2025 at 11:12 PM
cemulate.github.io
November 16, 2025 at 6:12 PM
et normal est dénombrablement paracompact, avant d'enchaîner sur le fait que tout de même, il y a quelques implications de base que tout mathématicien devrait connaître, comme le fait que tout espace T₁ pseudonormal est régulier (quelque chose de ce genre).
November 16, 2025 at 1:00 PM
J'ai récemment ouvert un bouquin de topologie, que je ne retrouve malheureusement pas, dans la préface duquel l'auteur se plaignait que la « botanique » des espaces topologiques était souvent portée à l'excès et qu'on se fichait vraiment de savoir si tout espace dénombrablement métacompact …
November 16, 2025 at 1:00 PM
Oui, ça s'appelle les maths synthétiques, et c'est un sujet de recherche actif et très intéressant. Et toute personne qui a fait des maths dans le supérieur a très probablement déjà rencontré ce type d'approche : qui a commencé l'analyse avec une construction des nombres réels ?
November 16, 2025 at 11:11 AM
C'est quoi, par curiosité ?
November 14, 2025 at 4:12 PM
I actually spent most of the time doing an informal introduction to realizability since the audience was not familiar with it. In case this interests anyone, the slides are here:

jean.abou-samra.fr/talks/2025-1...
jean.abou-samra.fr
November 10, 2025 at 10:32 PM
C'est sans doute un peu extrême mais personnellement, depuis quelques versions de Fedora (qui arrivent tous les 6 mois), je réinstalle complètement plutôt que de mettre à jour. L'intérêt est surtout de vérifier que les sauvegardes que j'ai suffisent à tout remettre en ordre en cas de pépin.
November 9, 2025 at 2:59 PM