Zulip Chat Archive

Stream: new members

Topic: Order for cartesian closed categories


Gareth Ma (Jul 10 2025 at 19:50):

(deleted)

Gareth Ma (Jul 10 2025 at 19:50):

(deleted) It seems I might be missing something indeed...

Gareth Ma (Jul 10 2025 at 20:23):

Actually, I believe part of my original question still holds.

Looking at the definition of closed monoidal categories

/-- An object `X` is (right) closed if `(X ⊗ -)` is a left adjoint. -/
class Closed {C : Type u} [Category.{v} C] [MonoidalCategory.{v} C] (X : C) where
  /-- a choice of a right adjoint for `tensorLeft X` -/
  rightAdj : C  C
  /-- `tensorLeft X` is a left adjoint -/
  adj : tensorLeft X  rightAdj

It's defined as adjunction X[X,]X \otimes - \dashv [X, -], but it seems the convention is Y[Y,]- \otimes Y \dashv [Y, -] (so Hom(XY,Z)=Hom(X,[Y,Z])Hom(X \otimes Y, Z) = Hom(X, [Y, Z])). Is that "intended"?

Matt Diamond (Jul 10 2025 at 22:26):

when I was working on a proof of Lawvere's fixed-point theorem I noticed that the evaluation map seemed to have the arguments reversed from what was in the literature... I wonder if this is why

Matt Diamond (Jul 10 2025 at 22:36):

@Bhavik Mehta might know why it's designed this way


Last updated: Dec 20 2025 at 21:32 UTC