Zulip Chat Archive

Stream: Is there code for X?

Topic: free or walking isomorphism?


Emily Riehl (Sep 16 2024 at 21:55):

Does mathlib know about the free category containing an isomorphism, which has two objects and a unique arrow in each hom-set? The quiver structure could be defined as follows:

/-- This is the free-living isomorphism as a category with objects called
`zero` and `one`. Perhaps these should have different names like `src` and `tgt`?-/
inductive walkingIso : Type u where
  | zero : walkingIso
  | one : walkingIso

open walkingIso

namespace WalkingIso

/-- The arrows in the walking iso category split into three cases.-/
inductive Hom : walkingIso  walkingIso  Type v where
  | id : (X : walkingIso)  Hom X X
  | hom : Hom zero one
  | inv : Hom one zero

/-- The quiver structure on `walkingIso`-/
instance : Quiver walkingIso where
  Hom := Hom

end WalkingIso

Adam Topaz (Sep 16 2024 at 22:01):

I don't think we have this.

Kim Morrison (Sep 16 2024 at 23:51):

What does this definition gain over just Hom X Y := PUnit?

Emily Riehl (Sep 17 2024 at 16:37):

I had thought it might be better to give explicit names for the key arrows? But then we ended up writing inductive proofs that were annoying and unenlightening. You and @Joël Riou have convinced me that using PUnit is better.

Emily Riehl (Sep 17 2024 at 16:39):

Relatedly, it seems that mathlib doesn't have a notion of "indiscrete" or "chaotic" or "contractible" category, defining the right adjoint to the functor ob : Cat -> Set. While "discrete" categories already exist, it seems that they haven't been shown to define the left adjoint to this functor either.

I have some students looking for a beginner category theory project, so I might suggest this to them. Which name do folks prefer?

Johan Commelin (Sep 17 2024 at 18:14):

I'll vote for Indiscrete

Dagur Asgeirsson (Sep 17 2024 at 19:48):

Emily Riehl said:

While "discrete" categories already exist, it seems that they haven't been shown to define the left adjoint to this functor either.

I'm pretty sure this is in mathlib, let me find it

Dagur Asgeirsson (Sep 17 2024 at 19:49):

docs#CategoryTheory.Cat.typeToCatObjectsAdj

Dagur Asgeirsson (Sep 17 2024 at 19:51):

But indeed we do not have the right adjoint

Kim Morrison (Sep 18 2024 at 00:21):

I like Indiscrete too, or Contractible just as much.


Last updated: May 02 2025 at 03:31 UTC