Zulip Chat Archive

Stream: Is there code for X?

Topic: MonCat is monoidal


Andrew Yang (Feb 03 2025 at 15:34):

Do we really not have this or do I need to open something

import Mathlib

open CategoryTheory MonoidalCategory

#synth MonoidalCategory MonCat

Markus Himmel (Feb 03 2025 at 15:35):

Seems like no one has given a docs#CategoryTheory.ChosenFiniteProducts instance for MonCat, but you can do

attribute [local instance] ChosenFiniteProducts.ofFiniteProducts

Andrew Yang (Feb 03 2025 at 15:38):

Is that the one we want? Or is the ChosenFiniteProducts given by actual PUnit and Prod better?

Markus Himmel (Feb 03 2025 at 15:40):

Yes, it would be better to have something like docs#CategoryTheory.typesChosenFiniteProducts in mathlib for the various other concrete categories.

Andrew Yang (Feb 03 2025 at 15:41):

Do we also want good defeqs for IsLimit.lift of the chosen limit cones? Or is it fine to get an arbitrary lift from ReflectsLimits?

Joël Riou (Feb 03 2025 at 15:42):

If we can have good defeq, I think it is better.

Andrew Yang (Feb 04 2025 at 00:37):

How does mathlib talk about cartesian monoidal categories? It seems like we use ChosenFiniteProducts, but what if the monoidal category instance is not constructed that way? e.g. CommMonCatᵒᵖ.

Adam Topaz (Feb 04 2025 at 00:43):

What’s the monoidal structure on CommMonCat^op?

Andrew Yang (Feb 04 2025 at 00:44):

The one coming from ChosenFiniteProducts CommMonCat (which is not yet in mathlib)

Andrew Yang (Feb 04 2025 at 00:44):

(and then docs#CategoryTheory.monoidalCategoryOp)

Adam Topaz (Feb 04 2025 at 00:49):

I guess you could introduce a ChosenFiniteCoproducts class, and show that C^op has such an instance provided that C has chosen finite products, then consider the cocartesian monoidal structure. Of course one has to be careful as we would very quickly hit some bad diamonds. Maybe we should also not make the ChosenFiniteProducts -> MonoidalCategory an instance?

Andrew Yang (Feb 04 2025 at 00:57):

Yeah that is one problem but the other problem I'm facing (or was; I don't need it anymore) is that the cartesian monoidal structure on CommMonCat is also co-cartesian because biproducts exists in CommMonCat.

Joël Riou (Feb 04 2025 at 14:48):

Andrew Yang said:

How does mathlib talk about cartesian monoidal categories? It seems like we use ChosenFiniteProducts, but what if the monoidal category instance is not constructed that way? e.g. CommMonCatᵒᵖ.

The general definition is MonoidalClosed!

Andrew Yang (Feb 04 2025 at 14:51):

I'm confused. Cartesian-monoidal and monoidal closed are different things?

Joël Riou (Feb 04 2025 at 14:53):

CartesianClosed is an abbrev.

Andrew Yang (Feb 04 2025 at 14:54):

Yes but I'm talking about not-necessarily-closed cartesian monoidal categories.


Last updated: May 02 2025 at 03:31 UTC