Zulip Chat Archive

Stream: lean4

Topic: pushing casts into structures


James Gallicchio (Jun 26 2023 at 05:53):

for a i : Fin n, if you cast with h : n = m, the cast can be pushed into the structure like so:

theorem push_cast (a : Fin n) (h : n = m) : h  a = a.1, h  a.2 :=
  by cases h; simp

is there a better way to deal with this sort of situation? marking this lemma simp solved my particular problem but I'm not sure if I'm totally off base.

James Gallicchio (Jun 26 2023 at 05:53):

I'm trying to reduce the number of suffices ... I write :laughing:

Kyle Miller (Jun 26 2023 at 06:07):

I find it's usually better to avoid explicit rewrites like this and instead develop API for rewriting type arguments in a more controlled way. In this case, there's already docs4#Fin.cast

Kyle Miller (Jun 26 2023 at 06:10):

The fact Fin.cast gives an order isomorphism complicates things a bit, but it also unlocks further simplification opportunities.

James Gallicchio (Jun 26 2023 at 06:10):

Yeah. I'm working in Std right now so I will need to move the Fin.cast infrastructure up to Std (this is fine).

James Gallicchio (Jun 26 2023 at 06:14):

oh, actually, it is not fine, since Fin.cast is an order isomorphism. hm

James Gallicchio (Jun 26 2023 at 06:16):

I don't think we're trying to move all of that infrastructure up to Std yet. @Mario Carneiro thoughts? maybe something akin to push_cast (but not @[simp]) is a reasonable addition to Std.

Mario Carneiro (Jun 26 2023 at 06:17):

Fin.cast can be a function

James Gallicchio (Jun 26 2023 at 06:18):

and then rename the mathlib isomorphism?

Mario Carneiro (Jun 26 2023 at 06:18):

once upon a time it was a function, but people started upgrading functions to equivs because it is notationally almost invisible, although this is less true in lean 4 than in lean 3

Mario Carneiro (Jun 26 2023 at 06:19):

it's only used as a function most of the time, IMO it was the wrong call to make it the default

Kyle Miller (Jun 26 2023 at 06:35):

Adding to this, I'd suggest that we rename mathlib's Fin.cast to say Fin.castIso or Fin.castEquiv, keep the API surrounding it, and add a lemma that Fin.cast n h = Fin.castEquiv n h. There are places where you do want the equivalence version.

James Gallicchio (Jun 26 2023 at 06:39):

I probably can't do the refactor today, so if someone is feeling like doing a little refactor feel free :) otherwise I'll work on it tomorrow.


Last updated: Dec 20 2023 at 11:08 UTC