From: Dan Licata <d...@cs.cmu.edu>
To: Andrew Polonsky <andrew....@gmail.com>
Cc: Homotopy Type Theory <HomotopyT...@googlegroups.com>
Subject: Re: [HoTT] A puzzle about "univalent equality"
Date: Mon, 5 Sep 2016 17:51:06 -0400 [thread overview]
Message-ID: <9064B371-C3B0-45F5-8720-90E6D10E211C@cs.cmu.edu> (raw)
In-Reply-To: <876efd6d-5c0b-488d-a72a-0a2d14ecb0ec@googlegroups.com>
yes, see https://github.com/dlicata335/hott-agda/blob/master/misc/Andrew.agda
Using the notation
e1 : (x : Bool) → (f True x) == (g True x)
e1 True = id
e1 False = id
eh : (x : _) → f x == g x
eh True = λ≃ e1
eh False = λ≃ (λ { True → id ; False → id })
e : f == g
e = λ≃ eh
the steps are
goal : fst Y == True
goal = fst Y ≃〈 ... 〉
transport (λ u → Phi (u True True)) e (fst X) ≃〈 ... 〉
transport Phi (ap (λ u → u True True) e) (fst X) ≃〈 ... 〉
transport Phi (ap (λ f₁ → f₁ True) (ap (λ f₁ → f₁ True) e)) (fst X) ≃〈 ... 〉
transport Phi (ap (λ f₁ → f₁ True) (λ≃ e1)) (fst X) ≃〈 ... 〉
True ∎
I would be very surprised if there was something like this that was not provable in "book HoTT”.
-Dan
> On Sep 5, 2016, at 12:54 PM, Andrew Polonsky <andrew....@gmail.com> wrote:
>
> Hi,
>
> There is a common understanding that the "right" concept of equality in Martin-Lof type theory is not the intensional identity type, but is a different notion of equality, which is extensional. The adjunction of the univalence axiom to standard MLTT makes the identity type behave like this "intended" equality --- but it breaks the computational edifice of type theory.
>
> More precisely, this "Hott book" approach only nails down the concept of equality with respect to its *logical properties* --- the things you can prove about it. But not its actual computational behavior.
>
> Since computational behavior can often be "seen" on the logical level, I am trying to get a better understanding of the sense in which this "Hott book" equality type is really (in)complete. How would a "truly computational" equality type be different from it? (Other than satisfying canonicity, etc.)
>
> One precise example is that, for the "right" notion of equality, the equivalences characterizing path types of standard type constructors proved in Chapter 2 of the book could perhaps hold definitionally. (The theorems proved there are thus "seeing" these equalities on the logical level.)
>
> I tried to look for a more interesting example, and came up with the following puzzle.
>
> f, g : Bool -> Bool -> Bool
> f x y = if x then y else ff
> g x y = if y then x else ff
>
> e : f = g
> e = FE(...) [using UA to get Function Extensionality]
>
> Phi : Bool -> Type
> Phi tt = Bool
> Phi ff = Unit
>
> Psi : (Bool->Bool->Bool)->Type
> Psi = \u. (u tt tt) x (u tt ff) x (u ff tt) x (u ff ff)
>
> X : Psi f
> X = (tt,*,*,*)
>
> Y : Psi g
> Y = subst Psi e X
>
> QUESTION.
> Can we prove, in "book Hott", that "proj1 Y = tt" is inhabited?
>
> Cheers!
> Andrew
>
> --
> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeThe...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
next prev parent reply other threads:[~2016-09-05 21:51 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-05 16:54 Andrew Polonsky
2016-09-05 21:40 ` [HoTT] " Michael Shulman
2016-09-05 21:51 ` Dan Licata [this message]
2016-09-06 7:30 ` Andrew Polonsky
2016-09-06 12:32 ` Michael Shulman
2016-09-06 12:56 ` Dan Licata
2016-09-06 12:57 ` Peter LeFanu Lumsdaine
2016-09-06 13:44 ` Andrew Polonsky
2016-09-06 22:14 ` Martin Escardo
2016-09-07 23:18 ` Matt Oliveri
2016-09-08 4:14 ` Michael Shulman
2016-09-08 6:06 ` Jason Gross
2016-09-08 9:11 ` Martin Escardo
2016-09-08 6:34 ` Matt Oliveri
2016-09-08 6:45 ` Michael Shulman
2016-09-08 9:07 ` Martin Escardo
2016-09-08 9:51 ` Thomas Streicher
2016-09-19 12:40 ` Robin Adams
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=9064B371-C3B0-45F5-8720-90E6D10E211C@cs.cmu.edu \
--to="d..."@cs.cmu.edu \
--cc="HomotopyT..."@googlegroups.com \
--cc="andrew...."@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).