Discussion of Homotopy Type Theory and Univalent Foundations
 help / Atom feed
From: Michael Shulman <shulman@sandiego.edu>
To: "Licata, Dan" <dlicata@wesleyan.edu>
Cc: "Anders Mörtberg" <andersmortberg@gmail.com>,
	"Homotopy Type Theory" <HomotopyTypeTheory@googlegroups.com>
Subject: Re: [HoTT] Different definitions of Sn
Date: Wed, 18 Sep 2019 12:19:16 -0700
Message-ID: <CAOvivQy+UGKt3q8P8ed9N6E+jOUcjgQ49pxp0rBuDM=SoV-QbQ@mail.gmail.com> (raw)
In-Reply-To: <DBF368F2-295C-4175-93F2-1EA01D49B95D@wesleyan.edu>

What I find more interesting is that it's even *possible* to
parametrize a definition of S^n with one n-loop by an internal n.  It
would be intriguing to try to find a schema or semantics for HITs that
allow paths of arbitrary internally-specified dimensions, and see
whether there are other examples that are more useful.  The only other
vaguely similar example that I know of is mentioned at the end of
https://homotopytypetheory.org/2014/11/02/universal-properties-without-function-extensionality/
(and implemented in the HoTT/Coq library), a definition of
localization that works in the absence of funext.

On Wed, Sep 18, 2019 at 9:27 AM Licata, Dan <dlicata@wesleyan.edu> wrote:
>
> Thanks Anders.  To be clear, I don’t think we have any schemas or semantics of HITs that accept this definition of S^n (with an n-loop for an internally-specified n) in any non-cubical settings either; at the time, I was thinking of it more as exploring what you can do with the definition.  Proving the two definitions equivalent should be the same (modulo which definitional equalities you get) as implementing the one-n-loop rules in terms of suspensions, so that would be one way to justify these rules.
>
> Also, the proof of pi_n(S^n) = Z that Guillaume mentioned, which predated Freudenthal, is a lot more work than the one that you get from Freudenthal, so I’m not sure we have any motivating examples for why the one-n-loop definition is better than the suspension definition for arbitrary n.
>
> -Dan
>
> > On Sep 18, 2019, at 8:00 AM, Anders Mörtberg <andersmortberg@gmail.com> wrote:
> >
> > Let me elaborate a bit on my answer. One might naively try to copy Dan and Guillaume's definition and write the following in Cubical Agda:
> >
> >
> > Omega : (A : Type₀) → A → Type₀
> > Omega A a = (a ≡ a)
> >
> > itOmega : ℕ → (A : Type₀) → A → Type₀
> > itOmega zero A a = Omega A a
> > itOmega (suc n) A a = itOmega n (Omega A a) refl
> >
> > data Sn (n : ℕ) : Type₀ where
> >   base : Sn n
> >   surf : itOmega n (Sn n) base
> >
> >
> > However Agda will complain as surf is not constructing an element of Sn. This might seem a bit funny as Cubical Agda is perfectly happy with
> >
> >
> > data S³ : Type₀ where
> >   base : S³
> >   surf : Path (Path (base ≡ base) refl refl) refl refl
> >
> >
> > But what is happening under the hood is that surf is a constructor taking i, j, and k in the interval and returning an element of S^3 with boundary "base" whenever i, j and k are 0 or 1. In cubicaltt we can write this HIT in the following way:
> >
> >
> > data S3 = base
> >  | surf <i j k> [ (i=0) -> base
> >                 , (i=1) -> base
> >                 , (j=0) -> base
> >                 , (j=1) -> base
> >                 , (k=0) -> base
> >                 , (k=1) -> base ]
> >
> >
> > The problem is then clearer: in order to write the surf constructor of Sn we would have to quantify over n interval variables and specify the boundary of the n-cell. This is what that is not supported by any of the cubical schemas for HITs.
> >
> > --
> > Anders
> >
> > On Wednesday, September 18, 2019 at 11:00:22 AM UTC+2, Guillaume Brunerie wrote:
> > Hi,
> >
> > We give a definition of S^n with one point and one n-loop by
> > introduction/elimination/reduction rules in our paper with Dan Licata
> > (https://guillaumebrunerie.github.io/pdf/lb13cpp.pdf), which can be
> > implemented in Agda (so Kristina’s question can be formulated and can
> > presumably be formalized in Agda) but I don’t think we actually proved
> > it.
> >
> > Best,
> > Guillaume
> >
> > Den ons 18 sep. 2019 kl 10:32 skrev Anders Mortberg <andersm...@gmail.com>:
> > >
> > > Hi Kristina,
> > >
> > > We have proofs for S^0, S^1, S^2 and S^3 in Cubical Agda:
> > > https://github.com/agda/cubical/blob/master/Cubical/HITs/Susp/Base.agda
> > >
> > > However, I don't think we can even write down the general version of
> > > S^n as a type with a point and an n-loop with the schema implemented
> > > in Cubical Agda. As far as I know no other schema for HITs support
> > > this kind of types either.
> > >
> > > --
> > > Anders
> > >
> > > On Tue, Sep 17, 2019 at 9:56 PM Kristina Sojakova
> > > <sojakova...@gmail.com> wrote:
> > > >
> > > > Hello everybody,
> > > >
> > > > Is it worked out somewhere that the two definitions of Sn as a
> > > > suspension on one hand and a HIT with a point and an n-loop on the other
> > > > hand are equivalent? This is also an exercise in the book. I know
> > > > Favonia has some Agda code on spheres but I couldn't find this result in
> > > > there.
> > > >
> > > > Thanks,
> > > >
> > > > Kristina
> > > >
> > > >
> >
> > --
> > 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 HomotopyTypeTheory+unsubscribe@googlegroups.com.
> > To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/db712ad2-396f-4328-bb73-898dcb956834%40googlegroups.com.
>
> --
> 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 HomotopyTypeTheory+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/DBF368F2-295C-4175-93F2-1EA01D49B95D%40wesleyan.edu.

-- 
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 HomotopyTypeTheory+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAOvivQy%2BUGKt3q8P8ed9N6E%2BjOUcjgQ49pxp0rBuDM%3DSoV-QbQ%40mail.gmail.com.

      reply index

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-17 19:56 Kristina Sojakova
2019-09-18  8:32 ` Anders Mortberg
2019-09-18  9:00   ` Guillaume Brunerie
2019-09-18 12:00     ` Anders Mörtberg
2019-09-18 16:27       ` Licata, Dan
2019-09-18 19:19         ` Michael Shulman [this message]

Reply instructions:

You may reply publically 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='CAOvivQy+UGKt3q8P8ed9N6E+jOUcjgQ49pxp0rBuDM=SoV-QbQ@mail.gmail.com' \
    --to=shulman@sandiego.edu \
    --cc=HomotopyTypeTheory@googlegroups.com \
    --cc=andersmortberg@gmail.com \
    --cc=dlicata@wesleyan.edu \
    /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

Discussion of Homotopy Type Theory and Univalent Foundations

Archives are clonable: git clone --mirror http://inbox.vuxu.org/hott

Newsgroup available over NNTP:
	nntp://inbox.vuxu.org/vuxu.archive.hott


AGPL code for this site: git clone https://public-inbox.org/ public-inbox