From: Michael Shulman <email@example.com> To: David Roberts <firstname.lastname@example.org> Cc: Bas Spitters <email@example.com>, homotopytypetheory <firstname.lastname@example.org>, Nicolas Alexander Schmidt <email@example.com> Subject: Re: [HoTT] Why did Voevodsky find existing proof assistants to be 'impractical'? Date: Sun, 3 Nov 2019 11:13:08 -0800 Message-ID: <CAOvivQz47kSm9WbKDmUsndrpAJMkNwiWmVABqOFrVqyTOvSAbw@mail.gmail.com> (raw) In-Reply-To: <CAFL+ZM_==iLS16Vy7sGiEqNkCxOMYL4j+ZFqKv5uJ-ivkuemEg@mail.gmail.com> But does univalence a la Book HoTT *actually* make it easier to reason about such things? It allows us to write "=" rather than "\cong", but to construct such an equality we have to construct an isomorphism first, and to *use* such an equality we have to transport along it, and then we get lots of univalence-redexes that we have to manually reduce away. My experience formalizing math in HoTT/Coq is that it's much easier if you *avoid* turning equivalences into equalities except when absolutely necessary. (I don't have any experience formalizing math in a cubical proof assistant, but in that case at least you wouldn't have to manually reduce the univalence-redexes -- although it seems to me you'd still have to construct the isomorphism before you can apply univalence to make it an equality.) On Sun, Nov 3, 2019 at 3:57 AM David Roberts <firstname.lastname@example.org> wrote: > > Forget even higher category theory. Kevin Buzzard now goes around telling the story of how even formally proving (using Lean) things in rather elementary commutative algebra from EGA that are stated as equalities was not obvious: the equality is really an isomorphism arising from a universal property. Forget trying to do anything motivic, when algebra is full of such equalities. This is not a problem with univalence, of course. > > David > > On Sun, 3 Nov 2019, 10:08 PM Bas Spitters <email@example.com> wrote: >> >> There's also VV homotopy lambda calculus, which he later abandoned for MLTT: >> https://www.math.ias.edu/~vladimir/Site3/Univalent_Foundations_files/Hlambda_short_current.pdf >> >> On Sun, Oct 27, 2019 at 6:22 PM Bas Spitters <firstname.lastname@example.org> wrote: >>> >>> I believe it refers to his 2-theories: >>> https://www.ias.edu/ideas/2014/voevodsky-origins >>> >>> On Sun, Oct 27, 2019 at 3:41 PM Nicolas Alexander Schmidt <email@example.com> wrote: >>>> >>>> In [this](https://www.youtube.com/watch?v=zw6NcwME7yI&t=1680) 2014 talk >>>> at IAS, Voevodsky talks about the history of his project of "univalent >>>> mathematics" and his motivation for starting it. Namely, he mentions >>>> that he found existing proof assistants at that time (in 2000) to be >>>> impractical for the kinds of mathematics he was interested in. >>>> >>>> Unfortunately, he doesn't go into details of what mathematics he was >>>> exactly interested in (I'm guessing something to do with homotopy >>>> theory) or why exactly existing proof assistants weren't practical for >>>> formalizing them. Judging by the things he mentions in his talk, it >>>> seems that (roughly) his rejection of those proof assistants was based >>>> on the view that predicate logic + ZFC is not expressive enough. In >>>> other words, there is too much lossy encoding needed in order to >>>> translate from the platonic world of mathematical ideas to this formal >>>> language. >>>> >>>> Comparing the situation to computer programming languages, one might say >>>> that predicate logic is like Assembly in that even though everything can >>>> be encoded in that language, it is not expressive enough to directly >>>> talk about higher level concepts, diminishing its practical value for >>>> reasoning about mathematics. In particular, those systems are not >>>> adequate for *interactive* development of *new* mathematics (as opposed >>>> to formalization of existing theories). >>>> >>>> Perhaps I am just misinterpreting what Voevodsky said. In this case, I >>>> hope someone can correct me. However even if this wasn't *his* view, to >>>> me it seems to be a view held implicitly in the HoTT community. In any >>>> case, it's a view that one might reasonably hold. >>>> >>>> However I wonder how reasonable that view actually is, i.e. whether e.g. >>>> Mizar really is that much more impractical than HoTT-Coq or Agda, given >>>> that people have been happily formalizing mathematics in it for 46 years >>>> now. And, even though by browsing the contents of "Formalized >>>> Mathematics" one can get the impression that the work consists mostly of >>>> formalizing early 20th century mathematics, neither the UniMath nor the >>>> HoTT library for example contain a proof of Fubini's theorem. >>>> >>>> So, to put this into one concrete question, how (if at all) is HoTT-Coq >>>> more practical than Mizar for the purpose of formalizing mathematics, >>>> outside the specific realm of synthetic homotopy theory? >>>> >>>> >>>> -- >>>> >>>> Nicolas >>>> >>>> >>>> -- >>>> 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 HomotopyTypeTheoryfirstname.lastname@example.org. >>>> To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/e491d38b-b50a-6172-dca9-40d45fe1b6d2%40fromzerotoinfinity.xyz. >> >> -- >> 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 HomotopyTypeTheoryemail@example.com. >> To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAOoPQuRQPMkCFKYtAbB%2BpNK90XtFk%2BaVT_aY59U_-9t17sBzeA%40mail.gmail.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 HomotopyTypeTheoryfirstname.lastname@example.org. > To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAFL%2BZM_%3D%3DiLS16Vy7sGiEqNkCxOMYL4j%2BZFqKv5uJ-ivkuemEg%40mail.gmail.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 HomotopyTypeTheoryemail@example.com. To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAOvivQz47kSm9WbKDmUsndrpAJMkNwiWmVABqOFrVqyTOvSAbw%40mail.gmail.com.
next prev parent reply index Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top 2019-10-27 14:41 Nicolas Alexander Schmidt 2019-10-27 17:22 ` Bas Spitters 2019-11-03 11:38 ` Bas Spitters 2019-11-03 11:52 ` David Roberts 2019-11-03 19:13 ` Michael Shulman [this message] 2019-11-03 19:45 ` Valery Isaev 2019-11-03 22:23 ` Martín Hötzel Escardó 2019-11-04 23:20 ` Nicolas Alexander Schmidt 2019-11-04 18:42 ` Kevin Buzzard 2019-11-04 21:10 ` Michael Shulman 2019-11-04 23:26 ` David Roberts 2019-11-05 15:43 ` Daniel R. Grayson 2019-11-05 20:29 ` Yuhao Huang 2019-11-06 23:59 ` Daniel R. Grayson 2019-11-05 23:14 ` Martín Hötzel Escardó 2019-11-06 0:06 ` Stefan Monnier 2019-11-11 18:26 ` Licata, Dan 2019-11-03 7:29 ` Michael Shulman
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=CAOvivQz47kSm9WbKDmUsndrpAJMkNwiWmVABqOFrVqyTOvSAbw@mail.gmail.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ /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 Example config snippet for mirrors Newsgroup available over NNTP: nntp://inbox.vuxu.org/vuxu.archive.hott AGPL code for this site: git clone https://public-inbox.org/public-inbox.git