(Beginner's question) Uses of HITs beyond homotopy theory

1 view
Skip to first unread message

Timothy Carstens

Aug 8, 2019, 2:33:06 PM8/8/19
to Homotopy Type Theory
Sorry for the broad & naive question. I'm a geometer by training but have been working in compsci for most of my career (with lots of time spent in Coq verifying programs).

I've got a naive question that I hope isn't too inappropriate for this list: can anyone suggest some papers that show applications of HITs? I'm embarrassed to admit it, but I don't know any applications outside of synthetic homotopy theory and higher categories.

Perhaps categorical semantics? But even still I'm not personally aware of any applied results from that domain (contrast with operational semantics; but I am extremely ignorant, so please correct me!)

All my best and apologies in advance if this is off-topic for this list,

Niels van der Weide

Aug 8, 2019, 2:52:48 PM8/8/19
to Homotopy Type Theory
Some established applications of HITs outside of synthetic homotopy theory:
- Partiality monad, see "Partiality Revisited" by Altenkirch, Danielsson, and Kraus
- Type Theory in Type Theory, see 'Type Theory in Type Theory with Quotient Inductive Inductive Types' and 'Normalization by evaluation for dependent types' by Altenkirch and Kaposi
- Free algebraic structures, see HoTT book 6.11 and the paper "Finite Sets in Homotopy Type Theory" by Frumin, Geuvers, Gondelman, Van der Weide
- Patch theory, see 'Homotopical Patch Theory' by Angiuli, Morehouse, Licata, and Harper

Steve Awodey

Aug 8, 2019, 4:18:13 PM8/8/19
to Timothy Carstens, Homotopy Type Theory
quotients by equivalence relations.
see HoTT Book 6.10

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.
To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAJGt_zG%2B04Rfbs_py%3DPYkubbwzeYb0TRhhfek-RT663uVUo%3D-A%40mail.gmail.com.

Michael Shulman

Aug 8, 2019, 5:49:56 PM8/8/19
to Timothy Carstens, Homotopy Type Theory
More generally, all colimits other than coproducts are HITs (of the
"non-recursive" variety). This includes both homotopy colimits and
ordinary colimits of sets (obtained by 0-truncating homotopy
colimits). Having colimits of sets is fairly essential for nearly all
ordinary set-based mathematics, even for people who don't care about
homotopy theory or higher category theory in the slightest. There
aren't really papers specifically about this, because it's so vast,
and because there's not much to say other than the observation that
colimits exist, since at that point you can just appeal to the
long-known fact that once the category of sets satisfies certain basic
properties (Lawvere's "Elementary Theory of the Category of Sets") it
suffices as a basis on which to develop a large amount of mathematics.
The verification of these axioms in HoTT with HITs can be found in
section 10.1 of the HoTT Book. (Before HITs, people formalizing
set-based mathematics in type theory used "setoids" to mimic quotients
and other colimits.)

Beyond this, in set-based mathematics HITs are used to construct free
algebraic structures, as Niels said. Some free algebraic structures
(free monoids, free groups, free rings, etc.) can be constructed based
only on the axioms of ETCS, but for fancier (and in particular,
infinitary) algebraic structures one needs more. In fact there are
algebraic theories for which free algebraic structures cannot be
constructed in ZF (at least, under a large cardinal assumption): the
idea is to use a theory to encode the existence of large regular
cardinals, which cannot be constructed in ZF (see Blass's paper
"Words, free algebras, and coequalizers"). But HITs suffice to
construct even free infinitary algebras of this sort; see e.g. section
9 of my paper with Peter Lumsdaine, "Semantics of higher inductive
types". Thus, HITs can be useful for doing (universal) algebra
constructively, where here "constructively" can even mean "with
classical logic but without the axiom of choice".
> To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/728FA1EA-014C-4242-8B34-33A17D7B9208%40gmail.com.

Timothy Carstens

Aug 8, 2019, 6:09:34 PM8/8/19
to Homotopy Type Theory
Thank you for the excellent replies! It looks like I was struggling with a lack of imagination while the answer was staring me right in the face.

Thorsten Altenkirch

Aug 9, 2019, 6:36:52 AM8/9/19
to Timothy Carstens, Homotopy Type Theory

I think it is an excellent question. However, looking at the examples it may seem that we only need QITs, that is set-truncated HITs. However, this is not true when you are dealing with higher structures that arise naturally like the type of sets. For example when you define the integers as a quotient, or nicer as a QIT, you can only eliminate into types that are sets, for example you cannot define a function from the Integers into Set. However, this can be addressed by replacing set-truncation with a coherence law, in this case you basically say that integers have 0 and suc and suc is an equivalence. You can prove that the HIT constructed by these principles is a Set (this is actually harder than it seems) – Luis Soccola and I have recently written a paper about  this (need to put it on arxiv).  Another example is the intrinsic presentation of type theory as the initial Category with Families which was already mentioned. Again the problem is that you need to set-truncate but then you cannot even define the set-interpretation. This can be again addressed by adding some coherence laws (need to check the details) and you get a coherent version of CWFs which enable us to eliminate into any 1-type, including the universe of sets.



This message and any attachment are intended solely for the addressee
and may contain confidential information. If you have received this
message in error, please contact the sender and delete the email and

Any views or opinions expressed by the author of this email do not
necessarily reflect the views of the University of Nottingham. Email
communications with the University of Nottingham may be monitored 
where permitted by law.

Reply all
Reply to author
0 new messages