```Discussion of Homotopy Type Theory and Univalent Foundations
help / color / mirror / Atom feed```
```From: Michael Shulman <shulman@sandiego.edu>
To: Jason Gross <jasongross9@gmail.com>
Cc: Nicolai Kraus <nicolai.kraus@gmail.com>,
Subject: Re: [HoTT] two's complement integers
Date: Thu, 4 Mar 2021 20:41:11 -0800
Message-ID: <CAOvivQxy2EJNScHu9Y2U=GLiHDrDxDq-Kg8Dz9TrFk3ZepiW8Q@mail.gmail.com> (raw)

I don't have a particular application in mind at the moment.  It just
struck me how 2-adic integers have such simple coinductive definitions
of the arithmetic operations, without any case distinctions at all,
and I wondered whether there was a definition of plain integers with a
similar property.  It seems annoying to have to constantly case-split
on a sign bit.

I guess a lower-inductive representation could take advantage of the
fact that the 2-adic representation of an integer other than 0 or -1
has a unique largest digit that differs from all larger digits, i.e.
it's either

...0000 (i.e. 0)
...1111 (i.e. -1)
...0001 + arbitrary bit string
...1110 +  arbitrary bit string

This involves more case splits when defining arithmetic, but it would
probably be an easy set to show that the HIT one is equivalent to.

On Thu, Mar 4, 2021 at 7:02 PM Jason Gross <jasongross9@gmail.com> wrote:
>
> Note that the Coq standard library Z is a binary representation of integers and Z.testbit gives access to the infinite twos-compliment representation.  Of course, it makes use of case-distinctions on sign to define it, but you go bit-by-bit; if the number is negative, you just invert the bit before returning it.  Is there something you're after by having the representation not encode sign bits separately?
>
> On Thu, Mar 4, 2021, 21:27 Michael Shulman <shulman@sandiego.edu> wrote:
>>
>> On Thu, Mar 4, 2021 at 3:16 PM Nicolai Kraus <nicolai.kraus@gmail.com> wrote:
>> > I'm not sure what the precise thing is that you're looking for because, without further specification, any standard definition of Z would qualify :-)
>>
>> Yes, that seems to be what Martin suggested too with ℕ + ℕ.  It seemed
>> to me as though the distance between ℕ + ℕ and my ℤ is greater than
>> the distance between his 𝔹 and 𝔹', but maybe not in any important
>> way.
>>
>> > The HIT is neat, but wouldn't it in practice behave pretty similar to a standard representation via binary lists? E.g. something like Unit + Bool * List(Bool), where inl(*) is zero, the first Bool is the sign, and you add a 1 in front of the list in order to get a positive integer. What's the advantage of the HIT - maybe one can avoid case distinctions?
>>
>> Is there a non-HIT binary representation that can be interpreted as
>> two's-complement (thereby avoiding case distinctions on sign)?  I
>> haven't been able to figure out a way to do that with mere lists of
>> booleans.
>>
>> --
>> 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/CAOvivQwtAMVnOG7D_A_s1EMAum4fv_x945MLB%2B01Y_pMdEKC2w%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 HomotopyTypeTheory+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAKObCaqiyUcB%2BFrbvPmR-eS5ZMsb2CTVCtVbAYZeMUcOzCpmbA%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 HomotopyTypeTheory+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/HomotopyTypeTheory/CAOvivQxy2EJNScHu9Y2U%3DGLiHDrDxDq-Kg8Dz9TrFk3ZepiW8Q%40mail.gmail.com.
```

```     prev parent reply	other threads:[~2021-03-05  4:41 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-04 20:43 Michael Shulman
2021-03-04 21:11 ` Martin Escardo
2021-03-04 22:05   ` Michael Shulman
2021-03-04 22:42     ` Martin Escardo
2021-03-04 23:16     ` Nicolai Kraus
2021-03-05  2:27       ` Michael Shulman
2021-03-05  3:02         ` Jason Gross
2021-03-05  4:41           ` Michael Shulman [this message]
```

```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,

Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

switches of git-send-email(1):

git send-email \
--to=shulman@sandiego.edu \
--cc=jasongross9@gmail.com \
--cc=nicolai.kraus@gmail.com \

https://kernel.org/pub/software/scm/git/docs/git-send-email.html

```

```Discussion of Homotopy Type Theory and Univalent Foundations

This inbox may be cloned and mirrored by anyone:

git clone --mirror https://inbox.vuxu.org/hott

# If you have public-inbox 1.1+ installed, you may
# initialize and index your mirror using the following commands:
public-inbox-init -V1 hott hott/ https://inbox.vuxu.org/hott \