Looking for a review manager for Real

newer
[JSON][review] Should there be an...

older
Do I need to use boost::mutex in...

Damian Vicino

2 Sep 2020 2 Sep '20

1:40 p.m.

Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it. Best regards, Damian

Show replies by date

Vinnie Falco

2 Sep 2 Sep

3:11 p.m.

On Wed, Sep 2, 2020 at 6:40 AM Damian Vicino via Boost wrote:

...

GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it.

Can you please provide a link to the repository? Thanks

Damian Vicino

3:44 p.m.

https://github.com/BoostGSoC20/Real

...

On Wed, Sep 2, 2020 at 6:40 AM Damian Vicino via Boost wrote:

...
GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it.

Can you please provide a link to the repository?

Thanks

https://github.com/BoostGSoC20/Real Thanks

Richard Hodges

5:06 p.m.

I'd be happy to volunteer for the role of Review Manager if there is no-one more suitable. *Domain Experience:* 7 Years working in investment banks writing pricers and risk systems, mainly for FX and Options. Minimising errors resulting from a sequence of floating point numerical operations was an important feature of that work, so I have some exposure to the domain. *Software Experience:* I have been writing software commercially in Assembler, C and C++ since I was 16 years old. So that would be 35 years. *Boost Review Manager Experience:* None. I would be happy with any advice that can be offered. *Qualifications within Boost:* I maintain the Boost.Beast library On Wed, 2 Sep 2020 at 15:40, Damian Vicino via Boost wrote:

...

Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it. Best regards, Damian

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

-- Richard Hodges hodges.r@gmail.com office: +442032898513 home: +376841522 mobile: +376380212

Mateusz Loskot

6:38 p.m.

On Wed, 2 Sep 2020 at 19:07, Richard Hodges via Boost wrote:

...

I'd be happy to volunteer for the role of Review Manager if there is no-one more suitable.

Richard, Thank you for volunteering. Damian, If you agree with Richard's offer, then please agree with Richard on possible dates of 10 day long period, and send the request to John Phillips and myself (e-mails on https://www.boost.org/community/reviews.html#Wizard) Best regards, -- Mateusz Loskot, http://mateusz.loskot.net

Mateusz Loskot

6:35 p.m.

On Wed, 2 Sep 2020 at 15:41, Damian Vicino via Boost wrote:

...

Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it.

Damian, If you are not an author of the library, but mentor only, you can manage the review. Best regards, -- Mateusz Loskot, http://mateusz.loskot.net

Damian Vicino

10:56 p.m.

El mié., 2 sept. 2020 a las 14:36, Mateusz Loskot via Boost (< boost@lists.boost.org>) escribió:

...

On Wed, 2 Sep 2020 at 15:41, Damian Vicino via Boost wrote:

...
Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real. We are looking for a volunteer to be a review manager for it.

Damian,

If you are not an author of the library, but mentor only, you can manage the review.

Best regards, -- Mateusz Loskot, http://mateusz.loskot.net

Thanks, but I think I'm too close to it even though I didn't write the code itself. I will prefer someone else to take the review management for this one.

sehe

4 Sep 4 Sep

7:05 p.m.

On wo, 02. sep 09:40, Damian Vicino via Boost wrote:

...

Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real.

Glanced over the library, and was wondering how much overlap there is with: - Boost Interval Arithmetic Library (https://www.boost.org/doc/libs/1_74_0/libs/numeric/interval/doc/interval.htm) which AFAIK is a part of Boost Numeric - Boost Interval Container Library If there are major differences in objectives/scope/approach it would be very helpful to existing Boost users if the documentation calls those out explicitly. Regards Seth

Robert Ramey

9:32 p.m.

On 9/4/20 12:05 PM, sehe via Boost wrote:

...

On wo, 02. sep 09:40, Damian Vicino via Boost wrote:

...
Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real.

Glanced over the library, and was wondering how much overlap there is with:

- Boost Interval Arithmetic Library (https://www.boost.org/doc/libs/1_74_0/libs/numeric/interval/doc/interval.htm) which AFAIK is a part of Boost Numeric - Boost Interval Container Library

Not to mention safe numerics library

...

If there are major differences in objectives/scope/approach it would be very helpful to existing Boost users if the documentation calls those out explicitly.

Regards Seth

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Damian Vicino

5 Sep 5 Sep

12:02 a.m.

El vie., 4 sept. 2020 a las 17:33, Robert Ramey via Boost (< boost@lists.boost.org>) escribió:

...

On 9/4/20 12:05 PM, sehe via Boost wrote:

...
On wo, 02. sep 09:40, Damian Vicino via Boost wrote:

...
Hi, GSOC finished and we are almost-ready for proposing to review a new library, Real.

Glanced over the library, and was wondering how much overlap there is with:

- Boost Interval Arithmetic Library ( https://www.boost.org/doc/libs/1_74_0/libs/numeric/interval/doc/interval.htm) which AFAIK is a part of Boost Numeric - Boost Interval Container Library

Not to mention safe numerics library

...
If there are major differences in objectives/scope/approach it would be very helpful to existing Boost users if the documentation calls those out explicitly.

Regards Seth

There is actually not much overlap with other libraries, it may in the implementation details be some overlap with boost::interval arithmetic, but not in the goals or features exposed. The goal in Real is to evaluate expressions that the required precision to evaluate them accurately is unknown in advance and explores how to achieve perfect accuracy for the evaluation, or fails noisily. The general idea of how to evaluate is defining numbers as functions that produce intervals around them and when iterated produce more constrained intervals around the same number, all numbers in the expression are adjusted until the intervals are small enough to determine an evaluation result. For example, let say we have the (3e < 2pi) expression. Lets evaluate with "1 digit". e is [2, 3], pi is [3, 4], then 3e < 2pi is [6, 9] < [6, 8], since the intervals overlap there is no way to know which side of the expression is the lower one. We iterate and use e= [2.7, 2.8] and pi [3.1, 3.2] then the expression is now [8.1, 8.4] < [6.2, 6.4] Then there is no overlap and no need to continue iterating, the answer is False. We are going to update the doc to make it clear that there is no relation to the other libraries.

Kostas Savvidis

4:03 p.m.

...

On Sep 5, 2020, at 03:02, Damian Vicino via Boost wrote:

...
There is actually not much overlap with other libraries, it may in the implementation details be some overlap with boost::interval arithmetic, but not in the goals or features exposed.

The goal in Real is to evaluate expressions that the required precision to evaluate them accurately is unknown in advance and explores how to achieve perfect accuracy for the evaluation, or fails noisily.

I am planning to do a full review, and certainly will try to see whether it can evaluate some hard numerical expressions. Unfortunately, it is clear already that the documentation is severely lacking. My suggestion right now is to delay the start of the review until we have proper documentation. Range arithmetic is useful mainly for two purposes: 1) numeric analysis of algorithms - to determine the accuracy of the final result given a finite accuracy of the input quantities. This sounds close enough to what you describe as your goal, so saying that your library does something different is strange. 2) in pure math - to derive strict bounds on the value of some trig and special functions. Your library cannot (yet?), but other range libraries can do it, notably the Arb library http://arblib.org. On 9/4/20 12:05 PM, sehe via Boost wrote:

...

Glanced over the library, and was wondering how much overlap there is with: - Boost Interval Arithmetic Library (https://www.boost.org/doc/libs/1_74_0/libs/numeric/interval/doc/interval.htm https://www.boost.org/doc/libs/1_74_0/libs/numeric/interval/doc/interval.htm) which AFAIK is a part of Boost Numeric - Boost Interval Container Library

The Interval Container Library has no relation with this business at all, it is about treating time slots for accounting software. Safe Numerics also has nothing to do with this, so any comparisons should be with Boost.Interval and Arb (which is in C). Best Regards, Kostas Savvidis ========================================= Institute of Nuclear and Particle Physics NCSR Demokritos http://inspirehep.net/author/profile/K.G.Savvidy.1 https://github.com/kotika/random https://mixmax.hepforge.org

pbristow＠hetp.u-net.com

4:38 p.m.

...

-----Original Message----- From: Boost On Behalf Of Kostas Savvidis via Boost Sent: 5 September 2020 17:04 To: boost@lists.boost.org Cc: Kostas Savvidis ; Robert Ramey Subject: Re: [boost] Looking for a review manager for Real

...
On Sep 5, 2020, at 03:02, Damian Vicino via Boost wrote:

...
There is actually not much overlap with other libraries, it may in the implementation details be some overlap with boost::interval arithmetic, but not in the goals or features exposed.

The goal in Real is to evaluate expressions that the required precision to evaluate them accurately is unknown in advance and explores how to achieve perfect accuracy for the evaluation, or fails noisily. <snip>

...

so any comparisons should be with Boost.Interval and Arb (which is in C).

And is GPL licenced, not Boost :-( Paul

Damian Vicino

9 Sep 9 Sep

2:27 a.m.

El sáb., 5 sept. 2020 a las 12:45, Paul A Bristow via Boost (< boost@lists.boost.org>) escribió:

...

...
-----Original Message----- From: Boost On Behalf Of Kostas Savvidis via Boost Sent: 5 September 2020 17:04 To: boost@lists.boost.org Cc: Kostas Savvidis ; Robert Ramey Subject: Re: [boost] Looking for a review manager for Real

...
On Sep 5, 2020, at 03:02, Damian Vicino via Boost < boost@lists.boost.org> wrote:

...
There is actually not much overlap with other libraries, it may in the implementation details be some overlap with boost::interval arithmetic, but not in the goals or features exposed.

The goal in Real is to evaluate expressions that the required precision to evaluate them accurately is unknown in advance and explores how to achieve perfect accuracy for the evaluation, or fails noisily. <snip>

...
so any comparisons should be with Boost.Interval and Arb (which is in C).

And is GPL licenced, not Boost :-(

Paul

Thanks for the comments.

I didn't have experience with the ARB library, but it seems very related. I think it will be super useful to use for comparing benchmarks and validate testing (same expression in both should have the same result kind of testing). I think the next step on our side would be to dig into ARB, automatize and compare them, and improve documentation. Sounds that we have work on that area for a while, I will post about review again after we have some advances there. BTW, great reference to the ARB, the libraries we knew about the topic are long dead, this one seems pretty much alive and it's very encouraging to see someone else working in the same area. About documentation. I think it would be really useful for us to have someone not involved with the project to point us what are the major areas of improvement we can target. Thanks! Damian

Vinnie Falco

11:16 a.m.

On Tue, Sep 8, 2020 at 7:28 PM Damian Vicino via Boost wrote:

...

...

I am greatly looking forward to seeing this library added to Boost, because then I can say... ...Boost just got Real!!!

Richard Hodges

10 Sep 10 Sep

7:18 a.m.

On Wed, 9 Sep 2020 at 13:16, Vinnie Falco via Boost wrote:

...

On Tue, Sep 8, 2020 at 7:28 PM Damian Vicino via Boost

wrote:

...
...

I am greatly looking forward to seeing this library added to Boost,

because then I can say...

...Boost just got Real!!!

Oh God, make it stop.

...

_______________________________________________

Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

--

Richard Hodges hodges.r@gmail.com office: +442032898513 home: +376841522 mobile: +376380212

Hans Dembinski

9 Sep 9 Sep

11:44 a.m.

...

On 5. Sep 2020, at 18:03, Kostas Savvidis via Boost wrote:

...
On Sep 5, 2020, at 03:02, Damian Vicino via Boost wrote:

...
There is actually not much overlap with other libraries, it may in the implementation details be some overlap with boost::interval arithmetic, but not in the goals or features exposed.

The goal in Real is to evaluate expressions that the required precision to evaluate them accurately is unknown in advance and explores how to achieve perfect accuracy for the evaluation, or fails noisily.

I am planning to do a full review, and certainly will try to see whether it can evaluate some hard numerical expressions. Unfortunately, it is clear already that the documentation is severely lacking. My suggestion right now is to delay the start of the review until we have proper documentation.

I second that. I am interested in reviewing this, but the docs and the medium article didn't explain very well what kind of problem this library solves. A simple educational example early on in the introduction would be helpful. The scope/target audience is not clear (as Robert already mentioned). Boost.Real does not seem to target high-performance computing, because the approach of using a variable number of intervals to represent floating point numbers does not appear to be particularly cache and/or SIMD friendly. Furthermore, the existing algorithms in high-performance computing are designed to deal with the limitations of float/double explicitly, by doing computations in a way that the FP error is kept under control. They do a good job. If high-performance is not the target anyway but accuracy is, then my first instinct would be to use a multi-precision type rather than Boost.Real. Best regards, Hans

1538

Age (days ago)

1546

Last active (days ago)

List overview

Download

15 comments

9 participants

participants (9)

Damian Vicino
Hans Dembinski
Kostas Savvidis
Mateusz Loskot
pbristow＠hetp.u-net.com
Richard Hodges
Robert Ramey
sehe
Vinnie Falco