Re: [EXTERNAL] Re: New writeup for intent examples

Hi Murray, all,

If we could reliably defer to Unicode to provide all mathematical
intent for single glyphs, that would have been one task less to
consider, but of course that's not something we can commit to.

An ascii "d" should be possible to remediate as a differential,
without requiring the authors to use the double-struck style.

Similarly we can't expect separate Unicode entries for all
mathematical operations that use arrows, vertical bars, and even other
single letters. E.g. in lambda calculus one finds a "λx" that serves a
very similar purpose to "dx" (or "ⅆx"), but we only have one Unicode
lambda.

So it would be nice if we can get to a place where we don't need to
re-examine Unicode reliance as a primary mechanism in the a11y spec.
Narrating the glyph is the baseline fallback, but taking it further
should be well-qualified in some limited scope of defaults, such as
subject areas (if we end up introducing such mechanisms).

To get a real world example with fully developed complexity, take a
look at equations (1) through (9) on page 2 of this arXiv PDF:
https://arxiv.org/pdf/1804.01919.pdf

You will find both "d" and "ⅆ" used as differential-d in integrals,
and you will also separately find both used as variables in the
equations for differential forms (which btw are way above my own
paygrade). They even make this distinction explicit in the same page -
the unicode glyph stands for "the deRahm differential", while the
simple d is used for "the space-time differential".

All to say that we shouldn't over-rely on the unicode glyphs
themselves as a primary "intent" mechanism.

Greetings,
Deyan


On Thu, Nov 12, 2020 at 10:27 AM Murray Sargent
<murrays@exchange.microsoft.com> wrote:
>
> The intent is conveyed in both cases by using 2146 for the d.
>
> Get Outlook for iOS
> ________________________________
> From: David Farmer <farmer@aimath.org>
> Sent: Thursday, November 12, 2020 4:05:52 AM
> To: public-mathml4@w3.org <public-mathml4@w3.org>
> Subject: [EXTERNAL] Re: New writeup for intent examples
>
>
> I'll repeat a point I made earlier about integrals with a
> weight function.
>
> The following expression is natural in the context of Chebyshev
> polynomials (similarly for any set of orthogonal polynomials):
>
> \int_{-1}^1 f(x) \frac{dx}{\sqrt{1-x^2}}
>
> The " 1/\sqrt{1-x^2} " is distinguished.  A similar situation
> arises in integral transforms.
>
> I suggest these "weighted integrals" should have a way of denoting
> the weight.
>
>
> I'd like to know how Sam's  "Differential alone in the numerator"
> compares to
>
> \int_0^1 \frac{1}{x^2 + 1} dx .
>
> Does that have the same intent?
>
>
> On Wed, 11 Nov 2020, Neil Soiffer wrote:
>
> > I figured I should read more carefully what you wrote in
> > https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmathml-refresh.github.io%2Fmathml%2Fdocs%2Fintent.html&amp;data=04%7C01%7Cmurrays%40exchange.microsoft.com%7Ccedd31445d00400b07b508d887035c21%7C72f988bf86f141af91ab2d7cd011db47%7C0%7C0%7C637407795789448872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=k3yMvi3zTKaDzc0gGXyfHusMMI7omdNYO2e4we%2BkQUk%3D&amp;reserved=0 even though you hadn't done an update yet.
> > In case you didn't fix it, the MathML for "Binomial as stacked numbers" is not right. Probably you
> > want it to be an mfrac, but an mtable could also be used. Kind of garbled in the version I read.
> >
> > I still don't like the way you handle plus/minus, but that's not really a criticism of the intent
> > idea...
> >
> > I don't think the "Differential alone in the numerator" is correct. The 'intent' on the mfrac should
> > block the higher level intent from seeing the 'x' inside it. Further, the 'x' should not be in an
> > <mtext>. Same issues for the next differential examples.
> >
> >     Neil
> >
> >
> >
> >
> > On Wed, Oct 21, 2020 at 12:41 PM Sam Dooley <samdooley64@gmail.com> wrote:
> >       Hello all,
> >
> >       Regrets for tomorrow's meeting, I will be having cataract surgery.
> >
> >       https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmathml-refresh.github.io%2Fmathml%2Fdocs%2Fintent.html&amp;data=04%7C01%7Cmurrays%40exchange.microsoft.com%7Ccedd31445d00400b07b508d887035c21%7C72f988bf86f141af91ab2d7cd011db47%7C0%7C0%7C637407795789448872%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=k3yMvi3zTKaDzc0gGXyfHusMMI7omdNYO2e4we%2BkQUk%3D&amp;reserved=0
> >
> >       On the plus side, I did my homework, and created a new document to describe as best I
> >       can what I believe is the latest consensus on the intent attribute.  Not the final word,
> >       as there are still things to discuss, and it is certainly biased toward my preferences,
> >       but hopefully not too badly.
> >
> >       I was able to include examples that should address Bruce's concerns with the handling of
> >       transpose.  To be continued.
> >
> >       If an element has sub elements with intent, then intent="fn" will collect them as
> >       arguments to fn.  If an element has no such sub elements, then intent="transpose" gives
> >       the intent of the transpose function itself, with no arguments.  If an operator has no
> >       arguments, and you want the intent of the application of the function, use
> >       intent="fn()".  Easy as pi, but we should discuss.
> >
> >       The operator name can be placed on the enclosing element for the apply, or on an element
> >       that gives markup for the operator.  This should allow for what folks want, but we
> >       should discuss.
> >
> >       I've included examples with both argument index references, and argument name
> >       references.  I'd really like to avoid XPath references.
> >
> >       I was able to expand on Bruce's examples where multiple infix/prefix/postfix operators
> >       appear in a single mrow, and I marked up both minimal-mrow and complete-mrow versions of
> >       each example.  To be discussed.
> >
> >       I've included examples with integrals of fractions where the differential is included in
> >       the fraction.  We should discuss scoping of argument name references.
> >
> >       I've not said anything about literal references, which I intend to add.
> >
> >       Oh yes, and I still need to convert this to markdown, once we stop adding examples to
> >       it.  I've not gone through the entire encyclopaedia.
> >
> >       This version is intended to be more descriptive than prescriptive.  The examples are
> >       informally grouped to illustrate how to use the syntax.
> >
> >       Enjoy,
> >       Sam
> >
> >
> >

Received on Thursday, 12 November 2020 16:37:06 UTC