Re: "They Use ∂ Differently in Math and Physics. Which is Better?"

Deyan --

You ask:

Should each of the 10+ LaTeX-to-HTML converters invent its own
system-specific macro dialect for depositing MathML Intent?


Just so it is clear to all, the GELLMU Project does not have a
LaTeX-to-HTML converter.  Its philosophy is that authors should write
"profiled LaTeX" (see my talk at TUG in 2010), which is to say they should
write LaTeX in such a way that it becomes completely straightforward to
translate it at the level of syntax to an SGML document, which in turn, can
be translated (with some dependence on vocabulary, e.g., looking up names
for positional command arguments) to an XML document having a tag
vocabulary, including math, that is as rich as that of LaTeX.  The XML
document is robust for converters to many formats.  The idea is that if
authors write profiled LaTeX, then conversion of documents to other formats
is fairly easy.  There can be many profiles.  Each profile has a fixed
command vocabulary, so a correctly written converter should never run into
things that it does not know how to handle.  Of course, each profile needs
its own set of converters.

This means, for example, authors must always write "\frac{1}{2}" instead of
"\frac12", and I've been told that's an unreasonable demand.  :-)

Best,

        -- Bill



William F Hammond
Email: gellmu@gmail.com
https://www.facebook.com/william.f.hammond
http://www.albany.edu/~hammond/

𝑻𝒉𝒆 𝒕𝒊𝒎𝒆 𝒕𝒐 𝒔𝒂𝒗𝒆 𝒂 𝒅𝒆𝒎𝒐𝒄𝒓𝒂𝒄𝒚 𝒊𝒔 𝒃𝒆𝒇𝒐𝒓𝒆 𝒊𝒕
𝒊𝒔 𝒍𝒐𝒔𝒕.   -- 𝐊𝐞𝐧 𝐁𝐮𝐫𝐧𝐬




On Fri, Jun 21, 2024 at 3:59 PM Deyan Ginev <deyan.ginev@gmail.com> wrote:

> Hi Bill,
>
> You've reminded me of an important question, which luckily is only
> tangential to MathML, but is truly a LaTeX one:
>
> Should each of the 10+ LaTeX-to-HTML converters invent its own
> system-specific macro dialect for depositing MathML Intent? And separately
> yet, there are 10+ formula-level LaTeX-on-the-web renderers. This is
> currently an intractable question - leaving it to the LaTeX Team itself
> seems to be an unfair burden at this late date. But who else?
>
> ---
> Separately:
>
> "What kinds of meta-information might you want?"
>
> Let me rephrase that to:
>
> "What kinds of applications do your authors want to enable via explicit
> annotations?"
>
> The answer may vary based on which community your tool is servicing.
>
> One of the nice aspects of MathML (and Intent) is that it can be generated
> by all kinds of methods, not just custom LaTeX macros. So you can imagine
> various other systems experimenting with the authoring syntax, or even
> (dare I say it) creating datasets for training AI models to contextually
> enrich plain MathML into intent-rich MathML.
>
> All sorts of opportunities...
>
> Greetings,
> Deyan
>
>
> On Fri, Jun 21, 2024 at 6:44 PM Hammond, William F <whammond@albany.edu>
> wrote:
>
>> (Sorry for finding myself in a mailer that makes it hard not to top-post..)
>>
>> This an aside about GELLMU, where there is a \newcommand variant
>> \mathsym, taking two arguments and an optional third argument, i.e., with
>> usage
>>
>>         \mathsym{ symbol-name }{ symbol-rendering }[symbol-meta-info ]
>>
>> So, in this case, as a very simple example,
>>
>>          \mathsym{\bdy}{\partial} ,
>>
>> might be used to flag "boundary" usage of the character ∂.  For
>> more information see section 6.4 of the GELLMU Manual, e.g.,
>> https://www.albany.edu/~hammond/gellmu/glman/glman.html#SU-6.4
>>
>> The optional third argument is for the future -- whatever meta
>> information might be handled in HTML (with MathML) output by (1) web
>> browsers or (2) computer algebra systems (in the case where HTML clips are
>> pasted from a browser).  Needless to say, the GELLMU Project, as it is,
>> contains no non-trivial code for dealing with the optional third argument.
>> What kinds of meta-information might you want?
>>
>> It would be up to authors to supply the meta information and up to you
>> (well, the MathML spec) to indicate what meta information should be passed,
>> and how it should be passed, in HTML/MathML output.
>>
>>         – Bill
>>
>> ------------------------------
>> *From:* Deyan Ginev <deyan.ginev@gmail.com>
>> *Sent:* Thursday, June 20, 2024 7:21 AM
>> *To:* Abbas Jaffary <abbas.jaffary2@gmail.com>
>> *Cc:* www-math@w3.org <www-math@w3.org>
>> *Subject:* Re: "They Use ∂ Differently in Math and Physics. Which is
>> Better?"
>>
>> Hi Abbas, all,
>>
>> Interesting additions, thank you.
>>
>> For k-forms, since those are "differential forms", I wonder if the use
>> already fits in the conventions described in the video.
>>
>> I see you are referring to overloading the notational use of "∂" with the
>> boundary example in homology.
>> I wasn't familiar with it until now, wikipedia has a nice overview here:
>>
>> https://en.wikipedia.org/wiki/Homology_(mathematics)#Construction_of_homology_groups
>>
>> Different "named concept" uses of ∂ would be relevant additions for the
>> Intent Open list, especially in cases where - quoting you - "it would be
>> nice to say boundary". That is useful for accessibility.
>>
>> The original youtube video is closer to the OpenMath questions, as it
>> reveals how an operator with the appearance of full formalization -
>> "partial derivative" - may represent (at least) two different formal
>> definitions.
>> In OpenMath terms, one could have cast the video exposition as two
>> symbols "convention-m:partial-derivative" and
>> "convention-p:partial-derivative".
>>
>> But I doubt generator tools can realistically infer these without some
>> very purposeful additional help from authors - such as a brand new
>> notational convention.
>>
>> Deyan
>>
>>
>> On Thu, Jun 20, 2024 at 8:56 AM Abbas Jaffary <abbas.jaffary2@gmail.com>
>> wrote:
>>
>> Very interesting!  There is also the differential geometry use of
>> derivatives as one-forms (and k-forms), and the as boundary operator in
>> algebraic topology. In homology, for example, one could likely infer
>> "partial x" as the "boundary of x", though it would be nice to have it say
>> "boundary".
>>
>> I imagine there will be some crowdsourced effort to accommodate the most
>> popular use cases for fundamental symbols.  Not sure where OpenMath is with
>> this?
>>
>> On Wed, Jun 19, 2024 at 8:49 AM Deyan Ginev <deyan.ginev@gmail.com>
>> wrote:
>>
>> Hello everyone,
>>
>> I stumbled on a very well-made exposition video, which covers a subtlety
>> in the meaning of  "partial derivative" between mathematics and physics:
>>
>> https://youtu.be/QFHSHhpbo00
>>
>> Aside: This topic is not directly related to current conversations about
>> derivative syntax for "intent".
>>
>> Instead, the presenter has some well-reasoned general discussion, a
>> community proposal for adding yet-another notation, and showcases some of
>> the problems that our generator tools also face when trying to infer
>> Content MathML expressions from human-authored math syntax.
>>
>> Greetings,
>> Deyan
>>
>>