@dginev@mathstodon.xyz
@dginev@mathstodon.xyz avatar

dginev

@dginev@mathstodon.xyz

Building NLP tools for science docs. LaTeXML dev.

Creator of http://ar5iv.org

This profile is from a federated server and may be incomplete. Browse more on the original instance.

bkardell, to random
@bkardell@toot.cafe avatar

It feels like kind of a problem that companies attempt to organize on a planet-wide scale, but protections, etc don't

dginev,
@dginev@mathstodon.xyz avatar

@bkardell

This rhymes with the general theme of how deciding something "on one's own" is so much easier than deciding it together.

christianp, (edited ) to random
@christianp@mathstodon.xyz avatar

You are shown a set of four cards placed on a table, each of which has a number on one side and a color on the other.

The visible faces of the cards show yellow, 5, 4, purple.

Which card(s) must you turn over in order to test that if a card shows an even number on one face, then its opposite face is purple?

dginev,
@dginev@mathstodon.xyz avatar

@christianp The poll results remind me of LLM output probabilities. Crowd intelligence is a curious thing.

pandoc, to typst
@pandoc@fosstodon.org avatar

See https://pandoc.org/typst-property-output for more info about the new #typst property support introduced in #pandoc 3.2.

dginev,
@dginev@mathstodon.xyz avatar

@pandoc

Is this also available for CSS style properties in HTML input? They often have clear mappings into Typst markup.

christianp, to random
@christianp@mathstodon.xyz avatar

@dginev I can't remember if I've asked you this before - do you have any stats on the relative popularity of LaTeX packages in papers submitted to the arXiv?

dginev,
@dginev@mathstodon.xyz avatar

@christianp Oh sure, I don't think you've asked.

Only the arXiv has the "real" answer to that, since one needs to parse the .log files produced by pdflatex and collect what was actually loaded (including deep dependency loads). So far they haven't had the free time to study those logs, but maybe it can happen with the recent hires...

In any case, we do have a "latexml answer", based on what latexml loads (or fails to find support for). It's somewhat different than the real pdflatex answer. We load this list of files over arXiv upto 03.2024:

https://corpora.mathweb.org/corpus/arxmliv/tex_to_html/info/loaded_file

and we fail to find support for this list of files:

https://corpora.mathweb.org/corpus/arxmliv/tex_to_html/warning/missing_file?all=true

The catch is that in certain situations where we lack support for a .cls file, LaTeXML will load something called "OmniBus" which will try to preemptively add a safety net of packages that may or may not be used by the actual article. Which skews the stats towards those most common packages, such as article.cls or graphicx.sty

christianp, to typst
@christianp@mathstodon.xyz avatar

This morning I'm looking at .
The first thing in the tutorial (https://typst.app/docs/tutorial/writing-in-typst/) is how to write a header, and it annoys me that it makes the same mistake HTML, and everything following HTML, made: you specify the level of the heading absolutely, and it's not scoped to a section of the document.

So when you want to have a heading one level lower, you have to know what level the previous heading was. And you can't tell how much of the document the heading applies to, only inferring it as going until the next header of the same or higher level.

I've always wondered why has
\section{name}
instead of
\begin{section}{name} ... \end{section}

Am I alone in wanting this?

dginev,
@dginev@mathstodon.xyz avatar

@christianp Opposite taste here - it's just so much shorter to write \section that I have always wondered why most of LaTeX couldn't have been moved away from the verbose \begin \end boilerplate.

But, funnily enough, LaTeX already supports {section}, {subsection} etc. used as environments, so if you like that syntax you can use it. They even work in latexml.

typst, to random
@typst@mastodon.social avatar

What's the best table you have seen lately? No matter what it was, you can build it with Typst 0.11!

A table with a training log of date, activity, and calories burned. There is a double stoke between the first and the second row.
A chess board. Black (!) moves its pawn from E2 to E4. The field it has moved to has a light blue outline.
A table showing which team member is in the office, remote, or on leave each weekday. Adjacent columns with the same value are merged.

dginev,
@dginev@mathstodon.xyz avatar

@typst

My favorite table so far is 219 pages long, split in 2 columns:

https://arxiv.org/pdf/quant-ph/0504002.pdf

Looking forward to a Typst document which conquers that!

christianp, to random
@christianp@mathstodon.xyz avatar

If a person wanted to use wikidata to find lists of things sorted by some orderable property they have in common, how would a person do that?

dginev,
@dginev@mathstodon.xyz avatar

@christianp

Would a SPARQL query with the Wikidata Query Service fit the need?

Here is an example query with ORDER BY:

https://query.wikidata.org/#%23All%20items%20with%20a%20property%0A%23%20Sample%20to%20query%20all%20values%20of%20a%20property%0A%23%20Property%20talk%20pages%20on%20Wikidata%20include%20basic%20queries%20adapted%20to%20each%20property%0ASELECT%0A%20%20%3Fitem%20%3FitemLabel%20%3FnotationItemLabel%0A%20%20%3FnotationItem%0A%23%20valueLabel%20is%20only%20useful%20for%20properties%20with%20item-datatype%0AWHERE%0A%7B%0A%20%20%3Fitem%20wdt%3AP913%20%3FnotationItem%0A%20%20%23%20change%20P1800%20to%20another%20property%0A%20%20SERVICE%20wikibase%3Alabel%20%7B%20bd%3AserviceParam%20wikibase%3Alanguage%20%22%5BAUTO_LANGUAGE%5D%2Cen%22.%20%7D%0A%7D%0AORDER%20BY%20%3FitemLabel%20OFFSET%20100%0A%23%20remove%20or%20change%20limit%20for%20more%20results%0ALIMIT%20100

dginev,
@dginev@mathstodon.xyz avatar

@christianp oh sure, it takes a little getting used to...

It is close to ordinary SQL at the secondary conditions.
And the primary part is a series of "subject predicate object" triples, with ?variables.

Apparently they also added a Query Builder to avoid learning SPARQL entirely:
https://query.wikidata.org/querybuilder/?uselang=en

christianp, to random
@christianp@mathstodon.xyz avatar

I've just realised that since I have a cold I could just... not be working

dginev,
@dginev@mathstodon.xyz avatar

@christianp

Think of it as working hard on implementing health 😅

Feel better soon!

dginev, to webdev
@dginev@mathstodon.xyz avatar

LaTeXML 0.8.8 was just released!

👉 support for tikz-cd
👉 upgraded subfigure HTML model
👉 refinements to MathML Core
👉 better TeX emulation (👿 in details)
👉 refinements for 35+ LaTeX packages

https://github.com/brucemiller/LaTeXML/releases/tag/v0.8.8

dginev, to random
@dginev@mathstodon.xyz avatar

🚨 tikz-cd support just landed in LaTeXML

dginev, to random
@dginev@mathstodon.xyz avatar

I am endlessly amused by the technical honesty of making PDF accessible by attaching an HTML page as a secondary tree.

dginev, to random
@dginev@mathstodon.xyz avatar

Thanks again to everyone for supporting our CSS+MathML issue for interop 2024!

Sadly, we didn't make the cut this year. The hard questions remain the same - how do we get traction for a niche topic without an obvious business use case?

For now: Patiently. (🐢 > 🐇)

Official interop 2024 focus areas:
https://webkit.org/blog/14955/the-web-just-gets-better-with-interop/#the-2024-focus-areas

If you don't recall the issue, it was at:
https://github.com/web-platform-tests/interop/issues/556

gregeganSF, (edited ) to random
@gregeganSF@mathstodon.xyz avatar

Does anyone on Mathstodon know how to stop the LaTeX in posts like the one below disappearing beyond the right margin?

A double backslash seems to have no effect. I guess I could split the LaTeX into two completely separate display-mode lines, but that seems like a very clunky solution.

https://mathstodon.xyz/@gregeganSF/111852966029958683

Edit: Using \begin{aligned} ... \end{aligned} enabled \, thanks to @glc

dginev,
@dginev@mathstodon.xyz avatar

@gregeganSF @glc

Automatic line-breaking for math on the web, wouldn't that be nice!

Mathjax 4 has some further work in that direction, but I suspect it has to be optionally enabled:
https://github.com/mathjax/MathJax-src/releases/tag/4.0.0-beta.2#line-breaking

And one of these years we'll get to line-breaking for MathML, hopefully in the 2020s...

tonofcrates, to random
@tonofcrates@mastodon.social avatar

I wrote some ideas for how to actually make HTML documents a viable replacement for PDFs.

https://willcrichton.net/notes/portable-epubs/

dginev,
@dginev@mathstodon.xyz avatar

@tonofcrates

Forwarding to @robin . It is great to see more brainstorming on which HTML dialect can get us to "portable" files.

5t3ph, to webdev
@5t3ph@front-end.social avatar

📝 New post:
"A Call for Consensus on Semantics"

Partly a rant, minorly educational, mostly a call for help.

https://thinkdobecreate.com/articles/a-call-for-consensus-on-html-semantics/

dginev,
@dginev@mathstodon.xyz avatar

@5t3ph

Thank you for the thoughtful post!

Structural semantics has an associated tragedy, especially in markup languages - there are never enough element names to go around, and most constructs fit a little too awkwardly when used for anything else than the main purpose they were designed for.

That said, consensus is really important for tools working together well, kudos for calling for that!

I suspect it may be healthy to establish several different "consensus dialects" of HTML, trying to agree on best practices for different communities - news, entertainment, academia, commerce... I myself am still wishing for a revival of the Scholarly HTML effort, which was one step in that direction:

https://w3c.github.io/scholarly-html/

dginev, to random
@dginev@mathstodon.xyz avatar

To this day, the strangest coincidence I have been a part of:

Of all the places ar5iv could have been independently developed, it happened just across the river from arXiv's new home at Cornell Tech.

... imagine our collective shock when we realized!

bkardell, to random
@bkardell@toot.cafe avatar

Do y'all still use elk? Does it get hung up for you a lot where it is just taking forever to load?

dginev,
@dginev@mathstodon.xyz avatar

@bkardell

Never used elk, but I once used Moose.

https://metacpan.org/pod/Moose

christianp, to random
@christianp@mathstodon.xyz avatar

Who is working on accessibility of maths research?
I know lots of people looking at improving maths education, but is there anything interesting happening in research?
The only ideas I have at the moment are to talk to @pkra; Volker Sorge's work on spoken maths in MathJax; and the arXiv team's work on converting preprints to HTML.

Are there any physical disability issues that affect research mathematicians?

dginev,
@dginev@mathstodon.xyz avatar

@christianp

MathML Intent is pretty interesting, on the standards side of current ongoing work.

It takes some tug-of-war with the people focused on education, but I think we've come up with a good open-ended initial design that can (in theory) cover arXiv-level maths.

Actually getting such a project completed is another story altogether of course... We're likely to need a lot more markup, or a lot more AI.

dginev, to random
@dginev@mathstodon.xyz avatar

Farewell arxiv-vanity! 🫡

Ben and Andreas made a fantastic contribution to the community, and reinvigorated some of us old timers to keep working on an HTML format for arXiv. I am nostalgic already...

"Our job here is done"
https://www.arxiv-vanity.com/

dginev, to random
@dginev@mathstodon.xyz avatar

📢 arXiv just landed HTML articles!

Try them out, report issues you run into, and expect a gradual process for increasing coverage.

They're starting small - HTML for new articles only.

That complements today's ar5iv nicely, which lags 1 month behind.

christianp, to random
@christianp@mathstodon.xyz avatar

Is the arXiv running slowly for everyone today, or am I being soft-blocked?

dginev,
@dginev@mathstodon.xyz avatar

@christianp Fine on my end... Also don't tell anyone 🕵️ but you can reach a different machine by using the export domain: https://export.arxiv.org/

Which can be a helpful last resort.

dginev,
@dginev@mathstodon.xyz avatar

@christianp Benefits of a less-known mirror is that it can be faster when traffic gets high on the main site... But if everyone starts using it, it might also crash 😅

The main arXiv opens for me with basically no lag at all, so you may be getting throttled en-route to that IP address - but that's a wild guess.

dginev, to webdev
@dginev@mathstodon.xyz avatar

📢 Do you enjoy math equation support as part of the open Web?

If so, would you give a github "thumbs up" 👍 to help us signal we have ongoing community interest in CSS+MathML working in unison?

This year we are attempting a small step forward in CSS support for MathML elements, as part of WPT's ecosystem "interop" effort:

https://github.com/web-platform-tests/interop/issues/556

"Math on the web" is a niche topic, where we have pockets of interest in community silos. So every 👍 counts!

P.S. The issues are not actually selected based on the number of upvotes or comments, but having an indication of community interest is helpful - and encouraging - for all parties involved. Thanks!

dginev, to random
@dginev@mathstodon.xyz avatar

Hey @JacquesC2 ,
(and everyone else who may have thoughts here)

Context: Presentation MathML is now available in all browsers and is here to stay, thanks to MathML Core. One would wish we could also achieve a Content counterpart to that.

Imagine we (a handful of W3C Math WG members) had some dwindling amount of spare time and motivation to refresh/reconsider/reinvent Content MathML. And were technically able to host that as part of a W3C Community Group, in preparation for an in-group discussion (planned for MathML 5). That would give us a 2 year work horizon for fleshing out some community proposal draft - until October 2025.

Do you think we should try? The first step here would be to again try and reach out to (+ recruit) a wide range of professionals working on CAS, ATP/ITP, math-related web applications, etc. Then prepare together the full laundry list of issues that need to be addressed and improved upon. I can imagine such an undertaking succeeding in theory. The very large unknown for me is whether the community has enough people who are willing to join in to help perfect the spec, while also adopting it in their systems.

Knowing what you know about the history of Content MathML and OpenMath, do you think the time is now close to appropriate, things having cooled down for a while? Or should we wait another decade? Thanks.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • tacticalgear
  • thenastyranch
  • ngwrru68w68
  • magazineikmin
  • khanakhh
  • rosin
  • mdbf
  • Youngstown
  • slotface
  • everett
  • cisconetworking
  • kavyap
  • DreamBathrooms
  • anitta
  • InstantRegret
  • Durango
  • osvaldo12
  • ethstaker
  • modclub
  • GTA5RPClips
  • Leos
  • cubers
  • tester
  • normalnudes
  • megavids
  • provamag3
  • lostlight
  • All magazines
    •