Use case for Latin Modern Math (LM Math)

Question

Since July 2011, there has been a beta release of LM Math. Some people seem to be quite excited about this. The release announcement states that this release

...completes the modernization of the Computer Modern family of typefaces .... the modernization was incomplete without the math fonts of the Comupter Modern family

I understand many reasons to use the lmodern text fonts:

• better support for diacritics and non-english glyphs than CM (allowing hyphenation, etc). See this FAQ answer for more details;
• allowing correct search and copy/paste from the pdfs unlike the virtual font-based solutions (\usepackage{ae}, \usepackage{zefonts});
• decent type 1 outlines, rather than EC-fonts' bitmapped type 3, or the slightly rough autotraced type 1 from CM-Super, etc;
• slightly improved metrics and diacritic positioning compared to EC, and a small number of redesigned/improved glyphs and ligatures;
• some additional shapes and weights compared to the bluesky/ams/y&y CM (eg: roman slanted smallcaps, light/dark/condensed typewriter). See the table at the end of this document.

As far as I can tell, none of these advantages apply to the LM Math fonts. So my question is, why did the GUST team feel the need to produce new math fonts? As far as I can tell the only advantage that they quote on their release announcement is that they are "modern", but I fail to find "new" a compelling argument. Why would I use a newer, less-tested font (that, if it is anything like the other LM fonts, is likely less well-hinted than bluesky/ams/y&y CM)?

I understand that this in an early beta release, so it is not suprising that there is not much (any?) documentation or rationale accompanying the fonts, but I hope someone here knows the reason for them.

EDIT: I looked a bit harder, and I saw that my installation of the lm package, version 2.004, 30.10.2009 already includes math fonts. In LaTeX I can use them by default \usepackage{lmodern}, or not by \usepackage[nomath]{lmodern}. The README says that these 20 math fonts are "at the moment a duplicate of PL math fonts", which presumably refers to these fonts. So, presumably, this new LM Math release is to replace these PL math fonts with something else. So, in addition to my question about what was wrong with the bluesky/ams/y&y CM math I'd also be interested to know: what was wrong with the PL math fonts?

Answer 1

The LM Math fonts are available in OTF Math format. As such they are usable with unicode-math and ConTeXt Mk IV. So you can finally use LM when using Xe/LuaTeX + unicode-math or ConTeXt (as LM + STIX Math just looks horrible). Presumably they also added additional symbols as Unicode defines lots of math symbols that aren't available in the CM math fonts.

As I read the announcement, you can't use the fonts without an OTF capable TeX engine. So with pdflatex, you have to continue to use the old fonts anyway (see the comments for a qualification of this statement).

Answer 2

Let me provide a different take on this. This has more to do with the fact why unicode enabled math typesetting is useful, but does illustrate why opentype math fonts help.

ConTeXt allows you to export a TeX document to XML. In such an export math is exported to MathML.

Using an opentype math font allows you to look at the final glyph that is inserted into the pdf, and use that in the exported MathML. For example, $\Gamma$, which is normally typeset upright, is exported to <math><mi>0x0393</mi></math>. However, if you set an appropriate option so that uppercase Greek letters are typeset in italic, then $\Gamma$ is exported to <math><mi>0x1D6E4</mi></math>. The export functions do not care how the mapping from $\Gamma$ to the appropriate font glyph is done. It simply uses the opentype slot of the final glyph in the exported MathML.

Of course, it is still possible to implement such an export with Type1 fonts. However, in that case, the export functions will need to understand the mappings from macros (or input characters) to unicode slots. Thus, the macro package needs to implement two mappings: (i) input character to font glyph; (ii) input character to opentype location. With an opentype math font, these two mappings collapse to one. For this reason, even type1 math fonts are converted into a fake opentype math font on the fly in ConTeXt MkIV.

It is also possible (although not currently implemented) to embed the converted MathML into the pdf file so that a screen reader will read the mathematical expression correctly. This will improve the accessibility of TeX documents. Another advantage is that if you copy a math expression from a pdf and pastes it in a text file, the resulting MathML will be pasted. Wrap it around an appropriate environment, and then you can compile the expression with TeX again and recover the original math expression. (ConTeXt has a module for typesetting MathML; I haven't checked whether the LaTeX XML packages allow typesetting of MathML). This is similar to using an appropriate cmap file so that you can copy and paste text.

Using an opentype math fonts makes it easier to implement such features. And, of course, if you want a good looking pdf, you need a good (and preferably free) math fonts. Currently they are in short supply; so a new opentype math font is a welcome addition.