43
$\begingroup$

In the edit queue, I've seen some improvements of non-MathJax code into MathJax. However, some Unicode characters were untouched and they were actually displaying fine.

Are there downsides in using $\sin(θ)$ \sin(θ) instead of $\sin(\theta)$ \sin(\theta)?

Is there a preferred course of action?

As a LaTeX guy, I'm in favor of consequently using TeX commands, but maybe this is just being old-fashioned.

$\endgroup$
12
  • 1
    $\begingroup$ Unless unicode is displaying incorrectly (which it shouldn't, but I don't know), this would be something I'd just reject as 'too minor'. If it's bundled with a more substantial edit, I'd take it. $\endgroup$
    – Milo Brandt
    Feb 18, 2016 at 23:50
  • $\begingroup$ @MiloBrandt: The main point of my question is not about the review queue, but rather about the mix of unicode and mathjax in general. The edit was generally introducing MathJax, but partially keeping the unicode symbols. I was wondering if there are any downsides to this. $\endgroup$
    – Roland
    Feb 19, 2016 at 0:21
  • 3
    $\begingroup$ Some of the things which can be influenced by the choice of Unicode or MathJax in the title are whether the question is eligible for the network-wide hot questions list and also which questions are shown among related questions in the sidebar. I have discussed this briefly (and given some examples illustrating this) here and here. $\endgroup$
    – Martin Sleziak
    Feb 19, 2016 at 2:09
  • 1
    $\begingroup$ This is only partially related, but I just noticed it the other day. Before switching to Common-HTML as the Mathjax renderer when Mathjax was updated, it was glaringly obvious when Unicode was used compared to Mathjax, and I liked to edit them away. But with Common-HTML, the two are indistinguishable (when the Unicode is in math mode!). I'll still edit Unicode away out of habit, but it feels (visually) pointless now. $\endgroup$
    – pjs36
    Feb 19, 2016 at 4:14
  • 1
    $\begingroup$ I think this is like asking if we should write some words in a manuscript with black ink and others with a pencil. And for pretty much the same reason I'd never do it with both. $\endgroup$
    – hjhjhj57
    Feb 19, 2016 at 7:14
  • 6
    $\begingroup$ IMO, θ looks ugly, and θ isn't much better. But either $\theta$ or $θ$ is fine. $\endgroup$
    – Ilmari Karonen
    Feb 20, 2016 at 0:56
  • 4
    $\begingroup$ At least for the less familiar Greek letters, like $\Xi, \xi, \Psi, \psi, \chi$ you can more easily find their name from the TeX source and not have to go to some other website. $\endgroup$
    – Robert Soupe
    Feb 20, 2016 at 5:33
  • $\begingroup$ @RobertSoupe Maybe, but most of the Greek unicode characters have almost the same code, so its almost easier because you don't have to know their names in LaTeX. $\endgroup$
    – Simply Beautiful Art
    Feb 28, 2016 at 2:19
  • 2
    $\begingroup$ @SimpleArt I for one want to have something to pronounce, even if it's only in my mind. I don't like saying "unfamiliar Greek letter squared plus 1" or whatever. $\endgroup$
    – Robert Soupe
    Feb 29, 2016 at 5:00
  • $\begingroup$ @RobertSoupe LoL, yes, but its just some weird fancy letters someone had to use because the English alphabet was overused. $\endgroup$
    – Simply Beautiful Art
    Feb 29, 2016 at 22:26
  • 1
    $\begingroup$ In Wikipedia articles, the version of TeX or LaTeX or whatever it is (it's not MathJax), when used in an inline setting rather than a displayed setting results in mismatches in font size and in misalignments. In math(dot)stackexchange(dot)com that's not a problem, so MathJax appears preferable to unicode in all situations. $\endgroup$
    – Michael Hardy
    Mar 3, 2016 at 4:32
  • 1
    $\begingroup$ @RobertSoupe This is exactly why $\phi$ is to be preferred to $\varphi$. Not everyone familiar with $\phi$ is familiar with $\varphi$, and even if someone's seen $\varphi$ they might not yet know that it means $\phi$. $\endgroup$
    – Rosie F
    Feb 26, 2019 at 11:21

1 Answer 1

Reset to default
43
$\begingroup$

First, as explained in this meta post, you should always mark your formulas as math using dollar signs or the such. This is for people with screen readers, automatic processing of posts... A bare θ, if it's supposed to represent a math variable, is no good. (Of course if you're writing something in Greek then you should not mark it as math.)

Now, MathJax has a feature that makes $θ$ and $\theta$ have the exact same output. So in this regard there's no difference. But consider these points:

  • Why use LaTeX commands for most things and Unicode characters for a small portion of them? It's not consistent. In your example $\sin(θ)$, you have one LaTeX command and a Unicode character. In a big formula it would be even more jarring, IMO, especially if you use LaTeX commands for some symbols which exist in Unicode.
  • Some people do not know how to input these Unicode characters, or even think of copy-pasting them, and if they edit your formulas it will become even more inconsistent.
  • If you use Unicode characters in your formula, then you cannot directly use the formula anymore in a LaTeX document.
  • MathJax has this feature now, what if they decide to drop it to be more in line with LaTeX syntax? Or, what if SE decides to use another math formatting library in the future that doesn't have this feature? (It's not very realistic, but it could happen.)
  • For some symbols the font is different: compare $\large 𝔸\mathbb{A}$ (obtained with $\large 𝔸\mathbb{A}$). At least on my device I see two different fonts, and it's probably dependent on your OS, browser and installed fonts.

These are small downsides, but to be honest I can't think of any upside of using Unicode characters: they're hard to input for most people, compared to ASCII.

$\endgroup$
16
  • 24
    $\begingroup$ The upside of using Unicode characters is that you can put more content into a single comment, as these allow only a certain number of characters. This is why I sometimes use × or ⟶ instead of \times or \longrightarrow, at least in comments. $\endgroup$
    – Stefan Hamcke
    Feb 19, 2016 at 16:54
  • 7
    $\begingroup$ I find it curious, that on my system the two versions of theta look different while the two versions of "large A" look exactly the same... (OS X / Safari) $\endgroup$
    – Stefan Mesken
    Feb 20, 2016 at 0:33
  • $\begingroup$ I completely understand what you mean here, but isn't most of ASCII part of Unicode now, the first block, in fact? Of course it would be cumbersome to say "Unicode beyond the first block." $\endgroup$
    – Robert Soupe
    Feb 20, 2016 at 5:03
  • 1
    $\begingroup$ @RobertSoupe It seems that you answered your own question. The first 128 characters of Unicode are the same as ASCII (the "Basic Latin" block), and if encoded with UTF-8, they're even encoded the same. Given the context, I thought it was clear that I was speaking about non-ASCII Unicode characters that replicate LaTeX commands. $\endgroup$
    – Najib Idrissi
    Feb 20, 2016 at 8:18
  • 2
    $\begingroup$ @NajibIdrissi You're right, "Unicode" here is a shorthand for an even more cumbersome expression, " non-ASCII Unicode characters that replicate LaTeX commands." $\endgroup$
    – Robert Soupe
    Feb 20, 2016 at 17:59
  • 1
    $\begingroup$ @Stefan: Maybe you are using HTML-CSS as renderer. With Common HTML its just the other way, at least on Linux/Firefox. To change the renderer, right-click on some formula, then go to Math Settings -> Math Renderer $\endgroup$
    – Stefan Hamcke
    Feb 20, 2016 at 19:11
  • $\begingroup$ @Stefan You are right - that's it! $\endgroup$
    – Stefan Mesken
    Feb 20, 2016 at 20:41
  • 17
    $\begingroup$ There is one obvious advantage of unicode characters: that the unrendered code looks nicer, which is good for editability. I personally don't type unicode characters when doing LaTeX math, but I increasingly think I should start doing so. Some may consider this “not in the spirit” of TeX, drifting in the wrong WYSIWYG direction, but I'd disagree: LaTeX separates presentation (which can't be written in plaintext) from content (which can), but single characters are content, and can be written in Unicode plaintext, hence it is just consequent to use them if that's possible. $\endgroup$
    – leftaroundabout
    Feb 21, 2016 at 21:11
  • 2
    $\begingroup$ Unicode is meant to be a universal language. We don't need that full power here. We only need ASCII + Latex. $\endgroup$
    – PyRulez
    Feb 22, 2016 at 13:19
  • 5
    $\begingroup$ Using Unicode makes the source shorter and more readable. And that source can be used in LaTeX docs if you use the Unicode Math package. $\endgroup$
    – bubba
    Feb 25, 2016 at 13:11
  • $\begingroup$ @bubba Only with XeTeX and LuaTeX. $\endgroup$
    – Najib Idrissi
    Feb 25, 2016 at 13:36
  • $\begingroup$ If you write math in sphinx it is readable and well formatted 1) in the source text, 2) on HTML because of this mathjax feature, 3) in latex using unicode-math and latex_engine = 'xelatex'. $\endgroup$
    – Roland Puntaier
    Dec 20, 2016 at 17:02
  • 1
    $\begingroup$ Related to your last point: quite a view browsers/devices do not (yet) support all those more uncommon unicode characters. They are then shown as a blank square or a square filled with the charactercode. $\endgroup$
    – quid Mod
    Mar 4, 2017 at 15:32
  • $\begingroup$ Note that $\thetaθ$ are not exactly the same. The first is in italic and the second is in roman. $\endgroup$
    – robjohn Mod
    Dec 6, 2018 at 5:43
  • $\begingroup$ @robjohn As far as I remember, this wasn't the case in February 2016. But thank you for letting me know about this crucial issue. $\endgroup$
    – Najib Idrissi
    Dec 6, 2018 at 8:17

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .