Indices with German Umlaut ä - \mathrm{} or \text{}?

Question

I'm wondering about the correct use of \mathrm{} and/or \text{} (or similar) in formulas. I've read these topics (and some more):

Is there a preference of when to use \text and \mathrm? and

Difference between \textrm{} and \mathrm{} and

Is \mathrm really preferable to \text? [duplicate]

but I'm not fully satisfied with my findings yet. As far as I understood it's best to use \text{} for something like:

$ a = b \text{if, and only if, ...} $ or $ C = \text{const.} $

while \mathrm{} would be used for indices, like

$ \rho_\mathrm{water} $ or $ m_\mathrm{main} $

because \text{} does not necessarily provide upright font (e.g. in an italic context).

My problem with \mathrm{} is that it doesn't accept German umlauts. I want to name a parameter $ m_{äq} $, which is short for for "m äquivalent" (an equivalent mass). When I use m_\mathrm{äq} the "ä" will not be visible, even m_\mathrm{{\"a}q} doesn't work. m_\text{äq} works fine, but it feels like \text{} is not the correct command in this case.

I'm using mathtools (including amsmath) and

\usepackage[ngerman]{babel}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

if this is of any importance.

Answer 1

You should do use \textup or \textrm as noted in comments.

The reason why it doesn't work is that even if you specify T1 for text math still uses the 7bit OT1 encoding for Roman letters.

You can change that:

\documentclass{article}

\usepackage[ngerman]{babel}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

\usepackage{mathtools}

\DeclareSymbolFont{operators}{T1}{cmr} {m}{n}
\makeatletter
\def\@inmathwarn#1{}
\makeatother

\begin{document}



$A_{\mathrm{Äpfel}}$



\end{document}

which works but try

$\Gamma=\Delta$

and you find that Uppercase greek (and several other commands, notably math accents) are expecting the OT1 encoding. Of course you could assign Greek to another font family and redefine all the commands, but many many packages would then need variant versions of their definitions to match the new setup.

---

A working version:

\documentclass{article}

\usepackage[ngerman]{babel}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

\DeclareSymbolFont{mathrmx}{T1}{cmr} {m}{n}

\DeclareSymbolFontAlphabet\mathrmx{mathrmx}
\usepackage{mathtools}

% make Ä follow the current math alphabet rather than
% use fam0 which os cmr which does not have the character.
\mathcode"C4="71C4


\makeatletter
\def\@inmathwarn#1{}
\makeatother

\begin{document}
\huge

$A_{\mathrmx{Äpfel}}$


$\Gamma=\Delta$

\end{document}

Answer 2

Just use \textrm{äq} for the subscript.

\documentclass{article}

\usepackage[ngerman]{babel}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

\usepackage{mathtools}

\begin{document}



$A_{\textrm{Äpfel}}$

\end{document}

Answer 3

If you have a few textual subscripts and no special characters in them, use \mathrm; otherwise use \textnormal

The second method, however, has a drawback: it doesn't honor \mathversion{bold}, as seen from the following example.

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}

\begin{document}

$A_{\mathrm{apple}}$

{\mathversion{bold}$A_{\mathrm{apple}}$}

$A_{\textnormal{Äpfel}}$

{\mathversion{bold}$A_{\textnormal{Äpfel}}$}

\end{document}

In the case of just a few of these subscripts, you can probably get away by adding \bfseries where needed; or, if you don't use bold math, there's nothing to worry about, probably. However, some classes (the Springer ones, for instance) add \mathversion{bold} (or the alias \boldmath) to the code for section titles.

Thus a personal command seems best:

\DeclareRobustCommand{\tsb}[1]{% textual subscript
  \textnormal{\setupforversion#1}%
}

\makeatletter
\newcommand{\setupforversion}{%
  \csname setupfor\math@version\endcsname
}
\makeatother

\newcommand{\setupfornormal}{}
\newcommand{\setupforbold}{\bfseries}
% set up also for possible other math versions you use

Full example

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}

\DeclareRobustCommand{\tsb}[1]{% textual subscript
  \textnormal{\setupforversion#1}%
}

\makeatletter % some syntactic sugar
\newcommand{\setupforversion}{%
  \csname setupfor\math@version\endcsname
}
\makeatother

% the commands to execute for the various versions
\newcommand{\setupfornormal}{}
\newcommand{\setupforbold}{\bfseries}

\begin{document}

$A_{\mathrm{apple}}$

{\mathversion{bold}$A_{\mathrm{apple}}$}

$A_{\tsb{Äpfel}}$

{\mathversion{bold}$A_{\tsb{Äpfel}}$}

\end{document}

Answer 4

Since I've been in doubt myself a few times (and not only with this), I think the best is to create an universal way of doing it so, in case one needs to change it, you just need to go back to the preamble.

With the code I post here the following syntax works: A_|whatever| (one could choose another like A_[foo] or A_(bar), but I prefer those for other things).

Basically we make _ active, and, if it's followed by a | then the argument is delimited by two |..| rather than {..}, and it's put in whatever font we choose.

\documentclass{scrartcl}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{mathtools}

\makeatletter
\def\sbtext|#1|{\sb{\textnormal{#1}}}
\begingroup\lccode`\~=`\_\lowercase{\endgroup\def~}{\@ifnextchar|\sbtext\sb}
\makeatother

\AtBeginDocument{\catcode`\_=12 \mathcode`\_="8000 }

\begin{document}

$A_{\mathrm{apple}}$

$A_|apple|$

$A_|Äpfel|$

$A_{\textnormal{Äpfel}}$

\end{document}