Why does LaTeX drop the first letter of my variable name in an equation?

Question

Friends: Can anybody tell me why the letter "e" of entropy does not show up in the following snippet?

\documentclass{article}
\usepackage{amsmath}
\begin{document}
\begin{equation}
\begin{aligned}
\mathcal entropy[9,5] & = \mathcal - [(\frac{9}{14})*lg(\frac{9}{14}) + (\frac{5}{14})*lg(\frac{5}{14})] \\
& = - [(\frac{9}{14})*((lg(9)-lg(14)) + (\frac{5}{14})*((lg(5) - lg(14))] \\
& = - [(\frac{9}{14})*(\frac{log(9)}{log(2)}) - (\frac{log(14)}{log(2)}) + (\frac{5}{14})*(\frac{log(5)}{log(2)}-\frac{log(14}{log(2)})]\\
& = 0.940
\end{aligned}
\end{equation}
\end{document}

and continue in the next line working with the numerical expression but only after the equal sign. I do not want to keep repteating entropty =. Thanks for your time — Ramon A. Mata-Toledo, Apr 28 '18 at 18:11
That doesn't really answer my question. What is your \mathcal supposed to do for you? — Alan Munn, Apr 28 '18 at 18:12
Also related more generally: How bad for TeX is omitting braces {}, even if the result is the same? — Alan Munn, Apr 28 '18 at 18:13
If you want to spell out words like log and entropy in upright letters, that’s \operatorname{} or if you want to insert snippets of text, that’s the \text{} command from amsmath. I’m guessing that’s what you meant instead of \mathcal, but I’m not sure. — Davislor, Apr 28 '18 at 21:54

Heiko Oberdiek · Answer 1 · 2018-04-28T23:18:42.773

11

The following example

removes \mathcal that does not seem to have a purpose,
fixes parentheses,
sets entropy as text word and not as product of variables e, n, t, ...,
sets the operators lg and log in upright font,
tries to improve the formatting by
- better sizes of the brackets,
- removal of unneeded brackets,
- removal of unneeded operators (*),
- using package siunitx for formatting the final number.

The example file:

\documentclass{article}
\usepackage{amsmath}
\usepackage{mleftright}
\usepackage{siunitx}
% For text in math, see comments.
\newcommand*{\tx}{\textnormal}
\begin{document}
\begin{equation}
\begin{aligned}
\tx{entropy}[9,5]
& = - \mleft[\frac{9}{14}\lg\Bigl(\frac{9}{14}\Bigr)
  + \frac{5}{14}\lg\Bigl(\frac{5}{14}\Bigr)\mright] \\
& = -
  \mleft[
    \frac{9}{14}\bigl(\lg(9)-\lg(14)\bigr)
    +
    \frac{5}{14}\bigl(\lg(5) - \lg(14)\bigr)
  \mright] \\
& = -
  \mleft[
    \frac{9}{14} \left(\frac{\log(9)}{\log(2)} - \frac{\log(14)}{\log(2)}\right)
    +
    \frac{5}{14} \left(\frac{\log(5)}{\log(2)}-\frac{\log(14)}{\log(2)}\right)
  \mright]\\
& \approx \num{0.9402859586}
\end{aligned}
\end{equation}
\end{document}

The formula between the \left and \right brackets are put in a "inner" math formula. This can cause additional horizontal space around the formula, for example, there is a thin space between an unary minus symbol and the opening bracket. \mleft and \mright of package mleftright avoids this.

edited Apr 28 '18 at 23:18

answered Apr 28 '18 at 18:20

Heiko Oberdiek

271,626

1

If you change all instances of \left[ and \right] to \biggl[ and \biggr], respectively, TeX will have a chance to treat the initiai - symbols as a unary rather than as a binary operators, by reducing the amount of whitespace between - and \biggl[. – Mico Apr 28 '18 at 20:06
3

@Mico The minus sign after a relation symbol can never be a binary math atom. The space (\thinmuskip) is added between the unary minus sign as ordinary math atom and the inner math formula by \left and \right. I have updated the answer to use \mleft and \mright of package mleftright instead. – Heiko Oberdiek Apr 28 '18 at 20:49
Probably \mathrm{entropy} is better. – egreg Apr 28 '18 at 22:26
@egreg Unhappily, there is no "best" way, just different ways with more or less advantages and disadvantages depending on the current situation. More details, e.g.: Write 'text' correctly in equations – Heiko Oberdiek Apr 28 '18 at 22:53
@HeikoOberdiek Sorry, but \text is wrong anyway. If the equation is part of a theorem statement, “entropy” would be typeset in italics. – egreg Apr 28 '18 at 23:01
1

@egreg \mathrm is also wrong, it fails for text with text accents, for example. For my taste, using italics in theorems is horrible. Depending on the font settings, \textnormal is often a good compromise, but the name is lengthy. The best is probably a user macro that can be defined depending on the use cases, formatting requirements, and taste. – Heiko Oberdiek Apr 28 '18 at 23:05

Przemysław Scherwentke · Answer 2 · 2018-04-28T23:59:25.440

10

\mathcal is a macro with an argument. In your example, because of lack of {...}, the argument is e. But \mathcal works only with capital letters, hence the side effect.

Additionally, \mathcal is typically used to denote families of sets. It seems that you wanted some other effect.

edited Apr 28 '18 at 23:59

answered Apr 28 '18 at 17:56

Przemysław Scherwentke

37,268

Thanks, I just wanted to say entropy = math expression and continue on the next line but only working at the right hand side of the equal sign. There are about 5 lines of simple math manipulation. I am newbie in LATEX. What do you mean by (...)? – Ramon A. Mata-Toledo Apr 28 '18 at 18:14
@RamonA.Mata-Toledo {...} was for an argument, e.g. \matcal{Z}. – Przemysław Scherwentke Apr 28 '18 at 19:13
In unicode-math, \mathcal is defined for small Latin letters. U+212E is the Unicode character for script small letter e. The legacy stix package defines a \mathscr{e}. Agreed that it’s being used in error here, however. – Davislor Apr 28 '18 at 21:49
"\mathcal is typically used to...". Well, no, people use it for many other types of objects. I agree that in this case it is a slightly weird choice, though. – Pedro Apr 29 '18 at 13:57

Why does LaTeX drop the first letter of my variable name in an equation?

2 Answers2