Math & Science ⇒ Center line not centered in cases environment

[Cham]

I'm getting a small glitch in a cases environment. As you should see with this MWE, the middle line is not exactly aligned with the equal sign on the left, and the vertical spacing isn't pretty :

Code: Select all

\documentclass[12pt,letterpaper,twoside]{book}
\usepackage{lmodern}
\usepackage{amsmath}
\usepackage{tensor}
\usepackage{pifont}

\begin{document}

		\begin{equation*}
		\delta(x) \; \Rightarrow \; \tensor{\delta}{_{\varepsilon}}(x) =
		\begin{cases}
		0, &\text{if $x < -\, \varepsilon$.} \\[8pt]
		\displaystyle{\frac{1}{2 \varepsilon}}, &\text{if $-\, \varepsilon < x < \varepsilon$.} \\[8pt]
		0, &\text{if $x > \varepsilon$.}
		\end{cases}
	\end{equation*}

\end{document}

Here's a preview :

cases.jpg (22.22 KiB) Viewed 8441 times

I want the center line to be properly aligned with the "=". How to solve this ?

[Stefan Kottwitz]

Hi Cham,

very good minimal example!

That glitch only occurs because of the manually inserted space of 8pt. The first line is made higher, the second line is made higher, the last one isn't. Somehow we got an asymmetry. If you omit this, it's still perfectly centered.

We can stretch it in another way. One way is redefining cases in order to get it stretched, such as by \arraystretch. I did this in my blog some time ago. This introduces an optional argument for stretching:

Code: Select all

\documentclass[12pt,letterpaper,twoside]{book}
\usepackage{lmodern}
\usepackage{amsmath}
\usepackage{tensor}
\usepackage{pifont}
\makeatletter
\renewcommand*\env@cases[1][1.2]{%
  \let\@ifnextchar\new@ifnextchar
  \left\lbrace
  \def\arraystretch{#1}%
  \array{@{}l@{\quad}l@{}}%
}
\makeatother
\begin{document}
\begin{equation*}
  \delta(x) \; \Rightarrow \; \tensor{\delta}{_{\varepsilon}}(x) =
  \begin{cases}[2]
    0, &\text{if $x < -\, \varepsilon$.} \\%[8pt]
    \dfrac{1}{2 \varepsilon\strut},
        &\text{if $-\, \varepsilon < x < \varepsilon$.} \\%[8pt]
    0, &\text{if $x > \varepsilon$.}
  \end{cases}
\end{equation*}
\end{document}

You may notice that all is stretched, so the brace is higher too.

Another way is adding an invisible line (0pt wide) that is vertically centered, in the middle line. So \vcenter{\rule{0pt}{8ex}} it preserves the centering:

Code: Select all

\documentclass[12pt,letterpaper,twoside]{book}
\usepackage{lmodern}
\usepackage{amsmath}
\usepackage{tensor}
\usepackage{pifont}
\begin{document}
\begin{equation*}
  \delta(x) \; \Rightarrow \; \tensor{\delta}{_{\varepsilon}}(x) =
  \begin{cases}
    0, &\text{if $x < -\, \varepsilon$.} \\
    \dfrac{1}{2 \varepsilon\strut},
        &\text{if $-\, \varepsilon < x < \varepsilon$.}
        \vcenter{\rule{0pt}{8ex}}\\
    0, &\text{if $x > \varepsilon$.}
  \end{cases}
\end{equation*}
\end{document}

Stefan

[Cham]

Thanks a lot Stefan !

I'm studying the second (simpler) method, since I don't want to make my preamble more complicated as it is already !

EDIT : Why the [cmd]\strut[/cmd] command ?

[Cham]

Arrgh ! Your second solution doesn't work in the mdframed environment :

Code: Select all

\documentclass[12pt,oneside]{article}
\usepackage[letterpaper,margin=1in]{geometry}
\usepackage{lmodern}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{amsfonts}
\usepackage{mathtools}
\usepackage{tensor}
\usepackage{xcolor}
\usepackage{mdframed}
\begin{document}

\begin{mdframed}[linecolor=gray,linewidth=1pt,leftmargin=0cm,rightmargin=0cm,topline=true,bottomline=true,innertopmargin=6pt,shadow=true,shadowsize=2pt,shadowcolor=gray]

	\begin{equation*}
		\delta(x) \; \Rightarrow \; \tensor{\delta}{_{\varepsilon}}(x) =
		\begin{cases}
			0, &\text{si $x < -\, \varepsilon$.} \\
			\dfrac{1}{2 \varepsilon}, &\text{si $-\, \varepsilon < x < \varepsilon$.} \\
			\vcenter{\rule{0pt}{8ex}} \\
			0, &\text{si $x > \varepsilon$.}
		\end{cases}
	\end{equation*}

\end{mdframed}

\end{document}

[Stefan Kottwitz]

At first I inserted the \strut (an invisible line over the line height) because the fraction line seemed to be a bit too close to the lower part, I think. It can be omitted.

Stefan

[Stefan Kottwitz]

In the mdframed code you just have an additional \\ that should be removed, before the \vcenter.

Stefan

[Cham]

Ah, yes, it's working great now. Thanks again !