align vs equation

Question

I always use align in my documents, and avoid equation. Is there anything wrong with that? My reasoning behind this: align > equation, so why not use it?

Answer 1

While not exactly a bad idea in principle, unfortunately it is a bad idea in practise because align doesn't have the same feature as equation whereby less vertical space is added if a small equation is displayed after a paragraph that ended early on the line. For example, consider

\documentclass[twocolumn]{article}
\usepackage{amsmath}
\begin{document}
hello
\[
a+b=c
\]

hello
\begin{align}
a+b&=c
\end{align}
\end{document}

You should easily be able to see the extra space after the second ‘hello’.

Answer 2

equation "squeezes" the math (i.e. uses less white space between symbols) in order to avoid badboxes but align does not.

\documentclass[11pt,draft]{article}
\usepackage{amsmath}
\begin{document}
 Using ``equation'':
 \begin{equation*}
  1+2+3+4+5+6+7+8+9+10
  = 11+12+13+14+15+16+17+18+19+20+21
 \end{equation*}

 Using ``align'':
 \begin{align*}
  1+2+3+4+5+6+7+8+9+10
  = 11+12+13+14+15+16+17+18+19+20+21\\
 \end{align*}
\end{document}

Answer 3

Will all due respect to Will Robertson, actually I think that it is a bad idea in principle, because they mean different things and markup (be it *TeX, HTML,...) should be logical as much as possible.

If you are typesetting an equation, then use equation and if you desire alignment you can use aligned blocks inside the equation.

You can decide easily whether you are typesetting an equation or not, by thinking how to want to reference it. If you only want one equation number to be displayed and see yourself suppressing by hand any additional equation numbers, (say by using \notag) then you are typesetting an equation and your markup should reflect that.

Answer 4

align is unable to replicate the label behavior of equation + split, which is to have a single vertically centered label. One can suppress labels for specific lines of align using either \nonumber or \notag, but all unaltered lines get their own label.

\documentclass[twocolumn]{article}
\usepackage{amsmath}
\begin{document}
 Using ``equation'':
 \begin{equation}
  \begin{split}
   a+b
   &= c\\
   &= c\\
   &= c\\
   &= c
  \end{split}
 \end{equation}

 Using ``align'':
 \begin{align}
  a+b
  &= c\\
  &= c \notag\\
  &= c \nonumber\\
  &= c
 \end{align}
\end{document}

(Answer inspired by this source.)

Answer 5

if you are using both amsmath and amsthm, you end a proof with a one-line equation, and want to move the qed box up onto the line with the equation, align will happily overprint the box onto the end of the math. equation behaves. so does \[ ... \].

\documentclass{article}
\usepackage{amsmath}
\usepackage{amsthm}
\begin{document}
\begin{proof}
one-line display with \verb+align*+
\begin{align*}
 a + b = c \qedhere
\end{align*}
\end{proof}
\begin{proof}
one-line display with \verb+equation*+
\begin{equation*}
 a + b = c \qedhere
\end{equation*}
\end{proof}
\begin{proof}
one-line display with \verb+\[+ \dots\verb+\]+
\[
 a + b = c \qedhere
\]
\end{proof}
\end{document}

Edit: This should have been stated in the original formulation of the answer, but was overlooked.

The real problem with this use of align is that align expects and requires an odd number of &s to work properly.

Briefly, the purpose of align is to set adjacent equations so that a specified point in each (marked by &) is positioned in the same horizontal location in all lines of a multi-line display. Since more than one component equation may appear on each line, these components are in turn separated by &, hence the requirement for an odd number of them. Furthermore, align should not be used for single-line displays; that's what equation (or \[ ... \] for the unnumbered variety) is for. See the amsmath user guide for details (texdoc amsldoc).