# Long equations breaking without using breqn

How to achieve nice (non-automatic) long equations breaking, like dmath environment from breqn package does automatically, but using a more common environment, e.g., multline, multlined (from mathtools), or split?

Compare dmath results with the rest. (Note that page size is dictated by the journal requirements.)

\documentclass[10pt,twocolumn]{article}
\usepackage[letterpaper,textwidth=18.5cm,textheight=22.9cm]{geometry}
\usepackage{mathtools}
\usepackage{breqn}

\DeclareMathOperator{\Real}{Re}
\let\oldhat\hat
\renewcommand{\vec}[1]{\mathbf{#1}}
\renewcommand{\hat}[1]{\oldhat{\mathbf{#1}}}
\newcommand{\del}{\boldsymbol{\nabla}}
\newcommand{\taub}{\boldsymbol{\tau}}

\begin{document}

\textbf{dmath:}
\begin{dmath}
\vec{U_{0}} = \left[ H^{-1} \hat{k} \times \del \left( fH^{-1} \right) \right]
\Real \left\{ \, \ointop_{fH^{-1}} \left( \rho^{-1} H^{-1} \taub \right) \cdot d \vec{l}
\div \ointop_{fH^{-1}} H^{-1} \left( H^{-1} R+ i \omega \right)
\left( \del \left( fH^{-1} \right) \cdot \vec{n} \right) \, d \vec{l} \right\}.
\end{dmath}

\textbf{multline:}
\begin{multline}
\vec{U_{0}} = \left[ H^{-1} \hat{k} \times \del \left( fH^{-1} \right) \right]
\Real \left\{ \, \ointop_{fH^{-1}} \left( \rho^{-1} H^{-1} \taub \right) \cdot d \vec{l} \right. \\
\div \left. \ointop_{fH^{-1}} H^{-1} \left( H^{-1} R+ i \omega \right)
\left( \del \left( fH^{-1} \right) \cdot \vec{n} \right) \, d \vec{l} \right\}.
\end{multline}

\textbf{multlined:}
$$\begin{multlined} \vec{U_{0}} = \left[ H^{-1} \hat{k} \times \del \left( fH^{-1} \right) \right] \Real \left\{ \, \ointop_{fH^{-1}} \left( \rho^{-1} H^{-1} \taub \right) \cdot d \vec{l} \right. \\ \div \left. \ointop_{fH^{-1}} H^{-1} \left( H^{-1} R+ i \omega \right) \left( \del \left( fH^{-1} \right) \cdot \vec{n} \right) \, d \vec{l} \right\}. \end{multlined}$$

\textbf{split:}
$$\begin{split} \vec{U_{0}} &= \left[ H^{-1} \hat{k} \times \del \left( fH^{-1} \right) \right] \Real \left\{ \, \ointop_{fH^{-1}} \left( \rho^{-1} H^{-1} \taub \right) \cdot d \vec{l} \right. \\ &\phantom{=} \div \left. \ointop_{fH^{-1}} H^{-1} \left( H^{-1} R+ i \omega \right) \left( \del \left( fH^{-1} \right) \cdot \vec{n} \right) \, d \vec{l} \right\}. \end{split}$$

\end{document}

-

I believe that the result can be improved:

\documentclass[10pt,twocolumn]{article}
\usepackage[letterpaper,textwidth=18.5cm,textheight=22.9cm]{geometry}
\usepackage{mathtools,bm}

\DeclareMathOperator{\Real}{Re}
\renewcommand{\vec}[1]{\mathbf{#1}}
\newcommand{\bhat}[1]{\hat{\mathbf{#1}}}
\newcommand{\del}{\bm{\nabla}}
\newcommand{\taub}{\bm{\tau}}

\begin{document}

$$\begin{multlined} \vec{U_{0}} = \bigl[H^{-1} \bhat{k} \times \del ( fH^{-1})\bigr] \Real \Biggl\{\ointop_{\,fH^{-1}}\!\! (\rho^{-1} H^{-1} \taub ) \cdot d \vec{l} \\ {}\div\ointop_{\,fH^{-1}}\!\! H^{-1} (H^{-1} R+ i \omega ) ( \del ( fH^{-1} ) \cdot \vec{n} ) \, d \vec{l} \Biggr\}. \end{multlined}$$

\end{document}


Notice that I defined \bhat, rather than redefining \hat (which I would never do); redefining \vec is fine. Notice also that I used no \left and \right commands. The braces results in a much nicer size and the spacing around parentheses is correct.

Final touches: \, in the subscript to the integral and \!\! to move the integrand a bit to the left.

The \boldsymbol command is obsolete and should be changed into \bm loading the bm package.

-
To reproduce the dmath output more closely, I would still go for a split. Additionally, when going from \left\{ to \Biggl\{ there is too little space after the \Real. –  mafp Dec 17 '12 at 16:31
It's questionable whether dmath gives the best result. I don't think so. There should be no space between an operator and a fence. –  egreg Dec 17 '12 at 16:32