# What is the correct way to properly align a split equation with continued parts prefixed with a binary operator?

Consider the following result with 2 methods. Don't hesitate to propose better ones if any.

The question is: What is the correct way to properly align a split equation in which binary operator appears at the beginning of the continued parts?

\documentclass[preview,border=12pt]{standalone}
\usepackage{mathtools}
\begin{document}
$\begin{split} \text{LHS} &= c + d\\ &\quad +e +f +g \end{split}$

\section*{phantom}
$\begin{split} \text{LHS} &= c + d\\ &\hphantom{{}=} +e +f +g \end{split}$
\end{document}

-
The first thing I would think is: is there a proper alignment? (Not talking about how to align, but rather the point of alignment.) I don't think so. Unlike manye equations where you can easily think that aligning at the = is right, here you can align at many different (and no so defined) points. – Manuel Jun 19 '14 at 22:02
The only rule I know says that you shouldn't align with the left side of the equal sign (tex.stackexchange.com/a/172110). So Manuel is right. There could be many points. I also use your style (vert. aligned right side of the =) and therefore would tend to the second version. This version yields a perfect alignment. Can't imagine that \quad always will. – LaRiFaRi Jun 19 '14 at 22:33
agreed that there's no definitive rule for such alignment. knuth uses \qquad after the alignment point for a continued formula (texbook, page 195). mathematics into type shows several possibilities, including those already mentioned (pages 46ff.). – barbara beeton Jun 20 '14 at 13:50

I prefer one of these 3 variants, that use the amsmath environments (the second one, not very convincing here, might be of interest for longer formulae):

\documentclass[preview,border=12pt]{standalone}
\usepackage{mathtools}

\begin{document}

\section*{variant 1:}
\begin{align*}
\text{LHS} &= \begin{aligned}[t]
c & + d\\
& +e + f +g
\end{aligned}
\end{align*}

\section*{variant 2:}
\begin{align*}
\text{LHS} & = \begin{multlined}[t]
c + d\\
+e +f +g
\end{multlined}
\end{align*}

\section*{variant 3:}
\begin{alignat*}{2}
& \text{LHS} & {} = c & + d \\
& & &+ e + f +g
\end{alignat*}

\end{document}


-
In the third one you ar missing a {}. And in the two others I would add \! before the …ed environment. – Manuel Jun 19 '14 at 22:03
Thanks, Manuel! You're right, I didn't check enough. Modified the answer. The funny point is that in another (not totally satisfactory) variant, I did have to add a {} in front of a binary operator… – Bernard Jun 19 '14 at 22:13
The code is not synced with the output image, I think. – kiss my armpit Jun 19 '14 at 22:49
You're right. I thought I had loaded the new output image, but I didn't for some reason (probably forgot to validate my upload°). Now it's sync'ed. – Bernard Jun 19 '14 at 22:54
Right, thanks. I think now it's OK. – Bernard Jun 19 '14 at 23:21