How to move a block of equations inside an enumeration down?

Question

In the following figure, the equations are well aligned in displayed mode but not in enumerated environment. I have done many permutations but I give up.

enter image description here

\documentclass{article}
\usepackage[a4paper,margin=1cm]{geometry}
\usepackage{mathtools}

\def\inner{%
\begin{multlined}[b][7cm]
\framebox[5cm]{Top LHS}\\
\framebox[5cm]{Middle LHS}\\
\framebox[5cm]{Bottom LHS}
\end{multlined}
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 1}\\
            \framebox[5cm]{Middle RHS 1}\\
            \framebox[5cm]{Bottom RHS 1}
        \end{multlined}\\
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 2}\\
            \framebox[5cm]{Middle RHS 2}\\
            \framebox[5cm]{Bottom RHS 2}
        \end{multlined}}

\begin{document}
\section*{displayed}
\begin{align*}
\inner
\end{align*}

\section*{enumerated}
\begin{enumerate}
    \item
    $\!
    \begin{aligned}[t]  
    \inner
    \end{aligned}
    $
\end{enumerate}
\end{document}

How to move a block of equations inside an enumeration down?

It is a good puzzle, isn't it? I don't find any information from mathmode.pdf about this pattern as well. — kiss my armpit, Apr 17 '13 at 14:59
It's the expected behaviour of [b] alignment, you would get the same if you went \item \begin{tabular}[b]..\end{tabular}..\begin{tabular}[t]..\end{tabular} (which is what you are doing, effectively. — David Carlisle, Apr 17 '13 at 15:10

score 7 · Answer 1 · edited Jun 16 '17 at 16:18

if you change the

[...]
\def\inner{%
\begin{multlined}[b][7cm]
[...]

to

[...]
\def\inner{%
\begin{multlined}[t][7cm]
[...]

the first line will align nicely in the enumerate, and it doesn't seem to make any difference to the displayed version.

edit: okay, i lied (or actually, didn't pay enough attention to what was happening to the right hand side).

the following leaves the optional position of the first multlined as [t] (which lines up its first line with the enumerate's \item label. then it follows the first multlined by a line break, which needs to be adjusted (upward) to align properly. i think this is closer to what you had in mind.

\documentclass{article}

\usepackage[a4paper,margin=1cm]{geometry}
\usepackage{mathtools}

\def\inner{%
\begin{multlined}[t][7cm]
\framebox[5cm]{Top LHS}\\
\framebox[5cm]{Middle LHS\vphantom{p}}\\
\framebox[5cm]{Bottom LHS\vphantom{p}}
\end{multlined}\\[-1.3\baselineskip]  %%%  <-- here's the change
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 1}\\
                \framebox[5cm]{Middle RHS 1\vphantom{p}}\\
                \framebox[5cm]{Bottom RHS 1\vphantom{p}}
        \end{multlined}\\
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 2}\\
            \framebox[5cm]{Middle RHS 2\vphantom{p}}\\
            \framebox[5cm]{Bottom RHS 2\vphantom{p}}
        \end{multlined}}

\begin{document}
\section*{displayed}
\begin{align*}
\inner
\end{align*}

\section*{enumerated}
\begin{enumerate}
    \item
    $\!
    \begin{aligned}[t]  
    \inner
    \end{aligned}
    $
\end{enumerate}
\end{document}

while we're at it, let's make all the boxes the same depth (with \vphantom{p} since the "p" in the "Top" line has the descender that's making them uneven), and also even up the spacing around the equal signs. (i haven't checked, but i suspect there's a \, at the beginning of multlined -- there are some peculiar "initial" thin spaces in some of the amsmath multi-line structures, and it wouldn't surprise me if that was carried over into multlined.) since the visual space is tight on account of the boxes, i added the thin space at the end of the left-hand multlined because i think it looks nicer than removing space on the right of the equals sign.

output of example code

but the alignment for remaining output on the right hand side is not what I want. :-) — kiss my armpit, Apr 17 '13 at 14:46
+1 for making it look nice even if I think my way's better:-) — David Carlisle, Apr 17 '13 at 14:51
@DavidCarlisle: The alignment on RHS does not conform to my requirement. Do you notice it? Please see this screenshot. — kiss my armpit, Apr 17 '13 at 14:53
@Bugbusters -- sad to say, it's the result of experimentation, starting with -1\baselineskip. but if you have to do this sort of thing often enough, the answer usually converges pretty quickly. oh -- it may be different with different type sizes applied to a document class, or with different document classes. and it definitely changes if you use something like \large or \small to adjust a particular region of the text. — barbara beeton, Apr 17 '13 at 18:29

David Carlisle · Answer 2 · 2017-06-03T13:20:38.317

I think you should use a display environment not inline math for a display

enter image description here

\documentclass{article}
\usepackage[a4paper,margin=1cm]{geometry}
\usepackage{mathtools}

\def\inner{%
\begin{multlined}[b][7cm]
\framebox[5cm]{Top LHS}\\
\framebox[5cm]{Middle LHS}\\
\framebox[5cm]{Bottom LHS}
\end{multlined}
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 1}\\
            \framebox[5cm]{Middle RHS 1}\\
            \framebox[5cm]{Bottom RHS 1}
        \end{multlined}\\
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 2}\\
            \framebox[5cm]{Middle RHS 2}\\
            \framebox[5cm]{Bottom RHS 2}
        \end{multlined}}

\begin{document}
\section*{displayed}
\begin{align*}
\inner
\end{align*}

\section*{enumerated}
\begin{enumerate}
    \item\mbox{}\par\vspace{\dimexpr-2\baselineskip-\abovedisplayskip\relax}
\begin{align*}
\inner
\end{align*}
\end{enumerate}
\end{document}

As the alignment point for the top part is b TeX doesn't really know where the first row of the inner box is, however if you assume that it is normal sized you can make a position that works reasonably well.

    \item\makebox[.5\linewidth]{ZZZZZZZZZZZZZZZZZZZ}\par
\vspace{\dimexpr-2\baselineskip-\abovedisplayskip\relax}
\begin{align*}

produces:

enter image description here

However if the top row is over-sized (as it is in the example because of the framebox) the alignment moves down to compensate so the baseline of the top row doesn't align with the item number, however this is (probably?) inevitable.

the alignment isn't very nice in the first line of the enumerated example. (a change of b to t in the original is all that's needed; see my answer.) — barbara beeton, Apr 17 '13 at 14:43
@barbarabeeton I'd agree it looks better but I still think that putting an alignment in $..$ has to be wrong on principle. I probably have the negative spacing wrong I'd have to look what space actually is needed (rather than baselineskip and abovedisplayskip) to get the baselines more aligned clearly by adjusting that space anything could be aligned. — David Carlisle, Apr 17 '13 at 14:50
i'm not disagreeing about putting an alignment in $...$ . but it's not the first time this has been asked for, and having multiple "items" of this sort strung out one after the other gets very spacy and ugly. they really look better closed up. (probably using a subequations approach would be better though, as well as more logical.) — barbara beeton, Apr 17 '13 at 18:23

score 2 · Accepted Answer · edited Jun 10 '17 at 12:31

Mission accomplished!

enter image description here

\documentclass{article}
\usepackage[a4paper,margin=1cm]{geometry}
\usepackage{mathtools}

\begin{document}

\begin{enumerate}
\item
$\!
\begin{aligned}[t]  
\mathrlap{\framebox[5cm]{Top LHS}}\rule{7cm}{0pt}\\
\rule{3.5cm}{0pt}\mathclap{\framebox[5cm]{Middle LHS}}\rule{3.5cm}{0pt}\\
\rule{7cm}{0pt}\mathllap{\framebox[5cm]{Bottom LHS}}
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 1}\\
            \framebox[5cm]{Middle RHS 1}\\
            \framebox[5cm]{Bottom RHS 1}
        \end{multlined}\\
    &= \begin{multlined}[t][7cm]
            \framebox[5cm]{Top RHS 2}\\
            \framebox[5cm]{Middle RHS 2}\\
            \framebox[5cm]{Bottom RHS 2}
        \end{multlined}
\end{aligned}
$
\end{enumerate}
\end{document}

Edit:

With better user interfaces.

\documentclass[twocolumn]{article}
\usepackage[a4paper,margin=1cm]{geometry}
\usepackage{mathtools}
\newlength\EWD
\newlength\HWD
\newlength\HHWD

\def\Init{%
    \settowidth{\EWD}{${}={}$}%
    \setlength{\HWD}{0.5\dimexpr\linewidth-\EWD\relax}%
    \setlength{\HHWD}{0.5\HWD}}

\def\TopLHS#1{\mathrlap{#1}\rule{\HWD}{0pt}}
\def\MiddleLHS#1{\rule{\HHWD}{0pt}\mathclap{#1}\rule{\HHWD}{0pt}}
\def\BottomLHS#1{\rule{\HWD}{0pt}\mathllap{#1}}

\begin{document}

\begin{enumerate}
\item\Init
$\!
\begin{aligned}[t]  
\TopLHS{\framebox[3cm]{Top LHS}}\\
\MiddleLHS{\framebox[3cm]{Middle LHS}}\\
\BottomLHS{\framebox[3cm]{Bottom LHS}}
    &= \begin{multlined}[t][\HWD]
            \framebox[3cm]{Top RHS 1}\\
            \framebox[3cm]{Middle RHS 1}\\
            \framebox[3cm]{Bottom RHS 1}
        \end{multlined}\\
    &= \begin{multlined}[t][\HWD]
            \framebox[3cm]{Top RHS 2}\\
            \framebox[3cm]{Middle RHS 2}\\
            \framebox[3cm]{Bottom RHS 2}
        \end{multlined}
\end{aligned}
$
\end{enumerate}
\end{document}

Note: I don't know the best place at which I should invoke the \Init. If you have any suggestion, please kindly let me know.

yes +1 putting a rule at the top of a box to fake a second alignment point is a standard trick, I still claim that putting an alignment in text math is the wrong thing to do, but since in this case it's hard to make the right markup do the right thing I won't complain too much:-) — David Carlisle, Apr 18 '13 at 20:21

How to move a block of equations inside an enumeration down?

3 Answers3

Edit: