How can I preserve vertical spacing around an equation in a minipage?

Question

The following example:

\documentclass{standalone}
\begin{document}
% outer minipages are not part of the test, and just demonstrate the inconsistency
\begin{minipage}[t]{0.5\textwidth}
This is an equation
\begin{equation}
    x = 1,
\end{equation}
which ends in a comma.


\end{minipage}%
\begin{minipage}[t]{0.5\textwidth}
This is a minipage equation\par
\begin{minipage}[t]{\textwidth}
    \begin{equation}
        x = 1,
    \end{equation}
    \par\nointerlineskip\xdef\tpd{\the\prevdepth}
\end{minipage}%
\par\prevdepth\tpd\noindent
which ends in a comma.


\end{minipage}%
\end{document}

where the prevdepth stuff is taken from https://tex.stackexchange.com/a/34982/41112, renders as

How can I make the column on the right match the one on the left, without removing the minipage? Note that the $x=1$ line is also misaligned, though only very slightly.

Edit: It turns out this is an XY problem, and the solution to my question above doesn't actually work for real problem. That is:

\documentclass{article}
\usepackage{amsmath}
\usepackage{xcolor}
\begin{document}
This is some text that is long enough to wrap across multiple lines, or at least it ought to be. This gap is too big:\par\nointerlineskip\noindent
\begin{minipage}[t]{.5\textwidth}%
    \begin{subequations}\label{eq:tensor_tensor_tensor_left}%
    \begin{alignat}{2}%
    &
        (M \otimes_{\color{red} R} N) \otimes_{\color{blue} A} (P \otimes_{\color{red} R} Q) \nonumber \
    \cong_{\color{blue} A} {}&
        ((M \otimes_{\color{red} R} N) \otimes_{\color{blue} A} P) \otimes_{\color{red} R} Q) \label{eq:ttt-left-assoc-1}
    \end{alignat}%
    \end{subequations}%
\end{minipage}%
\begin{minipage}[t]{.5\textwidth}%
    \begin{subequations}\label{eq:tensor_tensor_tensor_right}%
    \begin{alignat}{2}%
    &
        (M \otimes_{\color{red} R} N) \otimes_{\color{blue} A} (P \otimes_{\color{red} R} Q) \nonumber \
    \cong_{\color{blue} A} {}&
        M \otimes_{\color{red} R} (N \otimes_{\color{blue} A} (P \otimes_{\color{red} R} Q) \label{eq:ttt-right-assoc-1}
    \end{alignat}%
    \end{subequations}%
  \par\xdef\tpd{\the\prevdepth}%
  \vspace{\belowdisplayskip}\vspace{-\baselineskip}%
  \end{minipage}\par\prevdepth=\tpd\noindent
and some more unindented text. The gap above is too small. Toegther they add to the right amount.
\begin{alignat}{2}%
&
    (M \otimes_{\color{red} R} N) \otimes_{\color{blue} A} (P \otimes_{\color{red} R} Q) \nonumber \
\cong_{\color{blue} A} {}&
    M \otimes_{\color{red} R} (N \otimes_{\color{blue} A} (P \otimes_{\color{red} R} Q) \label{eq:ttt-right-assoc-1}
\end{alignat}%
These two gaps are just right, and is a bit bigger than the one above
\end{document}

which gives:

This is a completely different problem than the quoted question. — egreg, Jan 15 '24 at 21:33
@egreg: To my untrained eye they look the same, but I guess the type of skip around equation is unrelated to the baselineskip in that question? — Eric, Jan 15 '24 at 21:35
Exactly so. And the problem is exacerbated by the nested minipage. Perhaps it's better if you explain your motivation. — egreg, Jan 15 '24 at 21:47
The actual motivation is to put two sets of equations (and their numbering) -by-side without affecting the vertical spacing — Eric, Jan 15 '24 at 22:21
I suspected something like that. I believe that the problem has already been answered on the site. For instance https://tex.stackexchange.com/a/550010/4427 — egreg, Jan 15 '24 at 22:23
Unfortunately the approach in that question doesn't work for me; it turns out I have subequations environments in my minipages, not equation; and for whatever reason that seems to break the labels — Eric, Jan 15 '24 at 23:12
This question is relevant, but doesn't address the entire problem: Spurious space above align environment at top of page/minipage — barbara beeton, Jan 16 '24 at 00:30

egreg · Answer 1 · 2024-01-16T12:56:52.410

Let me modify your example, so the second outer minipage is just to show the effect side-by-side with the first minipage. In the second minipage, the first line and the equation are in a minipage and the text that follows is outside.

\documentclass{article}
\begin{document}
% outer minipages are not part of the test, and just demonstrate the inconsistency
\noindent
\begin{minipage}[t]{0.5\textwidth}
  This is an equation
  \begin{equation}
    x = 1,
  \end{equation}
  which ends in a comma.
\end{minipage}%
\begin{minipage}[t]{0.5\textwidth}
  \begin{minipage}[t]{\linewidth}
    This is a minipage equation
        \begin{equation}
            x = 1,
        \end{equation}
  \par\xdef\tpd{\the\prevdepth}%
  \vspace{\belowdisplayskip}\vspace{-\baselineskip}
  \end{minipage}\par\prevdepth=\tpd
   which ends in a comma.
\end{minipage}
\end{document}

However this could fail in several ways, because the skip after the equation might be \belowdisplayshortskip and a rather complex routine would be needed to ascertain it.

About the real problem, I suggest to manage along the lines of https://tex.stackexchange.com/a/550010/4427

Prefer \mathcolor{red}{R} to \color{red} R.

\documentclass{article}
\usepackage{amsmath}
\usepackage{xcolor}
\newcommand{\sidebysidesubequations}[5][0.5]{%
  $$% Yes! We're consciously doing it
  \begin{minipage}{#1\displaywidth}%
  \setlength{\abovedisplayskip}{0pt}%
  \begin{subequations}\label{#2}
  \noindent#3
  \end{subequations}
  \end{minipage}
  \begin{minipage}{\dimeval{\displaywidth-#1\displaywidth}}
  \setlength{\abovedisplayskip}{0pt}%
  \begin{subequations}\label{#4}
  \noindent#5
  \end{subequations}
  \end{minipage}
  $$%
}
\begin{document}
This is some text that is long enough to wrap across multiple lines, or 
at least it ought to be.
\sidebysidesubequations{eq:tensor_tensor_tensor_left}{
  \begin{align}
    & (M \otimes_{\mathcolor{red}{R}} N)
    \otimes_{\mathcolor{blue}{A}}
    (P \otimes_{\mathcolor{red}{R}} Q) \nonumber \
  \cong_{\mathcolor{blue}{A}} {}&
    M \otimes_{\mathcolor{red}{R}}
    (N \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \label{eq:ttt-left-assoc-1}
 \
    & (M \otimes_{\mathcolor{red}{R}} N)
    \otimes_{\mathcolor{blue}{A}}
    (P \otimes_{\mathcolor{red}{R}} Q) \nonumber \
  \cong_{\mathcolor{blue}{A}} {}&
    M \otimes_{\mathcolor{red}{R}}
    (N \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \label{foo1}
  \end{align}
}{eq:tensor_tensor_tensor_right}{
  \begin{align}
    & (M \otimes_{\mathcolor{red}{R}} N)
    \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \nonumber \
  \cong_{\mathcolor{blue}{A}} {}&
      M \otimes_{\mathcolor{red}{R}}
  (N \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \label{eq:ttt-right-assoc-1}
  \
    & (M \otimes_{\mathcolor{red}{R}} N)
    \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \nonumber \
  \cong_{\mathcolor{blue}{A}} {}&
      M \otimes_{\mathcolor{red}{R}}
  (N \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \label{foo2}
  \end{align}
}
and some more unindented text
and some more unindented text
and some more unindented text
and some more unindented text.
\begin{align}
  & (M \otimes_{\mathcolor{red}{R}} N)
    \otimes_{\mathcolor{blue}{A}} (P \otimes_{\mathcolor{red}{R}} Q) \nonumber \
  \cong_{\mathcolor{blue}{A}} {}&
    M \otimes_{\mathcolor{red}{R}}
    (N \otimes_{\mathcolor{blue}{A}}
    (P \otimes_{\mathcolor{red}{R}} Q) \label{eq:ttt-right-assoc-2}
  \end{align}
These two gaps are just right, and is a bit bigger than the one above
\end{document}

Unfortunately this didn't quite work for my real example. I've update my question, let me know if you think I should ask it as a new question. Thanks! — Eric, Jan 15 '24 at 23:34
"In the second minipage, the first line and the equation are in a minipage and the text that follows is outside." - oh, I completely missed this on first reading! Unfortunately I don't think this works for my two-column scenario — Eric, Jan 15 '24 at 23:36
Thanks! I really do want separate subequation numbering for the left and right column. In my real version there is more than one labeled line on each side, but I figured that wasn't relevant to my layout concern — Eric, Jan 16 '24 at 08:37
@Eric Added a new implementation. Of course, there can be no guarantee whatsoever on the inner alignments. — egreg, Jan 16 '24 at 12:57
Superimposing the first half of your screenshot onto the second half shows that the spacings are not consistent: https://i.stack.imgur.com/tL3ME.png; the space at the top of the first half is too big, while the one at the bottom of the first half is too small. — Eric, Jan 17 '24 at 11:54

How can I preserve vertical spacing around an equation in a minipage?

1 Answers1

Linked