1

I have a long expression (maximization problem) which I force down to the next line using //.

I also have two functions ("subject to") which I want to place underneath. I use & to make the expressions fit vertically.

Problem:

  • The third integral expression which I have to force down looks weird and isn't aligned with the two integral expression above. I don't know how to fix it.
  • The alignment is off. Fx "Another equation here" should be aligned with Max(n_G, 0) \leq n_{min} \leq Min(n_B,1). Using & obviously doesn't solve the problem.
  • Extra: The reason I use \def\mclimits_#1{\limits_{\mathclap{#1}}} is to make the long expression n_{min} fit in the limits of the integral.

Code:

\documentclass{article}
\usepackage[utf8]{inputenc}

\usepackage{mathtools}

\def\mclimits_#1{\limits_{\mathclap{#1}}}

\begin{document}

\begin{equation}
\begin{aligned}
\underset{n_{min}}{\text{argmax}} 
& & \Pi = \frac{1}{M}\[
\int\mclimits_{n_{min}}^M\! n_cP_c\, \mathrm{d}n_c
\] + \frac{p}{M}\[
\int\mclimits_{n_B}^M\! \delta\bigg(z\beta P_r\big(h(1-n_B)+(1-h)x_B\big)-D\bigg)\, \mathrm{d}n_c
\] \\
& & + \frac{1-p}{M}\[
\int\mclimits_{n_{min}}^M\! \delta\bigg(z\beta P_r\big(h(1-n_B)+(1-h)x_G\big)-D\bigg)\, \mathrm{d}n_c
\]  \\
\text{subject to}
& & Max(n_G, 0) \leq n_{min} \leq Min(n_B,1)\\
& &  \text{Another equation here}\\
\end{aligned}
\end{equation}

\end{document}

Current output:

enter image description here

  • Please make your example compilable. I get errors. – LaRiFaRi Jun 15 '16 at 9:39
  • Hmm, I ran it. I get errors too. But it compiles. – Saud Jun 15 '16 at 9:40
3

Your MWE has errors: you nested equation (\[ ... \]) in equation ... I try to figured out what is your problem. After cleaning of your code I obtain the following results:

enter image description here

Is this what you looking for? In your code I also added a split environment for the long first equation:

\documentclass{article}
\usepackage[utf8]{inputenc}
\usepackage{mathtools}
\DeclareMathOperator*{\argmax}{argmax} 

\begin{document}
\begin{align}
\begin{split}
\argmax_{n_{\min}}\Pi
    & = \frac{1}{M}\int\limits_{n_{\min}}^M\! n_cP_c\, \mathrm{d}n_c \\
    &\qquad   + \frac{p}{M}\int\limits_{n_B}^M\delta\bigg(z\beta P_r\big(h(1-n_B) + 
                (1-h)x_B\big)-D\bigg)\, \mathrm{d}n_c \\
    &\qquad   + \frac{1-p}{M}\int\limits_{n_{\min}}^M 
                \delta\bigg(z\beta P_r\big(h(1-n_B) + (1-h)x_G\big)-D\bigg)\, \mathrm{d}n_c
      \end{split}                                           \\
\text{subject to}
        &\ \max(n_G, 0) \leq n_{min} \leq \min(n_B,1)      \notag  \\
        &  \text{Another equation here}
\end{align}
\end{document}

Thanks to @egreg for the improvements he suggested in his comment.

  • Yes :) Split works great! I just removed \\ &\qquad between the first and second integral expression to place them on the same line, like in the original code. Just one small thing: Why is "Another equation here" slightly more to the left than the expression just above? – Saud Jun 15 '16 at 10:10
  • 1
    Because for expression above is shifted after ampersand to the right with \ . (otherwise it sticks with text before it). I assume that there ones wild be real equation, then this shift will looks slightly different. – Zarko Jun 15 '16 at 10:14
  • Ah, I see. Yes, will be replaced with a real equation. Thx again! – Saud Jun 15 '16 at 10:16
  • 2
    \DeclareMathOperator*{\argmax}{argmax} will allow typing \argmax_{n_{\min}\Pi instead of that awful concoction. – egreg Jun 15 '16 at 10:25
  • @egreg, yes, this is correct approach. I will consider this in eddit of my answer. – Zarko Jun 15 '16 at 10:29
4

I am not sure, if I got your right. Like this?

% arara: pdflatex

\documentclass{article}
\usepackage{blindtext}
\usepackage{mathtools}
\DeclareMathOperator{\argmax}{argmax}
\def\mclimits_#1{\limits_{\mathclap{#1}}}

\begin{document}
\blindtext  
\begin{align}
    \begin{split}
        \underset{n_{\min}}{\argmax}(\Pi) &= \frac{1}{M}\int\mclimits_{n_{\min}}^M n_c P_c\, \mathrm{d}n_c \\
        &\quad+ \frac{p}{M}\int\mclimits_{n_B}^M \delta\Bigl(z\beta P_r\bigl(h(1-n_B)+(1-h)x_B\bigr)-D\Bigr)\, \mathrm{d}n_c \\
        &\quad+ \frac{1-p}{M}
        \int\mclimits_{n_{\min}}^M \delta\Bigl(z\beta P_r\bigl(h(1-n_B)+(1-h)x_G\bigr)-D\Bigr)\, \mathrm{d}n_c
    \end{split}
        \shortintertext{subject to}
        &\max(n_G, 0) \leq n_{\min} \leq \min(n_B,1)\\
        &\text{Another equation here}
\end{align} 
\blindtext  
\end{document}

enter image description here

  • Yes, looks good :) – Saud Jun 15 '16 at 10:13
  • 1
    +1 for using \intertext. – Henri Menke Jun 15 '16 at 12:12
3

With ConTeXT MKIV

\setupmathlabeltext[argmax=argmax]
\definemathcommand[argmax][limop]{\mfunctionlabeltext{argmax}}
\starttext
\placeformula
\startformula
  \startmathalignment
    \NC \argmax_{n_{\min}}\Pi
    \NC = \frac{1}{M}\int\limits_{n_{\min}}^M\! n_cP_c\, \diff n_c \NR
    \NC\NC\quad + \frac{p}{M}\int\limits_{n_B}^M\delta\bigg(z\beta P_r\big(h(1-n_B) + (1-h)x_B\big)-D\bigg)\, \diff n_c \NR
    \NC\NC\qquad + \frac{1-p}{M}\int\limits_{n_{\min}}^M \delta\bigg(z\beta P_r\big(h(1-n_B) + (1-h)x_G\big)-D\bigg)\, \diff n_c \NR[+]
    \startintertext
      subject to
    \stopintertext
    \NC \max(n_G, 0) \NC \leq n_{\min} \leq \min(n_B,1) \NR
    \NC              \NC \text{Another equation here} \NR[+]
  \stopmathalignment  
\stopformula
\stoptext

enter image description here

  • What \quad in one line and \qquad in the other; won't \quad on both lines look better (aligned + signs) – Aditya Jun 16 '16 at 4:26
  • @Aditya This is a matter of taste, I guess. One could also leave out the horizontal space altogether to have = and + aligned. I'll leave it there, because I like it that way, but your proposed layout would look good as well. May the user decide. – Henri Menke Jun 16 '16 at 7:12
1

Using the geometry package, you have more sensible margins and it looks nicer with alignat:

\documentclass{article}
\usepackage[showframe]{geometry}
\usepackage{mathtools}
\DeclareMathOperator{\argmax}{argmax}
\def\mclimits_#1{\limits_{\mathclap{#1}}}

\begin{document}

\begin{alignat}{2}
    & \argmax_t{n_{\min}}(\Pi) &&=\!\begin{aligned}[t]\frac{1}{M}\int\mclimits_{n_{\min}}^M n_c P_c\, \mathrm{d}n_c + \frac{p}{M}\int\mclimits_{n_B}^M \delta\Bigl(z\beta P_r\bigl(h(1-n_B)+(1-h)x_B\bigr)-D\Bigr)\, \mathrm{d}n_c & \\
  + \frac{1-p}{M} \int\mclimits_{n_{\min}}^M \delta\Bigl(z\beta P_r\bigl(h(1-n_B)+(1-h)x_G\bigr)-D\Bigr)\, \mathrm{d}n_c &
  \end{aligned}\\
  & \text{subject to} \quad\mathrlap{\max(n_G, 0) \leq n_{\min} \leq \min(n_B,1)} \notag\\[1.5ex]
  & \rlap{Another equation here}
\end{alignat}

\end{document}

enter image description here

  • I think you mentioned before that this is due to your editor, but there are a lot of Unicode characters creeping into your answer, e.g. Π, β, and δ. Of course this would work if you used a Unicode engine in conjunction with unicode-math, but not with classic pdfTeX. – Henri Menke Jun 16 '16 at 7:15
  • Oh! yes. When it's late, I may forget to check. Thanks for pointing it! Hope it's all fixed now. – Bernard Jun 16 '16 at 8:27
  • BTW: Which editor are you using? – Henri Menke Jun 16 '16 at 8:30
  • WinEdt. I wrote a small package for it that does some sort of pretty-printing in order to have a an easier-to-understand displayed code in formulae (Greek letters, numerical exponents, a number of symbols). – Bernard Jun 16 '16 at 8:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.