Long summation symbol

Question

I need to put a long expression under summation symbol in single row. For example like this $(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})$ . How can I get a long summation symbol? Long but not tall. Minimal working example is

    \documentclass{article}
    \begin{document}
    \[\sum_{(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})}\]
    \end{document}

I want summation symbol have same length as string under it. Something like this

But, maybe, I ask something unusual and unapplicable.

Welcome to TeX.SX! Could you please turn your code fragment into a minimal working example (MWE) that illustrates your problem. Reproducing the problem and finding out what the issue is will be much easier when we see compilable code, starting with \documentclass{...} and ending with \end{document}. — leandriis, Jun 10 '18 at 14:32
I don't understand what you mean by long summation symbol, please clarify your question. If you think that could help draw it on a paper and post the photo. — gvgramazio, Jun 10 '18 at 15:40
What does \mathcal{P}(\overline{1,n}) signify? Depending on its meaning, a shorter expression may be available/desirable. — Mico, Jun 10 '18 at 15:59
I would absolutely avoid to use a symbol like the one you did in the question. — gvgramazio, Jun 10 '18 at 18:45
If you just want to sum the i values, then maybe \[\sum_{k=1}^{n}i_k\in\mathcal{P}(\overline{1,n})\] would be suitable? It might help (as @Mico also mentioned) to know a bit more about the mathematical context to know which notation would be useful here. — Marijn, Jun 10 '18 at 19:13
It can happen, when writing on the blackboard or on paper, to make longer summation signs for clarity, but this doesn't mean that reproducing this on printed material is advisable: the result, as you clearly see, is very dubious. Either live with the overlong subscript under a standard symbol, or try and split the subscript as suggested by Mico. You get something like an alligator trying to assault its prey. — egreg, Jun 10 '18 at 20:40
While I think the desired output looks quite terrible and would strongly advise against it (have you ever seen this in a printed work?), I don't think that the three downvotes are deserved. Especially after the edits the question is clear in its intent and I don't think lack of research is a problem here. — moewe, Jun 10 '18 at 21:05

score 15 · Answer 1 · answered Jun 10 '18 at 16:18

15

Instead of having one fairly long line below the summation symbol, you may want to insert a line break. This can be done with the help of the \substack macro of the amsmath package.

Observe that it isn't really necessary to surround i_1,\dots,i_n with round parentheses.

\documentclass{article}
\usepackage{amsmath} % for "\substack" macro
%% decicated macro "\Pset" -- note the extra spacing around "\overline{...}`
\newcommand\Pset{\mathcal{P}(\mkern1.5mu \overline{1,n} \mkern1.5mu)}

\begin{document}
\[
\sum_{\substack{i_1,\dots,i_n\\ \in\,\Pset}}
\]
\end{document}

answered Jun 10 '18 at 16:18

Mico

506,678

I think it is unacceptable. – Alex E. Jun 10 '18 at 16:47
4

@AlexE. - do please state what is unacceptable. – Mico Jun 10 '18 at 16:56
Line break before set membership symbol. – Alex E. Jun 10 '18 at 17:50
Of course it depends on the meanings of the symbols involved, but in general $(x_1,\dots,x_n)\in X$ is not equivalent to $x_1,\dots,x_n\in X$ . The former usually denotes one single element (implicitly named, say $x$ ) of $X$ whose coordinates are $x_1$ to $x_n$ ; the latter denotes $n$ different elements $x_1$ to $x_n$ of $X$ . I wouldn't exactly say that a break before \in is unacceptable, but it is certainly ... odd. FWIW my first though was something like \sum_{\substack{i \in\,\Pset\\i=(i_1,\dots,i_n)}} but that is not really better in any sensible meaning of the word. – moewe Jun 10 '18 at 21:07
1

Anyway \substack is a very sensible idea here, so +1. – moewe Jun 10 '18 at 21:19
@moewe - Notational conventions tend to differ considerably across fields, don't they? In my field (econometrics), leaving off the parentheses around i_1,\dots,i_n would be quite alright -- as would be using the \substack device... However, as you point out, people in other fields may well freak out over the difference between i_1,\dots,i_n and (i_1,\dots,i_n). – Mico Jun 10 '18 at 21:21
1

Interesting, I'm so used to $(x_1,\dots,x_n)\in X$ being a single element of $X$ and $x_1,\dots,x_n\in X$ being n elements that I did not seriously think that notation could be that different in at least this context. How would you distinguish between picking one vector $x$ with coordinates $x_1,\dots,x_n$ from $\mathbb{R}^n$ and picking n vectors $x_1,\dots,x_n$ from $\mathbb{R}^n$ ? Purely by other notational conventions, like lower-case letters are always scalar and can't be vectors or using different positions for the index? – moewe Jun 11 '18 at 07:24
@moewe - I had interpreted the OP's setup as a summation over some items which are indexed by the scalars i_1, i_2, ..., i_n, where each scalar is an element of the set \mathcal{P}; naturally, this implies that the elements of \mathcal{P} must be scalars. That's why I suggested leaving off the parentheses around i_1,\dots,i_n. I fully agree with you that leaving off the parentheses would be inappropriate if i_1,\dots,i_n were an n-tuple instead of a collection of scalars. Without further input from the OP, though, it's probably not possible to come up with a definitive ruling. – Mico Jun 11 '18 at 07:57

Manuel · Accepted Answer · 2018-06-11T20:51:07.653

13

Don't do this.

Not making it intelligent on purpose. But it could be made intelligent.

\documentclass{article}

\usepackage{adjustbox,expl3,etoolbox}

\letcs\replicate{prg_replicate:nn}

\newcommand*\longsum[1][1]{%
  \mathop{\textnormal{%
    \clipbox{0pt 0pt {.5\width} 0pt}{$\displaystyle\sum$}%
    \replicate{#1}{\clipbox{{.5\width} 0pt {.4\width} 0pt}{$\displaystyle\sum$}}%
    \clipbox{{.6\width} 0pt 0pt 0pt}{$\displaystyle\sum$}}}%
}

\begin{document}

\[
  \longsum[31]_{(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})}
\]

\end{document}

Here you use \longsum_{like always} but it has an optional argument which repeats a part of the sum to be longer. In your particular case \longsum[31]_{..} seems to do.

Oh, or may be what you want is to reduce space around the \sum symbol? You can do that automatically with \smashoperator, see How to extend the text under summation symbol without making extra space, or manually for middle grounds with \kern-1em \sum_{long expression} \kern-1em (adjust -1em to the size you want).

edited Jun 11 '18 at 20:51

answered Jun 10 '18 at 20:03

Manuel

27,118

4

technical: ok (+1), aesthetic: awfull ... – Zarko Jun 10 '18 at 20:26
2

Nice, but it seems like the space is not optimally used - maybe the expression can be put inside of the sum? – Marijn Jun 10 '18 at 20:46
@Marijn I'm not sure what you mean. – Manuel Jun 10 '18 at 20:53
1

@Manuel the summation symbol now sort of looks like an empty frame, so the (i_1, etc. could go in the middle next to the >shaped part of the sigma - it wasn't a fully serious suggestion :) – Marijn Jun 10 '18 at 21:04
@Marijn Ah, okey. – Manuel Jun 10 '18 at 21:07
4

+1. Good luck on getting a paper that contains this symbol accepted in any journal that exceeds even a minimal quality threshold and whose editors don't want to become the laughingstock of the entire academic community... – Mico Jun 10 '18 at 21:16
1

@Mico I actually don't dislike the symbol itself, but in the same vein as what Marijn says, it should be its own notation, not using it for \sum. – Manuel Jun 10 '18 at 21:47

gvgramazio · Answer 3 · 2018-06-16T20:42:16.370

Are you looking for something like this?

\documentclass{article}
\begin{document}
\[\sum_{(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})}\]
\end{document}

Or this?

\documentclass{article}
\begin{document}
\(\sum_{(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})}\)
\end{document}

Edit

I'm wondering... Are you bothered by the length of the subscript because of the white space between the sum and the next variable? Because if this is the case, then the command \mathclap from mathtools allow to overlap the subscript to the sum symbol. The difference with respect to the original is shown in the following image.

And this is the code to reproduce it:

\documentclass[convert]{article}
\usepackage{mathtools}
\begin{document}
\[\sum_{i_1,\ldots,i_n\in\mathcal{P}(\overline{1,n})}x_i\]
\[\sum_{\mathclap{i_1,\ldots,i_n\in\mathcal{P}(\overline{1,n})}}x_i\]
\end{document}

The use that code, maybe without the parenthesis. But, and I say this in your interests, don't go for stretched summation symbols or something like that. Another way could be to rewrite the expression to make it shorter or to use a different font-size. — gvgramazio, Jun 10 '18 at 18:50

score 3 · Answer 4 · answered Jun 10 '18 at 17:43

In case you want to stretch the summation symbol horizontally, you can use the \hstretch command from the scalerel package. You can measure the width of the summation symbol and the subscript using \settowidth and use pgf to calculate the scaling factor that needs to be supplied to \hstretch. Note that the widths are not fully consistent with the actual width of the symbol so you have to adjust the factor (by 0.6 in the example below). The summation itself is displayed using \underset from the amsmath package.

MWE:

\documentclass{article}
\usepackage{scalerel}
\usepackage{pgf}
\usepackage{amsmath}
\begin{document}

\def\mysubscript{(i_1,\ldots,i_n)\in\mathcal{P}(\overline{1,n})}

\newlength\mylen
\settowidth\mylen{${}_\mysubscript$}

\newlength\mysumlen
\settowidth\mysumlen{$\sum$}

\pgfmathsetmacro\scalefactor{\mylen / \mysumlen * 0.6}

\noindent width of subscript: \the\mylen\\
width of sum: \the\mysumlen\\
scale factor: \scalefactor\\


\[\underset{\mysubscript}{\hstretch{\scalefactor}{\sum}}\]

\end{document}

Result:

Note that the result is aesthetically rather questionable and I would advise against using it in any situation whatsoever.

Yes, I want to stretch, but not all summation symbol, I want to stretch only top and bottom bar of it. — Alex E., Jun 10 '18 at 18:04
@AlexE. That's ... not really how it works since the summation symbol is a character in a font. You could redraw the whole symbol using something like TikZ but it'd still be really ugly and out of proportion — Au101, Jun 10 '18 at 18:28
+1 -- for reminding readers (just in case they needed a reminder...) that the old saying that "the customer is always right" isn't always correct. — Mico, Jun 10 '18 at 18:33

Long summation symbol

4 Answers4

Edit