3

Using the MWE

\documentclass[fontsize=8pt, a4paper, DIV12]{scrbook}

\usepackage{amsmath}

\begin{document}
\[ e^t = x^{(5)} - 2x^{(3)} + \dot{x}
 = \sum_{k = 0}^{5} \binom{5}{k} p^{(k)}(t) e^t
 -2\sum_{k = 0}^{3} \binom{3}{k} p^{(k)}(t) e^t
 + \sum_{k = 0}^{1} \binom{1}{k} p^{(k)}(t) e^t
\]
\end{document}

I get the output

Formula with strange looking binomial coefficients

with strange looking binomial coefficients. Root cause is the font size.

Is there any way to get the usual look without changing font size?

arney
  • 2,023
  • 3
    What about using \tbinom instead of \binom? – Werner Oct 31 '14 at 20:21
  • I'm afraid not: it's an inherent problem with generalized fractions. – egreg Oct 31 '14 at 20:30
  • @Werner Noteworthy commands, but they yield the same result. – arney Oct 31 '14 at 20:33
  • @arney: Surely they don't yield the same result... your reference to "strange looking" is subjective then. – Werner Oct 31 '14 at 20:35
  • trying this out with three different sizes -- nominal, \small and \large -- and also with amsart using the same three sizes, i see that with scrbook, the parentheses, except with \large, the parentheses fail to "cover" the contents. using amsart, the parentheses are consistently tall enough to encompass both elements fully, and the result is much less objectionable. i also tried it with the (not well known} 9pt option of amsart, and in that test, the parens on the \small instance are undersized. so i think it's attributable to the size, as already suggested. – barbara beeton Oct 31 '14 at 20:55
  • @barbarabeeton Thanks for the indepth analysis. As I'm already over 150 pages, I'd rather not change font size now. As the strange look of the binomial gives it a clear visual distinction from 2D vectors (i.e. pmatrix), I might as well keep it. – arney Oct 31 '14 at 21:32
  • 1
    Possibly related: http://tex.stackexchange.com/questions/137141/strange-behaviour-of-binomial-coefficients-delimiters – Steven B. Segletes Oct 31 '14 at 21:49
  • 1
    @arney Sorry for sounding harsh, but setting font size to 8pt isn't the way to reduce the number of pages. I know that this doesn't mean the "bug" in amsmath is not a bug, but you can't expect setting a ridiculous font size and expecting no problems of this type. – yo' Oct 31 '14 at 23:42

1 Answers1

3

If you use the \mbinom command, from the nccmath package (medium-sized \binom, ~80 % of display style), it looks much better, in my opinion:

\documentclass[fontsize=8pt, a4paper, DIV12]{scrbook}

\usepackage{amsmath, nccmath}

\begin{document}

\[ e^t = x^{(5)} - 2x^{(3)} + \dot{x}
 = ∑_{k = 0}⁵ \mbinom{5}{k} p^{(k)}(t) e^t
 -2∑_{k = 0}³ \mbinom{3}{k} p^{(k)}(t) e^t
 + ∑_{k = 0}¹ \mbinom{1}{k} p^{(k)}(t) e^t
\]

\end{document} 

enter image description here

Bernard
  • 271,350
  • Well, I don't like this. Either you keep the numbers full size, or you make them smaller and bring them closer... – yo' Oct 31 '14 at 23:40
  • @tohecz: That's a matter of personal taste. I don't like very much the sizing of \tfrac in a context of display style elements, as the difference is too big. My opinion is that 80 % of display style is a sensible compromise, but it's only my opinion… – Bernard Oct 31 '14 at 23:47
  • There is a good reason why to have font sizes in discrete steps (usually 1.2 times), so that you clearly know of which font size is this or that part of the text or equation. So ok, you choose one step down for the binom, but then why the interline space that is almost one full baselineskip? And with \frac increasing the spacing is even worse, because you suddenly have the numerator and denominator floating in the middle of nowhere. But it's your choice to do that. – yo' Oct 31 '14 at 23:54
  • @tohecz: I usually have no change in \baselineskip because, as I also think the default is a little too tight compared to French usage, I use \setstretch{1.05} or so. But I do think it's a cultural habit. – Bernard Nov 01 '14 at 00:30
  • Have to add my usual warning about nccmath: Is there a replacement for nccmath?. – Peter Grill Nov 01 '14 at 00:53
  • @Peter Grill: I remember we had a discussion about problems with nccmath (that I personally never had, but I have a sparse use of it). Also think I found a possible solution for it, from another thread, but I can't remember which it was. Did you keep some trace of this discussion? – Bernard Nov 01 '14 at 01:03
  • @Bernard Sorry, the mentioning of baselineskip wasn't meant that way. I only wanted to say: You have the top and the bottom number miles away from each other. For no reason at all. – yo' Nov 01 '14 at 01:09
  • @Bernard: Sorry don't remember. As I had moved on from nccmath a long time ago I haven't really considered going back. – Peter Grill Nov 01 '14 at 01:13
  • @Bernhard Thanks! I like that fontsize adjustment a lot. And since I avoid parboxes and minipages, I should be able to run it seamlessly. – arney Nov 04 '14 at 15:03