I am using tikz to draw Feynman diagrams, and I learned how to draw a curved snake path.
The problem is that in general line segments are present (I'd like to avoid them).
I found a couple of post with could help to solve the problem, but none of them is working properly for me. The post are:
The second method results in the figure below,
\documentclass{beamer}
\usepackage{tikz}
\usetikzlibrary{decorations.pathreplacing,decorations.markings,snakes}
\newif\ifstartcompletesineup
\newif\ifendcompletesineup
\pgfkeys{
/pgf/decoration/.cd,
start up/.is if=startcompletesineup,
start up=true,
start up/.default=true,
start down/.style={/pgf/decoration/start up=false},
end up/.is if=endcompletesineup,
end up=true,
end up/.default=true,
end down/.style={/pgf/decoration/end up=false}
}
\pgfdeclaredecoration{complete sines}{initial}
{
\state{initial}[
width=+0pt,
next state=upsine,
persistent precomputation={
\ifstartcompletesineup
\pgfkeys{/pgf/decoration automaton/next state=upsine}
\ifendcompletesineup
\pgfmathsetmacro\matchinglength{
0.5*\pgfdecoratedinputsegmentlength / (ceil(0.5* \pgfdecoratedinputsegmentlength / \pgfdecorationsegmentlength) )
}
\else
\pgfmathsetmacro\matchinglength{
0.5 * \pgfdecoratedinputsegmentlength / (ceil(0.5 * \pgfdecoratedinputsegmentlength / \pgfdecorationsegmentlength ) - 0.499)
}
\fi
\else
\pgfkeys{/pgf/decoration automaton/next state=downsine}
\ifendcompletesineup
\pgfmathsetmacro\matchinglength{
0.5* \pgfdecoratedinputsegmentlength / (ceil(0.5 * \pgfdecoratedinputsegmentlength / \pgfdecorationsegmentlength ) - 0.4999)
}
\else
\pgfmathsetmacro\matchinglength{
0.5 * \pgfdecoratedinputsegmentlength / (ceil(0.5 * \pgfdecoratedinputsegmentlength / \pgfdecorationsegmentlength ) )
}
\fi
\fi
\setlength{\pgfdecorationsegmentlength}{\matchinglength pt}
}] {}
\state{downsine}[width=\pgfdecorationsegmentlength,next state=upsine]{
\pgfpathsine{\pgfpoint{0.5\pgfdecorationsegmentlength}{0.5\pgfdecorationsegmentamplitude}}
\pgfpathcosine{\pgfpoint{0.5\pgfdecorationsegmentlength}{-0.5\pgfdecorationsegmentamplitude}}
}
\state{upsine}[width=\pgfdecorationsegmentlength,next state=downsine]{
\pgfpathsine{\pgfpoint{0.5\pgfdecorationsegmentlength}{-0.5\pgfdecorationsegmentamplitude}}
\pgfpathcosine{\pgfpoint{0.5\pgfdecorationsegmentlength}{0.5\pgfdecorationsegmentamplitude}}
}
\state{final}{}
}
\tikzset{
% style to apply some styles to each segment of a path
on each segment/.style={
decorate,
decoration={
show path construction,
moveto code={},
lineto code={
\path [#1]
(\tikzinputsegmentfirst) -- (\tikzinputsegmentlast);
},
curveto code={
\path [#1] (\tikzinputsegmentfirst)
.. controls
(\tikzinputsegmentsupporta) and (\tikzinputsegmentsupportb)
..
(\tikzinputsegmentlast);
},
closepath code={
\path [#1]
(\tikzinputsegmentfirst) -- (\tikzinputsegmentlast);
},
},
},
% style to add an arrow in the middle of a path
mid arrow/.style={postaction={decorate,decoration={
markings,
mark=at position .5 with {\arrow[#1]{stealth}}
}}},
}
\begin{document}
\begin{frame}
\frametitle{Field and Mass Renormalization}
\begin{center}
\begin{tikzpicture}[thick,scale=.6]
\path [draw=blue,postaction={on each segment={mid arrow=blue}}]
(-4,0) -- (-2,0) -- (2,0) -- (4,0);
\draw[draw=blue,decorate, decoration=complete sines] (2,0) arc (0:180:2cm);
\end{tikzpicture}
\end{center}
\end{frame}
\end{document}
Any fresh ideas on this respect?
Best Answer
Here's a different approach that does not use a decoration, but rather a
to
path together with aplot
statement that connects two points using a semicircle with a superimposed sine wave.The number of complete periods to be drawn is specified using the key
wave count
. Another half period is added to this number so that the wave starts and ends on the outside (I think this looks better than starting on the inside or asymmetrically).The amplitude of the wave is set using
wave amplitude
, and the semicircle can be flipped to the opposite side of the path usingmirror semicircle=true
.This approach could reasonably easily be adapted to allow for circle sectors different from 180° (I don't know if that's ever needed in Feynman diagrams, though).