# Define tree recursively with Forest

I want to let forest draw trees similar to following one automatically in a way, that changing the root of the tree will lead to the correct tree for this root.

The root is any natural number. The rule is child1=floor((parent-1)/2) and child2=floor(parent/2).

For now I wrote a Fortran-95 programm that produces some Forest code, that makes me able to produce such trees. I would now like to do in a more elaborated way.

I found this code in another question and I think it is going in the right direction:

\begin{forest}
Stern Brocot/.style n args={5}{%
content=$\frac{\number\numexpr#1+#3\relax} {\number\numexpr#2+#4\relax}$,
if={#5>0}{% true
append={[,Stern Brocot={#1}{#2}{#1+#3}{#2+#4}{#5-1}]},
append={[,Stern Brocot={#1+#3}{#2+#4}{#3}{#4}{#5-1}]}
}{}}% false (empty)
[,Stern Brocot={0}{1}{1}{0}{3}]
\end{forest}


But I am not able to understand how these #1,#2,... are working.

• Welcome! What do you have so far? Right now, this is just another do-it-for-me without any ducks. – cfr Jul 21 '17 at 17:21
• I read the first answer of tex.stackexchange.com/questions/106895/…. I also read the tutorial from the forest documentation. But I have problems understanding how these features of Forest are working. What i have right now is a small program that writes the tree in Forestsyntax into a file. – Imperaton Jul 21 '17 at 17:28
• I actually just want to find out how I can do something like this, without browsing and reading around for more hours, An example of something similar with an explanation would also be sufficient. But I didn't find any. – Imperaton Jul 21 '17 at 17:36
• Please always post complete code people can compile and include links in your question. Comments are liable to deletion on a regular basis. – cfr Jul 21 '17 at 19:30
• The hashes are just the place holders for the arguments, as in regular macro definitions. The style shown takes 5 arguments, so the place holders are #1, #2, #3, #4 and #5. In the example, these gets the values 0, 1, 1, 0 and 3 respectively. – cfr Jul 21 '17 at 19:42

Note that I think I ought not answer do-it-for-mes, but sometimes do so anyway. When I do so, I do so for me. If the result happens to be of use, so be it. If not, not. Having started from scratch, I am typically less sympathetic to requests for fine-tuning, adjustments, explanations and other follow-up. Fine-tuning and adjustments are typically left as an exercise for the under-exercised reader. Explanations are for specific questions about particular problems, obstacles or code, where a user asks a question which lets me explain this bit, this change or this step. Since do-it-for-mes don't post code, there is no starting point from which to begin an explanation. In any case, I answer these solely for me. Any explanation is for me.

\documentclass[border=10pt]{standalone}
\usepackage{forest}
\forestset{
parhau/.style={
delay={
tempcounta/.process={O+nw+P {content}{int(floor((##1-1)/2))}},
tempcountb/.process={O+nw+P {content}{int(floor(##1/2))}},
if={
>Rn>{tempcounta}{0}
}{
prepend/.process={ RSn=?_ lw2  {tempcounta}{1}{}{parhau}{[##2, ##1]} },
}{},
append/.process={ RSn=?_ lw2  {tempcountb}{1}{}{parhau}{[##2, ##1]} },
},
},
}
\begin{document}
\begin{forest}
parhau
[20]
\end{forest}
\end{document}


• Togheter with the Forest Documentation I was able to learn alot from this. What still confuses me: What is the job of the first '>' in this line: >Rn>{tempcounta}{0}. – Imperaton Jul 21 '17 at 21:56
• @Imperaton See page 64. It is not using the .process handler, but the generic <forestmath>. Hence the switch. – cfr Jul 21 '17 at 23:45
• Do you always use ##1,##2,... (and not #1,#2) for the wrap process, or does it have a special reason here? – Imperaton Jul 22 '17 at 6:40
• @Imperaton No. It is in a style, so one hash would be an argument passed to the style. – cfr Jul 22 '17 at 13:06