Bdtyktl

Why is underbrace (and similar commands) in the form

underbrace{my formula}_{some text under my formula}

?

Shouldn't a standard macro be in the form

underbrace[some text under my formula]{my formula}

since the text to put under is optional?

Why is it treated like a math operator?

documentclass{article}

usepackage{amsmath}

begin{document}

[

underbrace{a+b=c}

]

[

underbrace{a+b=c}_{text{something}}

]

[

lim_{xto a} x= a

]

end{document}

The underbrace macro is exactly the same in LaTeX as in plain: it's one of those things which came with the original 'load on top of plain' approach, well before latex.ltx. Moreover, the _ 'argument' isn't an argument at all, it's a TeX core subscript. It looks like an argument, but the construction uses low-level TeX math mode primitives.

Re-implemented today, one would likely set up such that the alignment is done without the primitives, and thus the text part would be an argument. But that's a completely different question!

When you look at the definition of underbrace (or overbrace) you'll see the following:

> underbrace=macro:

#1->mathop {vtop {m@th ialign {##crcr $hfil displaystyle {#1}hfil $crcr

noalign {kern 3p@ nointerlineskip }upbracefill crcr noalign {kern 3p@ 

}}}}limits .

It shows that the first (and only) argument to underbrace is set as a math operator and closes with limits. And, math operators have their limits (superscripts and subscripts) set on top/below it. It's therefore no surprise that you can do:

$underbrace{abcd}_{dcba} underbrace{abcd}^{dcba}$

It's treated like a math operator because that's the primitive way of stacking elements below/above something so that it doesn't displace the someting vertically.

I think it's much easier to iterate with the underscore syntax. For example

documentclass{article}

usepackage{amsmath}

begin{document}

begin{equation*}

  underbrace{left(frac12+frac13right)}_{

    >underbrace{frac14+frac14}_{

      underbrace{2cdot frac14}_{

        frac12

        }

    }

  }

  quad + quad

    underbrace{left(frac14+frac15+frac16+frac17right)}_{

    >underbrace{frac18+frac18+frac18+frac18}_{

      underbrace{4cdot frac18}_{

        frac12

        }

    }

  }

  quad > quad frac12+frac12

end{equation*}



end{document}

Try to do the same with the "option" syntax. The point is that with the "underscore" syntax, you write the code for the iteration sequentially, from left to right, similarly to what the output will look like, while with the option syntax, you'd end up writing from right to left, so the bottom 1/2 in the output will be the leftmost formula after the three underbrace commands.