On Di, 2011-05-24 at 09:56 -0400, Aditya Mahajan wrote:
> On Tue, 24 May 2011, Paul Menzel wrote:

> > to markup the probability measure and the parenthesis around the
> > argument, I defined the following command.
> >
> > 	\define[1]\P{{\mathbf P}\left( #1 \right)}
> >
> > Unfortunately the space between the P and the left ( is a little big in
> > my opinion. Is that correct or should/can I fix that somehow?
> 
> Getting these spaces is tricky; especially if you consider spaces before 
> and after the definition. Choose your pick:
> 
> \starttext
>   \startformula
>    A{\mathbf P}\left( ABC \right)B \quad
>    A\mathop{\kern\zeropoint\mathbf P}\left( ABC \right)B \quad
>    A{\mathbf P}\mathopen{}\left( ABC \right)\mathclose{}B \quad
>    A\mathop{\kern\zeropoint\mathbf P}\mathopen{}\left( ABC \right)\mathclose{}B \quad
>   \stopformula
> \stoptext

That looks tricky and complex.

> Some information from the mathsets module (which does not work with MkIV).

I hope you or somebody will have time to port that module to MkIV.

> The entry is the latex bug database is not viewable now.
> 
> %D The \type{\left} and \type{\right} generate a math atom of type inner,
> %D while for math sets, we want a math open atom. To see the difference,
> %D consider
> %D
> %D \startbuffer
> %D \startformula
> %D  2\left(\frac {3}{4} \right) \qquad \hbox{ vs } \qquad
> %D  2\biggl( \frac {3}{4} \biggr)
> %D \stopformula
> %D
> %D and
> %D
> %D \startformula
> %D  \Pr\left(\frac {3}{4} \right) \qquad \hbox{ vs } \qquad
> %D  \Pr\biggl( \frac {3}{4} \biggr)
> %D \stopformula
> %D \stopbuffer
> %D \typebuffer
> %D
> %D which gives (notice the spacing before the parenthesis)
> %D
> %D \getbuffer
> %D
> %D I will assume that if \type{text} is something, then
> %D the default behaviour is desirable, if \type{text} is empty, then I add
> %D \type{\mathopen} and \type{\mathclose}.  Using \type{\mathopen} to correct
> %D the spacing is due to Frank Mittelbach, see
> %D \hyphenatedurl{http://www.latex-project.org/cgi-bin/ltxbugs2html?pr=latex/3853}
> %D
> %D Mathset module ensures that we get the correct spacing in both cases
> %D \startbuffer
> %D \definemathset[SET][left=(,right=)]
> %D \startformula
> %D  2\SET{\frac{3}{4}} \qquad \hbox{ and } \qquad
> %D  \PR{ \frac{3}{4} }
> %D \stopformula
> %D \stopbuffer
> %D \getbuffer[PR] \getbuffer which was typed as \typebuffer
> %D
> %D Also, if its argument is a single character, \type{\mathop} centers it 
> to
> %D with respect to the math||axis. Compare the outputs of
> %D
> %D \startbuffer
> %D \ruledhbox{$\mathop{y}\nolimits_x\left\{A\,\middle|\,B\right\}$}
> %D \ruledhbox{$\mathop{\kern\zeropoint y}\nolimits_x\left\{A\,\middle|\,B\right\}$}
> %D \stopbuffer
> %D
> %D \typebuffer
> %D \getbuffer
> %D
> %D I have added a \type{\kern\zeropoint} to prevent that.

Could you explain the advantage over `\!` which Wolfgang suggested at
least for the space after the definition.

In addition, I think AMSTeX defines a command `\DeclareMathOperator` [1]
which takes care of these things.


Thanks,

Paul


[1] http://en.wikibooks.org/wiki/LaTeX/Advanced_Mathematics#Custom_operators