I would like to change the default behaviour of limits to include \mathclap.
Rather than writing every time
\int\limits_{\mathclap{verylonglimit}} x \, dx
I would like to simply write
\int\limits_{verylonglimit} x \, dx
and to get the same behaviour.
I tried with (and similar forms) of:
\let\oldlimits\limits
\renewcommand{\limits_}[1]{\oldlimits_{\mathclap{#1}}}
The problem is that I do not know how to deal with the underscore.
Rather than redefine \limits -- which has the potential to cause lots of grief elsewhere in your document -- I'd like to suggest you define a new command called, say, \mclimits (short for "mathclap limits"...) as follows:
\def\mclimits_#1{\limits_{\mathclap{#1}}}
Note that with this definition, any use of \mclimits that doesn't employ the _ character will result in a syntax error, as TeX will be scanning the input material for this character in order to "know" what's supposed to be macro's argument (the "#1" part). This constraint shouldn't be a problem in practice, since you'll want to use the _ character in order to indicate the start of the material that should be placed in the lower-limit part of the full integral expression.
Here's the output of an MWE that uses the macro \mclimits:
\documentclass{article}
\usepackage{mathtools}
\def\mclimits_#1{\limits_{\mathclap{#1}}}
\begin{document}
\[
\int\mclimits_{\text{verylonglimit}}\! f(x)\,dx
\]
\end{document}
\def, it's possible to create a so-called "argument-delimited" macro, i.e., a macro whose argument is delimited by a specific symbol (such as _). Note that it's not possible (or, at least, not as easily possible) to do this with the LaTeX directives \newcommand and \renewcommand.
– Mico
Jun 3 '15 at 17:23
\def\limits_#1{..} then any use of \limits (including existing commands that use it internally) that is not followed by _ is a syntax error.
– David Carlisle
Jun 3 '15 at 18:37
\def\limits_ does not define a macro with name limits_ it redefines the command with name limits such that it has to be followed by _ so \limits_{1} would be OK as would \limits _ {1} but \limits^{1} is an error.
– David Carlisle
Jun 3 '15 at 19:14
This might seem like an overkill… but that's how xparse works.
With the “complex code” what I do is define new type of xparse arguments, k, K, a, b, A, B. Being a and A the correspondent to subscripts. The others are explained here
\documentclass{scrartcl}
\usepackage{mathtools,amssymb,xparse}
\ExplSyntaxOn
\cs_new_protected:Npn \__xparse_count_type_k:w #1
{
\__xparse_single_token_check:n { #1 }
\quark_if_recursion_tail_stop_do:Nn #1 { \__xparse_bad_arg_spec:wn }
\__xparse_count_mandatory:N
}
\cs_new_protected:Npn \__xparse_count_type_K:w #1 #2
{
\__xparse_single_token_check:n { #1 }
\quark_if_recursion_tail_stop_do:nn { #2 } { \__xparse_bad_arg_spec:wn }
\__xparse_count_mandatory:N
}
\cs_new_protected:Npn \__xparse_add_type_k:w #1
{ \exp_args:NNo \__xparse_add_type_K:w #1 { \c__xparse_no_value_tl } }
\cs_new_protected:Npn \__xparse_add_type_K:w #1 #2
{
\__xparse_flush_m_args:
\__xparse_add_grabber_optional:N K
\tl_put_right:Nn \l__xparse_signature_tl { #1 { #2 } }
\__xparse_prepare_signature:N
}
\cs_new_protected:Npn \__xparse_add_expandable_type_k:w #1
{
\exp_args:NNo \__xparse_add_expandable_type_K:w #1 { \c__xparse_no_value_tl }
}
\cs_new_protected_nopar:Npn \__xparse_add_expandable_type_K:w #1 #2
{
\__msg_kernel_error:nnx { xparse } { invalid-expandable-argument-type } { K }
\__xparse_add_expandable_type_m:w % May be create this?
}
\cs_new_protected:Npn \__xparse_grab_K:w #1 #2 #3 \l__xparse_args_tl
{
\__xparse_grab_K_aux:NnnNn #1 { #2 } { #3 } \cs_set_protected_nopar:Npn
{ _ignore_spaces }
}
\cs_new_protected:Npn \__xparse_grab_K_long:w #1 #2 #3 \l__xparse_args_tl
{
\__xparse_grab_K_aux:NnnNn #1 { #2 } { #3 } \cs_set_protected:Npn
{ _ignore_spaces }
}
\cs_new_protected:Npn \__xparse_grab_K_trailing:w #1 #2 #3 \l__xparse_args_tl
{
\__xparse_grab_K_aux:NnnNn #1 { #2 } { #3 } \cs_set_protected_nopar:Npn
{ _ignore_spaces }
}
\cs_new_protected:Npn \__xparse_grab_K_long_trailing:w #1 #2 #3 \l__xparse_args_tl
{
\__xparse_grab_K_aux:NnnNn #1 { #2 } { #3 } \cs_set_protected:Npn
{ _ignore_spaces }
}
\cs_new_protected:Npn \__xparse_grab_K_aux:NnnNn #1 #2 #3 #4 #5
{
\exp_after:wN #4 \l__xparse_fn_tl ##1
{
\__xparse_add_arg:n { ##1 }
#3 \l__xparse_args_tl
}
\use:c { peek_meaning_remove #5 :NTF } #1
{ \l__xparse_fn_tl }
{
\__xparse_add_arg:n { #2 }
#3 \l__xparse_args_tl
}
}
\prop_put:Nnn \c__xparse_shorthands_prop { a } { k \sb }
\prop_put:Nnn \c__xparse_shorthands_prop { b } { k \sp }
\prop_put:Nnn \c__xparse_shorthands_prop { A } { K \sb }
\prop_put:Nnn \c__xparse_shorthands_prop { B } { K \sp }
\ExplSyntaxOff
\let\originalint\int
\RenewDocumentCommand\int{ t\limits a }
{\originalint\IfBooleanT{#1}{\limits}\IfValueT{#2}{_{\mathclap{#2}}}}
\begin{document}
\[
\int\limits_{verylonglimit} x \, dx
\]
\end{document}
I redefine the \int command first to gobble (and use) any \limits following it (if you want a different behaviour it can be changed, for instance if you want all of them to have \limits so \int_a would equal \int\limits_{\mathclap{a}}) then the command checks if there's a subscript like _{whatever} then pases whatever as second argument, we then pass it through a _{\mathclap{#2}} automatically.
\limitsin a document – David Carlisle Jun 3 '15 at 16:49\limitsyou will redefine\stackrel,\buildrel,\doteqand any other command that uses\limitsinternally. – David Carlisle Jun 3 '15 at 16:52\int_{\mathclap{verylonglimit}}\frac{1}{x}\,dx? – egreg Jun 3 '15 at 16:52amsmath(ormathtools) package with the optionintlimits. (b) If typing\mathclapfrequently is too tedious, just create a shortcut named, say,\mcto ease the job. – Mico Jun 3 '15 at 17:01