I am examining a way to adjust the size of nested delimiters automatically.
This gives 'double subscript':
\documentclass[a4paper, 12pt]{article}
\usepackage[english]{babel}
\usepackage{amsmath}
\title{Untitled}
\date{Typeset on \today}
\newcommand{\p}[1]{\mathopen{}\left({#1}{{}{}},\negthickspace\right)\mathclose{}}
\begin{document}
\maketitle
$$ D_{v_p} = \p{ {\widehat e}_i } $$
\end{document}
Without the _i, the error vanishes.
How to understand that?
If the base of the math accent is just a single character node (or single character node that has been subscripted or superscripted) the outer braces are essentially lost to allow finer font-specified control over positioning. Note how even the empty {} in the first form avoids the error (but adversely affects the display in some cases.
\documentclass[a4paper, 12pt]{article}
\begin{document}
${{}{\widehat e}i}{} $
${{\widehat e}i}{} $
\end{document}
This is a tex-primitive behaviour not controlled by latex, see this plain tex example
${{}\mathaccent"7013 a_i}_{} $
${\mathaccent"7013 {a_i}}_{} $
% error ${\mathaccent"7013 {za}i}{} $
% error ${{\mathaccent"7013 a_i}}_{} $
% error $\begingroup \mathaccent"7013 a_i\endgroup_{} $
% error ${\mathaccent"7013 a_i}_{} $
\bye
