Problematic nested subscript

Hello and sorry if I'm reiterating an already asked question. The two i-subscripts have different positions in relation to m, the one on the right hand side of the inequality being the correct one presumably. Is there any way to fix this?

\documentclass[]{article}
\usepackage{mathtools}
\usepackage{amssymb}
\begin{document}
\begin{equation*}
\dim_{\mathbb{F}_{2^{m_{i}}}} W \geq \frac{n}{m_i}
\end{equation*}
\end{document}


-
Welcome to tex.sx please always post complete (small) documents showing all packages used. You can use the image upload button but remove (just) the ! in the generated markup: that will make a link to your image, someone with edit rights will put the ! back. –  David Carlisle Feb 2 '13 at 22:20
This is by design: the m in the left is smaller than the m in the right, so the output is a bit wider to enhance readability. But please don't use this sub-super-sub-superscript notation! It's not readable anyway ... –  Hendrik Vogt Feb 2 '13 at 22:22
I'll upload the image, it looks OK to me, subsubscripts are set in a more cramped style b default –  David Carlisle Feb 2 '13 at 22:24
I want to say something about the dimension of W as a vector space over the finite field with 2^{m_i} elements. I don't see any way around it, unless I use awkward abbreviated notation of some sort. –  Stefanos Aivazidis Feb 2 '13 at 22:29
You could force the superscriot to be set in textstyle (and so the sub-superscript in subscriptstyle but I'm not sure that would look better, I think you just accept that that is a choice of the font designer (Knuth in this case) and leave it that way or try another font set. –  David Carlisle Feb 2 '13 at 22:35

Generally, when a font is scaled to a lower point size, the glyphs can change to retain legibility. However, your unhappiness over the relation of the i and m indicates you don't like the choices made by the font.

Originally I used the scalerel package to take a normalsize m_i and scale it down to an appropriate size, but then Hendrik Vogt pointed out that the same could be more simply accomplished using a \scalebox{0.6}{$m_i$} for the sub-sub-superscript, since \usebox does not apply the \mathpalette wrapper.

The \scalebox creates the subscripted m_i to be an exact scale of the normal size m_i. Of course, the drawback is that, when scaled down, the strokes become thinner, and some may consider that more detrimental to the appearance that the relation of the i to the m. In any event, here it is:

\documentclass{article}
\usepackage{mathtools}
\usepackage{amssymb}
\usepackage{graphicx}
\begin{document}
\begin{equation*}
\dim_{\mathbb{F}_{2^{%
\scalebox{0.6}{$m_i$}%
}}} W \geq \frac{n}{m_i}
\end{equation*}
\end{document}


-

I strongly advise against using these four-level sub- and superscripts. Using two levels is OK, three levels is already dubious. In my opinion, the major problem in the output is not the distance between m and the index i but the fact that the \mathbb{F} and the m are almost on the same level. This way it almost looks like the product of \mathbb{F}_2 and m_i. If you really want to do it, here are some manual tweaks:

I pushed the index 2 further down by adding an empty superscript to the \mathbb{F}, and I reduced the horizontal space at three places, most notably before the W.

\documentclass{article}
\usepackage{amssymb}
\begin{document}
$\dim_{\mathbb{F}^{}_{\mkern-1mu 2^{m_{\mkern-1mu i}}}} \mkern-5mu W$
\end{document}


However, the better option is to use the recommendations from the comments to the question, e.g., to define \mathbb{F}(m_i) or \mathbb{F}(2,m_i) as a notation for the finite field with 2^{m_i} elements.

-