Tikz boxes do not appear of right size

Question

I have a beamer document with an eqnarray. I would like the box to appear nice and clean. Specifically, I would like that the boxes around the parts of the equations appear of the right size. The actual code (below) is giving me this output:

as you can see boxes are placed right but they are too short (the blue and the green one).

\begin{frame}

Details on computations leading to the following expression of $R_i$ are reported in the paper. Here it is worth noticing that the generalization error can be split into 6 components, namely:

\tikzstyle{na} = [baseline=-1.3ex]

\begin{itemize}[<+-| alert@+>]
    \item Coriolis acceleration
        \tikz[na] \node[coordinate] (n1) {};
\end{itemize}
%\begin{eqnarray}
%        \tikz[baseline]{
%            \node[fill=blue!20,anchor=base] (t1)
%            {$2\vec{\omega}_{ib}\times\frac{{}^bd}{dt}\vec{r}$};
%}
%\end{eqnarray}

{\footnotesize

\begin{eqnarray}\label{eq:generalization_error_FE_var}
&&{R_i\left(\{\underline{x}_{n,t}\}_{n=1,\ldots,N}^{t=1,\ldots,T}\right)} \nonumber \\
&& \tikz[baseline]{
            \node[fill=blue!20,anchor=base,height=3em, width=130mm] (t1)
{$\frac{\sigma^2}{{T}^2} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} {X_i'} \underline{1}_{{T}}$}}
+\tikz[baseline]{
            \node[fill=red!20,anchor=base] (t1) {$\frac{\sigma^2}{{T}^2}\underline{1}_{{T}}' {\Psi} \underline{1}_{{T}}$}}
 \nonumber \\
&&- \tikz[baseline]{
            \node[fill=green!20,anchor=base] (t1) {$\frac{2 {\sigma^2}}{{T}^2} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+}  \ddot{X}_n} \right)^{-1} {\ddot{X}_i'\Phi^{+} Q_T} {\Psi} \underline{1}_{{T}}$}}  \nonumber \\ 
&& + \sigma^2 \mathbb{E} \Bigg\{\left(\underline{x}_i^{test}\right)'\left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n} \right)^{-1} \underline{x}_i^{test} {\big| \{\underline{x}_{n,t}\}_{n=1,\ldots,N}^{t=1,\ldots,T}} \Bigg\} \nonumber \\
&&-\frac{2 \sigma^2}{{T}} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} \mathbb{E} \left\{\underline{x}_i^{test} \right\} \nonumber \\
&& +\frac{2 {\sigma^2}}{{T}} \left({Q_T} {\Psi} \underline{1}_{{T}}\right)' {\Phi^{+} \ddot{X}_i} \left(\sum_{n=1}^N {\ddot{X}_n' \Phi^{+} \ddot{X}_n} \right)^{-1} \mathbb{E} \left\{\underline{x}_i^{test} \right\}\,,
\end{eqnarray}

}

\end{frame}

How can I fix it? I would also like arrows to appear and display: First part (on blue rectangle), second part (on red) and third part (on green rectangle)

Answer 1

I consider only equation (it is so huge, that I doubt that you will be able to insert left out stuff this frame):

Is above what you after? I try to wee bit improve your equation layout, but in this I may introduce some math error. So check it if it is still correct.

\documentclass{beamer}
\usepackage{tikz}
\usepackage{mathtools}
    \tikzset{every picture/.append style = {baseline},
             box/.style = {fill=#1, minimum height=9ex, anchor=base},
             }


\begin{document}
\begin{frame}
    \footnotesize
\begin{align}\label{eq:generalization_error_FE_var}
    \MoveEqLeft
R_i\left(\bigl\{\underline{x}_{n,t}\bigr\}_{n=1,\ldots,N}^{t=1,\ldots,T}\right) \notag \\
    & = \tikz{\node[box=blue!20] (t1)
        {$\displaystyle\frac{\sigma^2}{{T}^2} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} {X_i'} \underline{1}_{{T}}$}}
      + \tikz{ \node[box=red!20] (t2) 
        {$\displaystyle\frac{\sigma^2}{{T}^2}\underline{1}_{{T}}' {\Psi} \underline{1}_{{T}}$}} \notag  \\
%
    & - \tikz{\node[box=green!20] (t3) 
        {$\displaystyle\frac{2 {\sigma^2}}{{T}^2} \underline{1}_{{T}}' {X_i} 
        \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+}  \ddot{X}_n} \right)^{-1} {\ddot{X}_i'\Phi^{+} Q_T} {\Psi} \underline{1}_{{T}}$}}  \notag \\
%
    & + \sigma^2 \mathbb{E} \left\{\left(\underline{x}_i^\mathrm{test}\right)'\left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n} \right)^{-1}
        \underline{x}_i^\mathrm{test} \middle| \bigl\{\underline{x}_{n,t}\bigr\}_{n=1,\ldots,N}^{t=1,\ldots,T} \right\} \notag \\
%
    & - \frac{2 \sigma^2}{{T}} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} \mathbb{E} \left\{\underline{x}_i^\mathrm{test} \right\} \notag \\
%
    & + \frac{2 {\sigma^2}}{{T}} \left({Q_T} {\Psi} \underline{1}_{{T}}\right)' {\Phi^{+} \ddot{X}_i} \left(\sum_{n=1}^N {\ddot{X}_n' \Phi^{+} \ddot{X}_n} \right)^{-1} \mathbb{E} \left\{\underline{x}_i^\mathrm{test} \right\}\,,
\end{align}
\end{frame}
\end{document}

Note: Ih you lake to have boxed layout as are shown in your question, than you only need to change box style to something like this:

             box/.style = {draw, fill=#1, minimum height=9ex, anchor=base,
                           drop shadow},

and load TikZ library shadows to document preamble.

Addendum:

If you use in other frame nodes styles with the same name as in this one, but have different options, which contradict used in above MWE, than the simple solution (when contents of other frame are unknown) is move \tikzset for this frame inside frame (to make it local) and add option `fragile to frame:

\documentclass{beamer}
\usepackage{tikz}
\usepackage{mathtools}

\begin{document}
\begin{frame}[fragile] % <---
    \tikzset{every picture/.append style = {baseline},
             box/.style = {fill=#1, minimum height=9ex, anchor=base},
             }
    \footnotesize
\begin{align}\label{eq:generalization_error_FE_var}
    \MoveEqLeft
R_i\left(\bigl\{\underline{x}_{n,t}\bigr\}_{n=1,\ldots,N}^{t=1,\ldots,T}\right) \notag \\
    & = \tikz{\node[box=blue!20] (t1)
        {$\displaystyle\frac{\sigma^2}{{T}^2} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} {X_i'} \underline{1}_{{T}}$}}
      + \tikz{ \node[box=red!20] (t2)
        {$\displaystyle\frac{\sigma^2}{{T}^2}\underline{1}_{{T}}' {\Psi} \underline{1}_{{T}}$}} \notag  \\
%
    & - \tikz{\node[box=green!20] (t3)
        {$\displaystyle\frac{2 {\sigma^2}}{{T}^2} \underline{1}_{{T}}' {X_i}
        \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+}  \ddot{X}_n} \right)^{-1} {\ddot{X}_i'\Phi^{+} Q_T} {\Psi} \underline{1}_{{T}}$}}  \notag \\
%
    & + \sigma^2 \mathbb{E} \left\{\left(\underline{x}_i^\mathrm{test}\right)'\left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n} \right)^{-1}
        \underline{x}_i^\mathrm{test} \middle| \bigl\{\underline{x}_{n,t}\bigr\}_{n=1,\ldots,N}^{t=1,\ldots,T} \right\} \notag \\
%
    & - \frac{2 \sigma^2}{{T}} \underline{1}_{{T}}' {X_i} \left(\sum_{n=1}^N {\ddot{X}_n'\Phi^{+} \ddot{X}_n}\right)^{-1} \mathbb{E} \left\{\underline{x}_i^\mathrm{test} \right\} \notag \\
%
    & + \frac{2 {\sigma^2}}{{T}} \left({Q_T} {\Psi} \underline{1}_{{T}}\right)' {\Phi^{+} \ddot{X}_i} \left(\sum_{n=1}^N {\ddot{X}_n' \Phi^{+} \ddot{X}_n} \right)^{-1} \mathbb{E} \left\{\underline{x}_i^\mathrm{test} \right\}\,,
\end{align}
\end{frame}
\end{document}

Result is the same as before.