My problem is that the outputs of \bmatrix are a bit weird and not of the same height and the linespacings are different depending on element heights.
I have put these on a blank document and I got a satisfactory result therefore I think maybe some packages are interfering with each other.
Also my other bmatrices are not affected and removing the align environment doesn't solve the problem.
These are the packages I use
\usepackage{amsthm,amssymb,amsmath}
\usepackage{epsf,graphicx}
\usepackage[a4paper]{geometry}
\usepackage{titlesec}
\usepackage{setspace}
\usepackage[stable,bottom]{footmisc}
\usepackage{placeins}
\usepackage[colorlinks,citecolor=blue]{hyperref}
\hypersetup{linkcolor=blue}
\usepackage{zref-perpage}
\zmakeperpage[1]{footnote}
\usepackage{cancel}
\usepackage{tikz}
\usetikzlibrary{fit,positioning}
\usetikzlibrary{bayesnet}
\usepackage{adjustbox}
\usepackage{pgfplots}
\pgfplotsset{every tick label/.append style={font=\tiny}}
\usepackage{algorithm}
\usepackage{algorithmic}
\usepackage{booktabs}
\usepackage{units}
And this is the tex code
\begin{align}
\begin{bmatrix}
0 & \Sigma^r_{xy} \\
\Sigma^r_{yx} & 0 \\
\end{bmatrix}
\begin{bmatrix}
l_{x} \\
l_{y} \\
\end{bmatrix}
=\lambda
\begin{bmatrix}
\Sigma^r_{xx} & 0 \\
0 & \Sigma^r_{yy} \\
\end{bmatrix}
\begin{bmatrix}
l_{x} \\
l_{y} \\
\end{bmatrix} \\ %%%%%%
\begin{bmatrix}
0 & \Sigma^l_{xy} \\
\Sigma^l_{yx} & 0 \\
\end{bmatrix}
\begin{bmatrix}
r_{x} \\
r_{y} \\
\end{bmatrix}
=\lambda
\begin{bmatrix}
\Sigma^l_{xx} & 0 \\
0 & \Sigma^l_{yy} \\
\end{bmatrix}
\begin{bmatrix}
r_{x} \\
r_{y} \\
\end{bmatrix}
\end{align}


articledocument class. – Mico Apr 16 '16 at 13:37\arraystretchand/or\extrarowheightsomewhere in the document? – Mico Apr 16 '16 at 13:42\begin{document}) that modifies some important parameters. Given how cramped the screenshot looks, I would suspect that\baselineskip,\arraystretch, or\extrarowheighthave been modified. – Mico Apr 16 '16 at 13:44