I am adding an answer for a remark (the OP got the answer from @Manuel's comment to use \bm from bm package; one can also use \boldsymbol from amsmath package).
The remark here is that there was a problem with \mu but not with \Sigma. Why?
The point is that the cmr font contains 11 uppercase greek letters in slots 0 to 10, in the OT1 encoding. This is a bizarre legacy, but it does make \mathbf{\Sigma} work. I.e. the font used for operator names (log, sin, cos,...) also contains the uppercase (non-Latin) Greek letters. The idea of the \math.. alphabet commands is to use the text font in math mode. That it works for those 11 Greek letters is counter-intuitive and bound to the OT1 encoding used by the "operator" font. It is not robust against changes of the "operator" font.
Hence I would not recommend using \mathbf{\Sigma}, because it twists too much the idea of the LaTeX NFSS math alphabet commands. It ties it to peculiarity of Knuth's OT1 encoding.
\newcommand*\vector[1]{\mathbf{#1}}and that way you get clearer code an easy to change. In any case, for\muyou would be better using\bmfrombmpackage. – Manuel Nov 30 '17 at 10:58capitalize but to get a bold\mu`. – Nov 30 '17 at 10:58\mathbfis not expected to do anything good on\muin traditional pdflatex context. You can use also\boldsymbol{\mu}if using amsmath. Admittedly your\mathbf{\Sigma}works... turns out indeed that in the default setting\Sigmais of variable family type but not\mu. – Nov 30 '17 at 11:54