The error is multinomially distributed with mean zero and covariance matrix $A$.
There is also a decrease in clustering as $n$ increases, particularly for normally distributed $X$.
Go to the list of words starting with: a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
y
z