How to show that for any constant $\mu \in (0,1)$ the following inequality holds \begin{equation} \sum_{t=M}^N \sum_{s=M}^N \mu^{|t-s|} \leq C (N-M) \end{equation} in which $C$ is a constant that might depend on $\mu$.
I started as follows: write \begin{align} \sum_{t=M}^N \sum_{s=M}^N \mu^{|t-s|} &= \sum_{t=M}^N \left( \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t}^t \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \right)\\ &= \sum_{t=M}^N \sum_{s=t}^{t} \mu^{|t-s|} +\sum_{t=M}^N \left( \sum_{s=M}^{t-1} \mu^{|t-s|}+ \sum_{s=t+1}^N \mu^{|t-s|} \right)\\ & = (M-N) + \sum_{t=M}^N \left( \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \right) \end{align} Not sure how to get a multiplicative constant $C$ (independent of $N$ and $M$) to establish the inequality.
EDIT: few more steps. From above we see that
\begin{align} \sum_{t=M}^N \sum_{s=M}^N \mu^{|t-s|} &= (M-N) + \sum_{t=M}^N \left( \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \right) \end{align} Now consider \begin{equation} \left( \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \right) \end{equation} Because $\mu\in(0,1)$, the maximum term in each of these two sums are $\mu^{|t-(t-1)|} = \mu$ and $\mu^{|t-(t+1)|} = \mu$ respectively. Therefore we can have the following bound \begin{equation} \left( \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \right) \leq \left( \mu \sum_{s=M}^{t-1} 1 + \mu \sum_{s=t+1}^N 1 \right) = \mu ( |t-1 -M| + |N - t-1|) \end{equation}
now applying the second sum it seems that I get \begin{equation} \mu \sum_{t=M}^N \left(\sum_{s=M}^{t-1} 1 + \sum_{s=t+1}^N 1 \right)\stackrel{?}{\leq} 2\mu (N-M-1) \end{equation}
You're almost there. In the expression \begin{equation} \sum_{s=M}^{t-1} \mu^{|t-s|} + \sum_{s=t+1}^N \mu^{|t-s|} \end{equation} both sums are geometric sums with common ratio $\mu$ and largest term $\mu$. Go nuts and upper bound each sum by the infinite sum $$ \mu+\mu^2+\mu^3+\cdots =\frac\mu{1-\mu}, $$ which is a constant depending on $\mu$ but not on $M$ and $N$.
I suspect the RHS of the putative inequality should be $C(N-M+1)$. (Look at the case $M=N$.) The above argument gives $$C=1+\frac{2\mu}{1-\mu}=\frac{1+\mu}{1-\mu}.$$