Is $B_t + B_{2t}$ a Markov process?

145 Views Asked by At

If $B = (B_t)_{t\geq 0}$ is a standard brownian motion, is $B_t + B_{2t}$ a markov process?

My intuition tells me that it's not a markov process because we don't have independent increments (I think). As far as I know, if we take $t_{k} < t_{k+1} < t_{n} < t_{n+1}$, then nothing assures that $[t_{2(n+1)}, t_{2n}]\cap[t_{2{k+1}}, t_{2k}] = \emptyset$. However, I'm not sure whether this is enough or not. Any help, please?

1

There are 1 best solutions below

3
On BEST ANSWER

You are right because the increments are not independent. You can explicitly compute the covariance of the increments. For example for the increments between $t=0$ to $t=1$ and between $t=1$ to $t=2$. Assuming $B_0=0$, we compute: \begin{align} C_t &= B_t + B_{2t} \\ \mathrm{Cov}(C_1-C_0, C_2-C_1) &= \mathrm{Cov}(B_1+B_{0.5}, (B_2+B_1)-(B_1+B_{0.5})) \\ &= \mathrm{Cov}(B_1+B_{0.5}, B_2-B_{0.5}) \\ &= \mathrm{Cov}(B_1,B_2)+\mathrm{Cov}(B_{0.5}, B_2)-\mathrm{Cov}(B_1,B_{0.5})-\mathrm{Cov}(B_{0.5},B_{0.5}) \\ &= \mathrm{Cov}(B_1,B_1)+\mathrm{Cov}(B_{0.5}, B_{0.5})-\mathrm{Cov}(B_{0.5},B_{0.5})-\mathrm{Cov}(B_{0.5},B_{0.5}) \\ &= \mathrm{Cov}(B_1,B_1)-\mathrm{Cov}(B_{0.5}, B_{0.5}) \\ &=1-0.5\neq 0 \end{align} Well, not the most beautiful calculation, but it works. In particular we used the fact that $\mathrm{Cov}(B_a,B_b)=\mathrm{Cov}(B_c,B_c)$ with $c=\min(a,b)$. (This is true because $B$ has independent increments.)