Proof verification / Help understanding a step in a proof about probability and brownian motions

69 Views Asked by At

I would like some help to understand a step in a proof or/and a proof verification:

I've tried doing it by myself but I can't justify exactly the same as in the proof (is mine correct as well?)

Setup: For $t\in\left[\frac{k-1}{n},\frac{k}{n}\right]$ where $k$ is the unique integer such that $k-1\leq nt<k$, and $T_k$ are increasing $T_k\geq T_{k-1}$ random variables (stopping times here) of expectations $k$.

Here we have take $\delta\in(0,1)$ and $n>\delta/2$.

The step in the proof I dont understand:

$$\mathbb P\left(\exists t\in [0,1):\max\left(\left|\frac{T_k}{n}-t\right|,\left|\frac{T_{k-1}}{n}-t\right|\right)\geq\delta\right)\leq\mathbb P\left(\sup_{1\leq k\leq n}\frac{\max(T_k-(k-1),k-T_{k-1})}{n}\geq\delta\right)$$ $$\leq \mathbb P\left(\sup_{1\leq k\leq n}\frac{T_k-k}{n}\geq \delta/2\right)+\mathbb P\left(\sup_{1\leq k\leq n}\frac{k-1-T_{k-1}}{n}\geq\delta/2\right)$$

I've tried developing the maximum as $\frac{1}{2n}\left(|T_k-nt|+|T_{k-1}-nt|+|T_k-T_{k-1}|\right)$ and other reasonings but none lead me to the same result. The further I've gotten is:

My proof: $$\{\exists t\in [0,1]:\max\left(\left|\frac{T_k}{n}-t\right|,\left|\frac{T_{k-1}}{n}-t\right|\right)\geq\delta\}=\bigcup_{k=1}^n\{\exists t\in [(k-1)/n,k/n]: \max\left(\left|\frac{T_k}{n}-t\right|,\left|\frac{T_{k-1}}{n}-t\right|\right)\geq\delta\}=(\star)$$ Using the fact that $T_i$ is increasing and decomposing the absolute value, I obtain that $$\mathbb P(\star)\leq\mathbb P\left(\sup_{1\leq k\leq n}\max\left(\frac{T_k-(k-1)}{n},\frac{k-T_{k-1}}{n}\right)\geq\delta\right)$$ $$\leq \mathbb{P}\left(\sup_{1\leq k\leq n}\frac{T_k-(k-1)}{n}\text{ or }\sup_{1\leq k\leq n}\frac{k-T_{k-1}}{n}\geq\delta\right)$$ And using the union bound for measures: $$\leq \mathbb P\left(\sup_{1\leq k\leq n}\frac{T_{k-1}-(k-1)}{n}\geq \delta\right)+\mathbb P\left(\sup_{1\leq k\leq n}\frac{k-T_k}{n}\geq\delta\right)$$

The proof is by Peter Mörters and Yuval Peres on Brownian motions.

NB: I have seen a similar question asked here but the answer has not been approved and has me puzzled.

Thanks in advance for the proof verification or the help understanding the steps.

1

There are 1 best solutions below

1
On BEST ANSWER

Yes, that is correct. In the end there is one extra step where the inequality $n >\delta/2$ is used.