In Karatzas&Shreve Brownian Motion and Stochastic Calculus Problem 3.27: Let $T$ be a bounded stopping time of the filtration $\left\{ \mathscr{F}_{t}\right\} $, which satisfies the usual conditions, and define $\mathscr{\widetilde{F}}_{t}=\mathscr{F}_{T+t}$. Then $\left\{ \mathscr{\widetilde{F}}_{t}\right\} $ also satisfies the usual conditions. And if $\left\{ \widetilde{X}_{t},\mathscr{\widetilde{F}}_{t}:0\le t<\infty\right\}$ is a right-continuous submartingale with $\widetilde{X}_{0}=0$, a.s. P, then $\left\{ X_{t}=\widetilde{X}_{\max\left\{ 0,t-T\right\} },\mathscr{F}_{t}:0\le t<\infty\right\}$ is also a submartingale.
I have solved this problem except one part. My question is that: how can we show that $\widetilde{X}_{\max\left\{ 0,t-T\right\} }$ is adapted $\left\{ \mathscr{F}_{t}\right\}$.
If $t\le T$, $X_{t}=0$. So knowing all the past before $t$ allows you to know $X$ until that time.
If $t\ge T$, $X_{t}=\widetilde{X}_{t-T}$. So knowing all the past before $t$ for the filtration $\mathscr{{F}}$ allows you to know all the past before $t-T$ for the filtration $\mathscr{\widetilde{F}}$, which allows you to know $\widetilde X$ until time $t-T$ and therefore $X$ until time $t$.
Formally, let $A\in\mathcal{B}(\mathbb{R})$ and $t\ge T$. We need to prove that $\left\{X_t\in A\right\}\in\mathscr{F}_t$:
\begin{alignat*}{2} \left\{\omega\in\Omega, X_t(\omega)\in A\right\} = \left\{\omega\in\Omega, \widetilde X_{t-T}(\omega)\in A\right\} \end{alignat*} but $\left\{\widetilde X_{t-T}\in A\right\}$ is $\widetilde{\mathscr{F}}_{t-T}$-measurable, which means it is $\mathscr{F}_{t}$-measurable or equivalently $\left\{X_t\in A\right\}\in\mathscr{F}_t$.