Finding $\mathbb E(X\mid Y)$ where $P(X > t) =e^{−t}$ with $t >0$ and $Y=\min(X,t)$

158 Views Asked by At

Find $\mathbb E(X\mid Y)$ where $X$ is a random variable such that $P(X > t) =e^{−t}$ with $t >0$ and $Y=\min(X,t)$.

I have not done a problem before where we condition on the minimum or maximum value. I thought about breaking it up into cases as follows:

$$\mathbb E(X\mid X>t)\mathbb P(X>t)+\mathbb E(X\mid X<t)\mathbb P(X<t)$$

but this gives $\mathbb E(X)$ rather than $\mathbb E(X\mid Y)$. I tried thinking of this conceptually but what makes this problem tricky is that (if I'm understanding the problem correctly) we are not told whether $X<t$ or $X>t$ but rather we're just given the smaller of the two values. If we are (unknowingly) given the information that $Y=\min\{X,t\}=t$ with probability $e^{-t}$ then by the memoryless property, $\mathbb E(S\mid t)=t+1$ and if we are (unknowingly) given the information that $Y=\min\{X,t\}=X$ with probability $1-e^{-t}$ then $\mathbb E(X\mid X=x)=x$ so that

$$\mathbb E(X\mid Y)=(t+1)e^{-t}+x\left(1-e^{-t}\right)$$

but here I'm basically doing the same thing as above so I'm not sure how to correctly think about this type of problem. Any help would be appreciated!

1

There are 1 best solutions below

1
On BEST ANSWER

First observe that $Y=\min\{X,t\}$ is supported on $(0,t]$ and that $E(X|Y=y)$ is necessarily a function of $y$. On one hand, if $y\in (0,t)$, we have $\{Y=y\}=\{X=y\}$ giving us $$E(X|Y=y)=E(X|X=y)=y$$ On the other hand, if $y=t$, then $\{Y=t\}=\{X\geq t\}$ so in this case $$E(X|Y=y)=E(X|X \geq t)$$ The RHS evaluates to $$\int_{t}^{\infty}\frac{xf_X(x)}{P(X\geq t)}dx=t+1$$ Here we're using the fact that $X\sim \text{Exponential}(1)$.