Integration of Gaussian random variables.

1.1k Views Asked by At

Suppose $g(t)$ is Gaussian random variable for real $t \in D $. Assume $[t,t+\Delta t] \subseteq D $.

I want to prove that, for \begin{equation} \xi = \int_t ^{t+\Delta t} g (\tau ) d\tau \end{equation} $\xi$ is Gaussian random variable.

I already know that for random variables $X_1, X_2$ with normal distribution, \begin{equation} X = X_1 + X_2 \end{equation} is have normal distribution. However, I have difficulty on continuous and integral instead of discontinuous and sum.

Does anyone have idea or counter examples of my question?

1

There are 1 best solutions below

4
On

It is not true in general that the sum of Gaussian random variables is Gaussian - this is only true if they are independent, as well. In a similar way, your claim about the integral is not true without some further assumptions on the covariance structure of your Gaussian process $g(t)$.

A well-known counterexample goes as follows. Let $X,Y$ be independent standard Gaussians, and let $$ Z=\text{sign}(X)\cdot |Y|. $$ Then $Z$ is also standard Gausian, but it can be shown that $X+Z$ is not Gaussian at all.