1
$\begingroup$

I am trying to evaluate the following integral: $$ \int_{-\infty}^{\infty}e^{-\frac{x^2}{2}}\Phi(x-t)dx $$ where $$ \Phi(y) = \frac{1}{2} + \frac{1}{2}erf\left(\frac{y}{\sqrt{2}}\right) $$ I have tried integration by parts, a substitution, but nothing useful came out of this

$\endgroup$
6
$\begingroup$

$$ \begin{align} \int_{-\infty}^{\infty} e^{-\frac{x^2} {2}} \Phi(x-t) dx & = \sqrt{2\pi}\int_{-\infty}^{\infty} \phi(x) \Phi(x-t) dx \\ & = \sqrt{2\pi}\Pr\{Z_1 \leq Z_2 - t\} \\ & = \sqrt{2\pi}\Pr\left\{\frac {Z_1 - Z_2} {\sqrt{2}} \leq - \frac {t} {\sqrt{2}}\right\} \\ & = \sqrt{2\pi}\Phi\left(- \frac {t} {\sqrt{2}}\right) \end{align} $$ where $\phi$ and $\Phi$ is the standard normal pdf and CDF respectively, $Z_1, Z_2$ are independent standard normal random variables.

$\endgroup$
1
$\begingroup$

The key is to express as integral of a gaussian. $$ I(t) = \int_{-\infty}^\infty \exp\left(-\frac{x^2}{2} \right)~\Phi(x-t)~dx\\ = \int_{-\infty}^\infty \exp\left(-\frac{x^2}{2} \right)~\left[ 1-\Phi(t-x) \right]~dx\\ =\sqrt{2\pi}-\int_{-\infty}^\infty{ \exp\left(-\frac{x^2}{2} \right)~ \Phi(t-x) ~dx } $$ Plug in $$ \Phi(t-x)=\int_{\infty}^{t}{ \frac{1}{\sqrt{2\pi}}\exp\left( -\frac{(y-x)^2}{2} \right) ~dy } $$ to get $$ I(t)=\sqrt{2\pi}-\int_{-\infty}^\infty{ \exp\left(-\frac{x^2}{2} \right)~ \int_{\infty}^{t}{ \frac{1}{\sqrt{2\pi}}\exp\left( -\frac{(y-x)^2}{2} \right) ~dy } ~dx }\\ =\sqrt{2\pi}- \int_{\infty}^{t}{ \frac{1}{\sqrt{2}}\exp \left(-\frac{y^2}{4} \right) ~dy }\\ =\sqrt{2\pi}- \sqrt{2\pi}\Phi\left(\frac{t}{\sqrt{2}} \right)\\ =\sqrt{2\pi}\Phi\left( -\frac{t}{\sqrt{2}} \right) $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.