Let $ X,Y$ be two standard normal random variables with correlation coefficient $ r\in[-1,1]$ .
I’m trying to compute the moment generating function of the maxima using Mathematica:
Expectation[Exp[t Max[X, r X + Sqrt[1 - r^2] Y]], {X \[Distributed] NormalDistribution[], Y \[Distributed] NormalDistribution[]}]
Since the distribution of the maximum is known to follow the Skew Normal Distribution, I can easily compute the expectation myself to be
$ $ e^{t^2/2} (1 + \text{erf}(\sqrt{1-r}\,t/2),$ $
but Mathematica seems to get bogged down.
I have some slightly more complicated expectations I would like to evaluate, so I wonder if there is some way I may help Mathematica solve things like this? I’m using Mathematica 12.0.0, and I’ve tried adding $ Assumptions = -1 < r < 1
.