# mixture of complex gaussians

I would like to compute the following probability

$$P\left( ax \geq \sum_{i=1}^n b_i y_i \right)$$

where $a, b_i$ are constant coefficients (in my case, they are positive too) and $x, y_i$ are independent complex gaussian random variables with mean $0$ and variance $1$.

-
Do you really mean complex Gaussians? If so, mean, variance do not give full specification. –  André Nicolas Jul 28 '11 at 19:40
Could you mean circularly symmetric complex Gaussian, and be interested in the behaviour of norms, or maybe projections on the real axis? –  André Nicolas Jul 28 '11 at 23:36

Like André Nicolas, I do not understand your use of complex, especially with the $\ge$ symbol. So what follows ignores it. I will also use Capitals for random variables
$aX$ is a Gaussian random variable with mean $0$ and variance $a^2$.
Similarly $\sum_{i=1}^n b_i Y_i$ is a Gaussian random variable with mean $0$ and variance $\sum_{i=1}^n b_i^2$ because of independence.
And $aX - \sum_{i=1}^n b_i Y_i$ is a Gaussian random variable with mean $0$ and variance $a^2+\sum_{i=1}^n b_i^2$ because of independence.
So $\Pr( aX \geq \sum_{i=1}^n b_i Y_i ) = \frac{1}{2}$.