Returns the first derivative of the error function.
num dxErf(num x) => 2.0 * invSqrtPi * math.pow(math.e, -x * x);