Generate 100 random data points from some hypothetical climate variable x~N(2,2) - in other words, from a normal distribution with u = 2 and o = 2. Now suppose there is an impact- relevant variable that is related to the climate variable via the following non-linear function: y = x. Calculate values of y from all the values of x, and then calculate the expected value of y (i.e. E(y)) Calculated the expected value of x (i.e. E(x)), and then calculate the expected value of y from this value(i.e. E(y) = [E(x)]). Question 2a (0.5%). Comment on whether the two different ways of calculating E(y) yield the same or different answers. Given that most real-world systems have a non-linear relationship with climate (or, indeed, with other drivers of risk), what does this suggest about the manner in which climate projections should be summarized to support impact modelling?