2017.3.13 Question 13

We have

V (x) = E [(X x)2] = E (X2 2xX + x2) = E (X2) 2xE (X) + x2 = σ2 + μ2 2 + x2.

Therefore, if Y = V (X), then

E (Y ) = E (V (X)) = E (σ2 + μ2 2 + X2) = σ2 + μ2 2μE (X) + E (X2) = σ2 + μ2 2μ2 + μ2 + σ2 = 2σ2.

Let X U[0,1], we have μ = E (X) = 1 2, and σ2 = Var (X) = 1 12. Therefore,

V (x) = 1 12 + 1 4 x + x2 = x2 x + 1 3.

The c.d.f. of X is F, defined as

P (X x) = F(x) = { 0,x 0, x,0 < x 1,

Let the c.d.f. of Y be G, we have G(y) = P (Y y).

Since V ([0,1]) = [ 1 12, 1 3 ], we must have G(y) = 0 for y 1 12 and G(y) = 1 for y > 1 3.

For y (1 12, 1 3 ], we have

G(y) = P (Y y) = P (V (X) y) = P ( (x 1 2 )2 + 1 12 y) = P ( |x 1 2 | y 1 12 ) = P (1 2 y 1 12 x 1 2 + y 1 12 ) = F (1 2 + y 1 12 ) F (1 2 y 1 12 ) = (1 2 + y 1 12 ) (1 2 y 1 12 ) = 2y 1 12.

Therefore, the p.d.f. of y, g satisfies that for y (1 12, 1 3 ],

g(y) = G(y) = 1 y 1 12

and 0 everywhere else.

Hence, we have

E (Y ) =yf(y) dy =1 12 1 3 y y 1 12 dy =y= 1 12 y=1 3 2y dy 1 12 = [2yy 1 12 ] 1 12 1 3 21 12 1 3 y 1 12 dy = [2yy 1 12 4 3 (y 1 12 )3 2 ] 1 12 1 3 = 2 1 3 1 2 4 3 1 8 = 1 6.

Also, 2σ2 = 2 1 12 = 1 6 = E (Y ), so the formula we derived holds in this case.