Section 9 – Functions and Transformations of Random Variables Distribution of a transformation of continuous RV: X • Y = u(X) – Y is defined as a function “u” of X • v(u(x))=x – Function “v” is defined as function the inverse function of “u” – Obtain v(u(x)) by solving the given Y=u(x) for x fY (y) f X (v(y)) | v'(y) | Distribution of a Sum of RV’s • If Y = X1 + X2 – E[Y] = E[X1] + E[X2] – Var[Y] = Var[X1] + Var[X2] + 2Cov[X1, X2] • Use the convolution method to find the distribution of a sum of independent RV’s – Note: X1 & X2 must be independent P[X1 X 2 k] P[X1 X 2 Y] k f (x , x ) 1 fY (y) 2 x1 0 f (x1, x 2 )dx1 k f (x ,k x ) 1 1 x1 0 k f (x ) f (x ) 1 1 2 2 x1 0 k f (x ) f (k x ) 1 x1 0 1 2 1 f (x1, y x1 )dx1 f1 (x1 ) f 2 (x 2 )dx1 f1 (x1 ) f 2 (y x1 )dx1 Central Limit Theorem • If asked for a probability involving a sum of a large number of independent RV’s, you usually need to use the normal approximation – X1, X2, … Xn are independent RV’s with the same distribution • X has mean and standard deviation E[Yn ] n Var[Yn ] n 2 Y n~ N(n,n 2 ) Distribution of Maximum or Minimum of a Collection of Independent RV’s • Suppose X1 and X2 are independent RV’s – U = max{X1, X2} – V = min{X1, X2} (trickier) – We know F1(x) and F2(x) such that F1(x)=P(X1<=x) FU (u) P[U u] P[max{X1, X 2} u] P[(X1 u) (X 2 u)] (independent) P[X1 u] P[X 2 u] F1(u) F2 (u) FU (u) F1 (u) F2 (u) FV (v) P[V v] 1 P(V v) 1 P[min{X1, X 2 } v] 1 P[(X1 v) (X 2 v)] (independent) 1 P[X1 v] P[X 2 v] 1 [1 F1 (v)][1 F2 (v)] FV (v) 1 [1 F1 (v)][1 F2 (v)] Mixtures of RV’s • X1 and X2 are RV’s with independent (but usually different) probability functions • X is a mixture of X1 and X2 – Where 0<a<1 – Weight “a” on X1 – Weight (1-a) on X2 • Expectation, probabilities, and moments follow a “weighted-average” form: E[X] aE[X ] (1 a)E[X ] 1 2 E[X 2 ] aE[X12 ] (1 a)E[X 22 ] FX (x) aF1 (x) (1 a)F2 (x) M X (t) aMX 1 (t) (1 a)M X 2 (t) • Variance is NOT calculated from a “weighted-average approach” • Use the above approach to find E[X] & E[X^2] Var[X] aVar[X1 ] (1 a)Var[X 2 ] Var[X] E[X 2 ] (E[X])2 Mixtures of RV’s • There is a difference between a “sum of RV’s” and a “mixture of RV’s” – Sum of RV’s: X = X1 + X2 – Mixture of RV’s: X is defined by probability function: f (x) a f (x) (1 a) f (x) X 1 2 • So, don’t make the mistake of thinking: X = a*X1 + (1-a)*X2 (this is wrong)