Let X1, X2, X3 be uniform random variables on the interval (0, 1) with Cov(Xi , Xj ) = 1/24 for i, j = 1, 2, 3, i i != j. Calculate the variance of 2X1 + X2 − X3.
Since "X_1, X_2, X_3" are uniform random variables on the interval (0, 1), their probability distribution functions are,
"f(X_1)=f(X_2)=f(X_3)= 1,\\space 0\\lt X_i\\lt1, \\space i=1,2,3"
"0, \\space elsewhere"
"E(X_1)=\\displaystyle\\int^1_0 X_1dx_1={X_1^2\\over 2}|^1_0={1\\over2}"
"E(X_1^2)=\\displaystyle\\int^1_0 X_1^2dx_1={X_1^3\\over3}|^1_0={1\\over3}"
"Var(X_1)=E(X_1^2)-(E(X_1))^2={1\\over3}-({1\\over2})^2={1\\over12}"
Since "X_2" and "X_3" have the same probability distribution function as "X_1", they both will have the same variance as "X_1", that is,
"Var(X_1)=Var(X_2)=Var(X_3)={1\\over12}" and
"Cov(X_i,X_j)={1\\over24}\\space for\\space i=1,2,3,\\space j=1,2,3, \\space i\\not=j"
Now,
"Var(2X_1 + X_2 \u2212 X_3)=E[(2X_1 + X_2 \u2212 X_3)-(2\\mu_1+\\mu_2-\\mu_3)]^2" , from definition
"=E[2(X_1-\\mu_1)+(X_2+\\mu_2)-(X_3-\\mu_3)]^2"
"=4Var(X_1)+Var(X_2)+Var(X_3)+4Cov(X_1,X_2)-4Cov(X_1,X_3)-2Cov(X_2,X_3)"
"=4({1\\over12})+{1\\over12}+{1\\over12}+4({1\\over24})-4({1\\over24})-2({1\\over24})={6\\over12}-{2\\over24}={5\\over12}"
Therefore, the variance of 2X1 + X2 − X3 is "{5\\over12}".
Comments
Leave a comment