Solutions Stat755 Homework 4 4.1 consider a bivariate normal distribution with μ1 = 1, μ2 = 3, σ 11 = 2, σ 22 = 1 and ρ12 = −.8. (a) Write out the bivariate normal density. (b) Write out the squared statistical distance expression ( x − μ ) ' Σ −1 ( x − μ ) as a quadratic function of x1 and x2. Sln: (a) the density function of bivarate normal distribution is the following: f ( x1 , x2 ) = = ⎧ ⎡ exp ⎨− 2 1−1ρ 2 ⎢ ( 2 12 ) ⎣ ⎩ 2π σ 11σ 22 (1 − ρ12 ) 1 5 6π 2 { 25 ⎡ exp − 18 ⎢⎣ ( ) x1 −1 2 2 ( ) + ( ) − 2ρ ( )( )⎤⎥⎦ ⎭⎫⎬ x1 − μ1 + ( x2 − 3) + 85 2 2 x2 − μ 2 σ11 2 σ 22 ( )( x x1 −1 2 2 12 x1 − μ1 x2 − μ 2 σ11 σ 22 } − 3) ⎤ ⎥⎦ (b) The quadratic funciton is the following: ( x − μ ) ' Σ −1 ( x − μ ) = 1 2 1− ρ12 ( ⎡ ⎢⎣ ⎡ ) ⎢⎣ ( ) + ( ) − 2ρ ( )( )⎤⎥⎦ x1 − μ1 ( ) x1 −1 2 σ11 2 x2 − μ2 2 σ 22 + ( x2 − 3) + 85 x1 − μ1 x2 − μ2 σ11 σ 22 12 ( )( x x1 −1 − 3) ⎤ ⎥⎦ = 25 9 = 475 20 2 25 x1 20 2 x1 25 x12 50 x2 20 2 x2 20 25 x22 + − − + − − + 2 x1 x2 + 18 3 9 3 18 3 9 9 9 2 2 2 2 4.2 consider a bivariate normal population with μ1 = 0, μ2 = 2, σ 11 = 2, σ 22 = 1 and ρ12 = .5. (a) Write out the bivariate normal density (b) Write out the squared generalized distance expression ( x − μ ) ' Σ −1 ( x − μ ) as a function of x1 and x2. (c) Determin (and sketch) the constant‐density contour that contains 50% of the probability. Sln: (a) the density function of bivarate normal distribution is the following: 1 Solutions Stat755 f ( x1 , x2 ) = = ⎧ ⎡ exp ⎨ − 2 1−1ρ 2 ⎢ ⎩ ( 12 ) ⎣ 2π σ 11σ 22 (1 − ρ122 ) 1 1 2π 3 2 { exp − 23 ⎡ ⎣⎢ ( ) x1 2 2 Homework 4 ( ) ( ) − 2ρ ( )( )⎤⎥⎦ ⎭⎫⎬ 2 x1 − μ1 σ11 + ( x2 − 2 ) − 1⋅ 2 ( )( x σ 22 x1 2 2 2 x2 − μ2 + 12 x1 − μ1 x2 − μ2 σ11 σ 22 } − 2)⎤ ⎦⎥ (b) ( x − μ ) ' Σ −1 ( x − μ ) ( ) + ( ) − 2ρ ( )( )⎤⎥⎦ ( ⎡ ) ⎢⎣ = 23 ⎡ ⎢⎣ ( ) = 1 2 1− ρ12 x1 2 x1 − μ1 σ11 2 2 x2 − μ2 2 σ 22 12 + ( x2 − 2 ) − 2 ( )( x x1 2 2 x1 − μ1 x2 − μ2 σ11 σ 22 − 2)⎤ ⎥⎦ 8 2 2 x1 x12 8 x2 1 2x2 = + + − − 2 x1 x2 + 2 3 3 3 3 3 3 (c) The solid ellipsoid of (x1,x2) satisfy ( x − μ ) ' Σ −1 ( x − μ ) ≤ χ p2 (α ) = c 2 will has probability 50%, where p = 2. So, α = 50%. From distribution table, we have χ 22 ( 0.5 ) =1.39, therefore, c = 1.18. σ 11 σ 21 ⎤ ⎡ ⎡σ The covariance matrix Σ = ⎢ 11 =⎢ ⎥ ⎣σ 21 σ 22 ⎦ ⎣⎢ ρ 21 σ 11σ 22 ρ12 σ 11σ 22 ⎤ ⎡ 2 σ 22 ⎥=⎢1 ⎦⎥ ⎣ ⎤ . 1 ⎥⎦ 1 2 2 ⎛ ⎛ ⎡ 0.460 ⎤ ⎞ ⎡ −.888⎤ ⎞ The eigen pairs are ⎜ 2.37, ⎢ ⎟ and ⎜ .634, ⎢ ⎥ ⎟ . Therefore, we have the axes ⎥ .460 .0.888 − − ⎣ ⎦⎠ ⎣ ⎦ ⎝ ⎝ ⎠ as: c λ1 = 1.18 ⋅ 2.37 = 1.81 and c λ2 = 1.18 ⋅ 0.634 = 0.937. 2 Solutions Stat755 Homework 4 ⎡ 1 −2 0 ⎤ 4.3 Let X be N 3 ( μ , Σ ) with μ ' = [ −3 1 4] and Σ = ⎢ −2 5 0 ⎥ . Which of the ⎢ ⎥ ⎢⎣ 0 0 2 ⎥⎦ following random variables are independent? Explain. (a) X1 and X2. (b) X2 and X3 . (c) (X1,X2) and X3. (d) X 2 − 52 X 1 − X 3 . X1 + X 2 and X3. (e) X2 and 2 Sln: (a) σ 12 = σ 21 = −2, X1 and X2 are dependent. (b) σ 13 = σ 31 = 0, X2 and X3 are independent. ⎡ X1 ⎤ ⎡ Y1 ⎤ ⎡1 1 0 ⎤ ⎢ ⎥ ⎡ X 1 + X 2 ⎤ ⎡1 1 0 ⎤ (c) Let A = ⎢ ⎥ . Therefore, ⎥ , so ⎢Y ⎥ = ⎢0 0 1 ⎥ ⎢ X 2 ⎥ = ⎢ X ⎣0 0 1 ⎦ ⎦⎢ ⎥ ⎣ 3 ⎣ 2⎦ ⎣ ⎦ ⎣ X3 ⎦ ⎡ Y1 ⎤ T ⎢Y ⎥ ∼ N ( Aμ , AΣA ) . ⎣ 2⎦ ⎡ 1 −2 0 ⎤ ⎡1 0 ⎤ ⎡1 0 ⎤ ⎡1 1 0 ⎤ ⎢ ⎡ −1 3 0 ⎤ ⎢ ⎥ ⎢ ⎥ ⎥ = ⎡2 0⎤ . AΣA = ⎢ − = 2 5 0 1 0 1 0 ⎥ ⎢ ⎥ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎣0 0 1 ⎦ ⎢ 0 0 2 ⎥ ⎢0 1 ⎥ ⎣ 0 0 2 ⎦ ⎢0 1 ⎥ ⎣0 2 ⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦ T σ 12 = 0, Thus, (X1,X2) and X3 are independent. ⎡ X1 ⎤ ⎡ 12 12 0 ⎤ ⎡ Y1 ⎤ ⎡ 12 12 0 ⎤ ⎢ ⎥ ⎡ 12 X 1 + 12 X 2 ⎤ (d) Let A = ⎢ ⎥ . Therefore, ⎥ , so ⎢Y ⎥ = ⎢0 0 1 ⎥ ⎢ X 2 ⎥ = ⎢ X3 ⎣0 0 1 ⎦ ⎣ 2⎦ ⎣ ⎦⎢X ⎥ ⎣ ⎦ ⎣ 3⎦ ⎡Y1 ⎤ ⎢Y ⎥ ∼ N ( Aμ , ⎣ 2⎦ AΣAT ) . ⎡ 1 −2 0 ⎤ ⎡ 12 0 ⎤ 0 ⎡ − 12 ⎡ ⎤⎢ ⎢ ⎥ ⎥ T 1 AΣA = ⎢ ⎥ ⎢ −2 5 0 ⎥ ⎢ 2 0 ⎥ = ⎢ ⎣0 0 1 ⎦ ⎢ 0 0 2 ⎥ ⎢0 1 ⎥ ⎣ 0 ⎣ ⎦⎣ ⎦ 1 2 1 2 σ 12 = 0, Thus, X1 + X 2 and X3 are independent. 2 3 ⎡ 1 0⎤ 0 ⎤ ⎢ 12 ⎡ 12 0 ⎤ ⎥ 0⎥ = ⎢ . 0 2 ⎥⎦ ⎢ 2 0 2 ⎥⎦ ⎣ ⎢⎣0 1 ⎥⎦ 3 2 Solutions Stat755 Homework 4 ⎡ X1 ⎤ X2 ⎡ 0 1 0⎤ ⎤ ⎡ Y1 ⎤ ⎡ 0 1 0 ⎤ ⎢ ⎥ ⎡ , so ⎢ ⎥ = ⎢ 5 (e) Let A = ⎢ 5 X = 2 ⎥ ⎥ ⎢ ⎥ . Therefore, 5 ⎢ ⎥ ⎣Y2 ⎦ ⎣ − 2 1 −1⎦ ⎢ X ⎥ ⎣ − 2 X 1 + X 2 − X 3 ⎦ ⎣ − 2 1 −1⎦ ⎣ 3⎦ ⎡ Y1 ⎤ T ⎢Y ⎥ ∼ N ( Aμ , AΣA ) . ⎣ 2⎦ ⎡0 AΣAT = ⎢ 5 ⎣− 2 ⎡0 − 52 ⎤ ⎡ 1 −2 0 ⎤ ⎡0 − 52 ⎤ 1 0 ⎤⎢ ⎡ −2 5 0 ⎤ ⎢ ⎡ 5 10 ⎤ 1 1 ⎥⎥ = ⎢ . −2 5 0 ⎥⎥ ⎢⎢1 1 ⎥⎥ = ⎢ 9 ⎥ ⎥ ⎢ ⎢ 1 −1⎦ 10 23 14 ⎥⎦ − 2 10 −2 ⎦ ⎣ ⎣ ⎢⎣0 −1 ⎥⎦ ⎢⎣ 0 0 2 ⎥⎦ ⎢⎣0 −1 ⎥⎦ σ 12 = 10, Thus, X1 + X 2 and X3 are dependent. 2 ⎡1 1 1 ⎤ 4.4 Let X be N 3 ( μ , Σ ) with μ ' = [ 2 −3 1] and Σ = ⎢1 3 2 ⎥ . ⎢ ⎥ ⎢⎣1 2 2 ⎥⎦ (a) Find the distribution of 3X1 2X2+ X3. (b) Relabel the variables if necessary, and find a 2 by 1 vector a such that X2 and ⎡X ⎤ X 2 − a ' ⎢ 1 ⎥ are independent. ⎣ X3 ⎦ ⎡ X1 ⎤ Sln: (a) Let a ' = [3 −2 1] , so [3 −2 1] ⎢ X 2 ⎥ = [3 X 1 ⎢ ⎥ ⎢⎣ X 3 ⎥⎦ [3 X 1 −2 X 2 −2 X 2 X 3 ] . Therefore, ⎡2⎤ X 3 ] . ∼ N ( a ' μ , a ' Σa ) . Where a ' μ = [3 −2 1] ⎢ −3⎥ = 13, ⎢ ⎥ ⎣⎢ 1 ⎦⎥ ⎡1 1 1 ⎤ ⎡ 3 ⎤ ⎡3⎤ ⎢ ⎥ ⎢ ⎥ a ' Σa = [3 −2 1] ⎢1 3 2 ⎥ ⎢ −2 ⎥ = [ 2 −1 1] ⎢⎢ −2 ⎥⎥ = 9. The distribution is N (13, 9 ) . ⎢⎣1 2 2 ⎥⎦ ⎢⎣ 1 ⎥⎦ ⎢⎣ 1 ⎥⎦ (b) Let a ' = [ a1 ⎡ X1 ⎤ a2 ] , so Y = X 2 − a ' ⎢ ⎥ = X 2 − [ a1 ⎣ X3 ⎦ ⎡ 0 1 0 ⎤ ⎡X ⎤ Therefore, A = ⎢ . Thus, ⎢ 2 ⎥ ∼ N ( Aμ , ⎥ ⎣Y ⎦ ⎣ −a1 1 −a2 ⎦ 4 ⎡X ⎤ a2 ] ⎢ 1 ⎥ = [ −a1 X 1 ⎣ X3 ⎦ AΣAT ) . Where, X2 −a2 X 3 ] . Solutions Stat755 Homework 4 ⎡1 1 1 ⎤ ⎡0 − a1 ⎤ ⎡ 0 1 0 ⎤⎢ AΣA = ⎢ 1 3 2 ⎥⎥ ⎢⎢1 1 ⎥⎥ ⎥ ⎢ ⎣ − a1 1 −a2 ⎦ ⎢1 2 2 ⎥ ⎢0 − a ⎥ 2⎦ ⎣ ⎦⎣ T ⎡0 − a1 ⎤ ⎤⎢ 1 1 ⎥⎥ 3 − a1 − 2a2 2 − a1 − 2a2 ⎥⎦ ⎢ ⎢⎣0 −a2 ⎥⎦ 3 3 − a1 − 2a2 ⎡ ⎤ =⎢ ⎥ 2 2 ⎣3 − a1 − 2a2 a1 − 2a1 − 4a2 + 2a1a2 + 2a2 + 3⎦ 1 ⎡ =⎢ ⎣1 − a1 − a2 3 2 ⎡X ⎤ Since 3 − a1 − 2a2 = 0 implies X2 and X 2 − a ' ⎢ 1 ⎥ are independent. So we have vector ⎣ X3 ⎦ ⎡3⎤ ⎡ −2 ⎤ a = ⎢ ⎥ + c ⎢ ⎥ ,c ∈ . ⎣0⎦ ⎣1⎦ ⎡ 4 0 −1⎤ 4.6 Let X be N 3 ( μ , Σ ) with μ ' = [1 −1 2] and Σ = ⎢ 0 5 0 ⎥ . Which of the following ⎢ ⎥ ⎢⎣ −1 0 2 ⎥⎦ random variables are independent? Explain. (a) X1 and X2. (b) X1 and X3. (c) X2 and X3. (d) (X1, X3) and X2. (e) X1 and X1 + 3 X 2 − 2 X 3. Sln: (a) σ 12 = σ 21 = 0, X1 and X2 are independent. (b) σ 13 = σ 31 = −1, X1 and X3 are dependent. (c) σ 23 = σ 32 = 0, X2 and X3 are independent. ⎡ X1 ⎤ ⎡ Y1 ⎤ ⎡1 0 1 ⎤ ⎢ ⎥ ⎡ X 1 + X 3 ⎤ ⎡1 0 1 ⎤ (d) Let A = ⎢ ⎥ . Therefore, ⎥ , so ⎢Y ⎥ = ⎢0 1 0 ⎥ ⎢ X 2 ⎥ = ⎢ X ⎣0 1 0 ⎦ ⎦ ⎢X ⎥ ⎣ ⎣ 2⎦ ⎣ ⎦ 2 ⎣ 3⎦ ⎡ Y1 ⎤ T ⎢Y ⎥ ∼ N ( Aμ , AΣA ) . ⎣ 2⎦ ⎡ 4 0 −1⎤ ⎡1 0 ⎤ ⎡1 0 ⎤ ⎡1 0 1 ⎤ ⎢ ⎡3 0 1⎤ ⎢ ⎥ ⎢ ⎥ ⎥ = ⎡4 0⎤ . = AΣA = ⎢ 0 5 0 0 1 0 1 ⎥ ⎢ ⎥ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎣0 1 0 ⎦ ⎢ −1 0 2 ⎥ ⎢1 0 ⎥ ⎣0 5 0 ⎦ ⎢1 0 ⎥ ⎣ 0 5 ⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦ T σ 12 = 0 Thus, (X1,X3) and X2 are independent. 5 Solutions Stat755 Homework 4 ⎡ X1 ⎤ X1 ⎡ Y1 ⎤ ⎡1 0 0 ⎤ ⎢ ⎥ ⎡ ⎤ ⎡1 0 0 ⎤ (e) Let A = ⎢ , so ⎢ ⎥ = ⎢ ,⎢X2 ⎥ = ⎢ ⎥ . Therefore, ⎥ ⎥ ⎣1 3 −2 ⎦ ⎣Y2 ⎦ ⎣1 3 −2 ⎦ ⎢ X ⎥ ⎣ X 1 + 2 X 3 − 2 X 2 ⎦ ⎣ 3⎦ ⎡ Y1 ⎤ T ⎢Y ⎥ ∼ N ( Aμ , AΣA ) . ⎣ 2⎦ ⎡ 4 0 −1⎤ ⎡1 1 ⎤ ⎡1 1 ⎤ ⎡1 0 0 ⎤ ⎢ ⎡ 4 0 −1⎤ ⎢ ⎡4 6 ⎤ ⎥ ⎢ ⎥ 0 5 0 ⎥ ⎢0 3 ⎥ = ⎢ 0 3 ⎥⎥ = ⎢ . AΣA = ⎢ ⎥ ⎥ ⎢ ⎢ 6 15 −5⎦ 6 61⎥⎦ ⎣1 3 −2 ⎦ ⎢ ⎣ ⎣ ⎢⎣0 −2 ⎥⎦ ⎣ −1 0 2 ⎥⎦ ⎢⎣0 −2 ⎥⎦ T σ 12 = 6, Thus, X1 and X 1 + 3 X 2 − 2 X 3 are dependent. 6