CHAPTER 1 Matrix Algebra ( Linear Algebra ) A matrix is a collection of elements arranged in an array of rows and columns. 2 1 3 1 2 4 Ex: A Amn [aij ] Ex: a12 1 i 1,2, . . m . , , j 1,2,..., m , a23 4 a11 a 12 . column vector . a n1 An1 Amn 1 0 B 0 1 , a11 a12 a a 22 21 . . . . a n1 a n 2 A1n [a11 , a12 ,..., a1n ] row vector ... a1n ... a 2 n ... . square matrix ... . ... a nn Matrix Operations: 1.Addation:adding elements in corresponding position.(element-by-element) 1 2 3 Ex: A23 2 0 4 , 2 4 5 B23 1 1 7 1 2 2 4 3 5 3 6 8 A B 2 1 0 1 4 7 3 1 11 2.Subtraction:(element-by-element) 1 2 2 4 3 5 1 2 2 2 1 0 1 4 7 1 1 3 1 2 1 1 2 , M 23 0 4 4 5 6 Ex: A B Ex: M 22 1 Note: Only matrices of the same size can be added or subtracted. 3.Multiplication: 1 2 3 3 6 9 2 0 4 6 0 12 Ex: 3 A 3 Ex: M 23 Multiply each a ij by 3 3 5 1 1 2 , N 32 0 1 2 0 3 2 4 1 3 1 0 2 2 1 5 1 1 2 4 7 14 P22 M N 2 3 0 0 3 2 2 5 0 1 3 4 12 22 註: Pqt M qr N rt 1 4 1 0 1 4 1 Ex: A 2 3 , B 2 1 1 , C 5 2 1 0 1 0 3 A B Can’t do it ! 1 4 4 5 1 1 4 2 24 9 A C 2 4 3 5 2 1 3 2 23 8 1 4 0 5 1 1 0 2 4 1 Note:乘法性質 1 0 2 1 , B 2 1 3 4 Ex: A 1 2 0 3 1 1 0 4 2 1 A B 2 2 1 3 2 1 1 4 7 6 2 1 1 2 2 0 1 1 4 1 B A 3 1 4 2 3 0 4 1 11 4 A B B A (也有相等的時候) 2 1 0 1 0 1 0 , B → A B B A (a b b a ) 2 1 0 1 2 1 ex: A 2. AB AC B C (ab ac b c) ( a 0 ) 1 1 1 1 1 0 , B , C 1 1 0 0 0 1 Ex: A 1 1 1 1 AB AC 1 1 1 1 but B C 3. A( BC ) ( AB)C Identity Matrix:單位矩陣 ( always square ) I×A=A×I=A I 22 1 0 0 1 I 33 1 0 0 0 1 0 0 0 1 ( 1 on diagonal and 0 elsewhere ) Ex: 2 5 8 A 3 6 9 4 7 10 A I 33 2 5 8 3 6 9 4 7 10 1 0 0 2 5 8 0 1 0 3 6 9 I A 0 0 1 4 7 10 1 0 0 2 1 3 2 1 3 0 1 0 Ex: B I 33 4 6 5 4 6 5 0 0 1 3 4. Transposition: Ex: A23 1 3 1 2 1 列 行 A' 32 2 0 3 0 4 1 4 Ex: B24 1 3 1 3 4 5 B ' 42 4 2 4 2 7 5 Note:1. ( A B)' A' B' 2 4 2 7 2. ( AB)' A' B' 3. ( AB)' B' A' 4. ( A' )' A 5. Matrix Partitioning: Ex: A44 B43 a11 a12 a a 22 21 a31 a32 a41 a 42 b11 b12 b b 21 22 b31 b32 b41 b42 a14 a24 A11 a34 A21 a44 a13 a23 a33 a43 b13 b23 B11 b33 B21 b43 A12 A22 B12 B22 Then, A A B 11 A21 A12 B11 A22 B21 B12 A11 B11 A12 B21 B22 A21 B11 A22 B21 A11 B12 A12 B22 A21 B12 A212 B22 Diagonal Matrices: 1.Always square ( m=n ) 2.with one or more nonzero elements on the main diagonal ([\]) and zeroes elsewhere. 2 0 Ex: 0 0 22 , 1 0 0 0 2 0 0 0 5 33 4 1 2 3 If M 4 5 6 , 7 8 9 If M Mˆ M , 1 0 0 then M̂ 0 5 0 0 0 9 0 2 3 then M 4 0 6 7 8 0 Ex: 3 0 0 1 2 3 3 10 3 1 3 2 5 3 2 0 5 0 4 1 2 12 5 4 4 1 2 0 0 2 2 0 1 2 3 2 4 6 2 1 2 3 0 3 4 1 2 12 3 6 3 4 1 2 Ex: Vectors: 1.Summation Vectors: 1 1 e : 1 or 1 1 1 : 1 1 1 2 3 1 2 3 6 Ex: 1 4 5 6 1 4 5 6 15 1 1 1 2 3 1 4 2 5 3 6 5 7 9 4 5 6 2.Inner Product: u v u v uv 5 1 4 Ex: u 2 , v 2 3 1 5 u v 3 2 1 1 15 2 4 21 4 3.Outer Product: u v uv 1 4 Ex: u 2 , v 2 3 1 1 4 2 1 u v u v 2 4 2 1 8 4 2 3 12 6 3 Linear Equation Systems Linear -Contain only variables to its first power - 2 x 5 , 3x1 2 x2 1 -No x 2 , x 3 , e x , ln t, sin x -No ( x1 x2 x3 )! 2 x 5 0 point 2 x1 3x2 7 0 line 3x1 x2 x3 5 0 plane Ex:Job Retraining Programs Total budget = $1,000,000 Program1:$2500/人 expect to earn $10000/yr Program2:$3500/人 expect to earn $20000/yr 6 Target tax revenue = $115000/yr (from program1 and program2) Tax rate = 2.5% 5x 7 x2 2000 Ans: 2500x1 3500x2 1000000 1 Ax B (1 0 0 0x0 1 1 5 0 0 0 2 2 0 0 0x0 2 ) 0.0 2 5 x1 2 x2 460 x 5 7 2000 , X 1 , B A 460 1 2 x2 Inverse:反矩陣 If AA1 A1 A I , then A1 is the inverse of A . 1 1 2 x 6 (2 x) 2 1 (2 x) 1 x (6) 3 2 2 1 1 Ax B A ( Ax) Ix x A B (1) Existence? 3 31 1 3 , 5 51 1 5 , 1 0 (x) 0 Some matrices may not have inverses , we call such matrices singular, otherwise , nonsingular (inverse exists!) (2) Uniqueness? Proof:Suppose B and C are inverses of A AC BA I B( AC ) ( BA)C B I I C BC How to fine the inverse A 1 ? 5 7 A 1 ? 1 2 Ex: A A A 1 I 7 s 5 7 s t 1 0 1 2 u v 0 1 t Let A 1 u v 5s 7u 1 1 2 u , s 3 3 s 2u 0 5s 7v 0 5 v 3 t 2v 1 a , t 2 A 1 3 1 3 7 3 7 3 5 3 a In general , A 11 12 A 1 ? a 21 a 22 s t Let A 1 u v a A A1 I 11 a21 a12 a22 s t 1 0 u v 0 1 a s a12u 1 ( a22 ) 11 a21 s a22 u 0 ( a12 ) a t a12 v 0 ( a22 ) 11 a21t a22 v 1 ( a12 ) ( a a a21a12 ) s a22 a22 11 22 s a11a22 a21a12 ( a11a22 a21a12 ) t a12 t u a 21 a11a 22 a 21a12 A1 , 1 a11a22 a21a12 v a12 a11a22 a21a12 a11 a11a 22 a 21a12 a22 a 21 a12 a11 1 3 2 4 Ex: A 8 A 1 2 4 3 1 4 6 2 1 1 1 3 A , 2 4 2 1 3 2 4 1 3 2 1 0 1 0 1 2 A 4 6 2 a12 a A 11 a21 a22 Note: A1 3 2 1 2 , A a11a22 a21a12 1 Some matrix A How to fine the determinant A ? 2 6 A 4 9 A 2 9 4 6 6 1 4 2 A 2 0 3 3 5 7 A 1 0 7 2 5 2 3 3 4 - 2 0 3 - 3 5 1 - 7 2 4 -15 Def: 1. Minor of an element a ij = m ij = the determinant of the matrix remaining when row i and column j are removed. 1 3 Ex: A m11 4 4 2 4 1 4 3 A 2 5 1 m 23 6 12 6 3 6 6 9 1 4 A m 22 1 1 3 6 Def 2:Cofactor of an element a ij = A ij (1) i j mij 1 3 A 12 (1) 2 4 Ex: A 1 2 4 2 3 A 0 5 2 A 31 (1) 1 4 1 Note:For any n n matrix 2 2 31 ( m12 2 ) (11) 11 ( m31 2 2 3 5 ) A n (1) A a ij A ij :for any row i j1 n (2) A a ij A ij :for any column j i 1 1 3 A 4 6 2 2 4 Ex: A Row 1:1 A 11 3 A 12 (1) 11 4 3 (1) 12 2 4 6 2 Column 2:3 A 12 4 A 22 6 4 (1) 2 2 1 6 4 1 4 2 Ex: A 2 1 5 3 A 31 3 A 33 3 18 3 (7) 33 (Row 3) 3 0 3 Def 3:The adjoint of a matrix A = adj A A ij 10 ( Cofactors of A ) 1 3 1 2 4 3 A adj A 2 4 3 4 2 1 Ex: A 1 4 2 Ex: A 2 0 1 3 5 6 5 14 4 adj A 9 0 3 10 7 8 1 adj A A Note: A1 A : Properties of 1. 1 2 3 A 4 0 5 2 1 6 A A 1 3 A 4 6 2 2 4 Ex: A 1 2 A A 4 6 2 3 4 2. If any row or column of A contains all zeroes, than A 0. 1 0 5 Ex: A 2 0 9 0 3 0 18 3. If B is obtained by multiplying some row or column of A by a constant C, than |B|=C |A|. 4. If any 2 rows or columns of A are interchanged, than |A| changes sign. 1 3 2 4 B , 1 3 2 4 Ex: A A 4 6 2 , B 642 11 5. If two or more rows or columns in A are equal, than | A |=0. 1 1 0 Ex: A 3 3 , 1 1 4 B 2 2 9 0 3 3 100 6. If 2 rows or columns in A are proportional, then | A |=0. 2 3 Ex: A 0 4 6 1 A 9 2 4 0 8 5 9 0 1 7 7. Evaluation of the determinant by alien cofactors always yields a value of zero. n a i j A i j 0 j1 n a A 0 ij ij i 1 n a ij A ij j1 A n a A ij ij i 1 1 1 1 Ex: A 2 0 6 3 7 1 f o r a n y i i f o r a n y j j for any i for any j A 2 ( 1 ) ( 6 ) 0 6 ( 1 ) ( 4) 1 2 Note: adj A A ij 1 5 1 6 6 5 A11 A adj A , 5 3 3 1 A 3 6 12 Ex: A 12 A 21 A 22 Theorem: A1 adj A A a11 a proof: A adj A 21 : a n1 n a1i A1j j1 n a 2i A1j j1 : n a ni A1j j1 a12 a 22 : an2 ... a1n A11 ... a 2n A 21 : : ... a nn A n1 n a1i A 2 j j1 n a j1 2i A2 j : n a j1 ni A2 j A12 A 22 : An 2 ... A1n ... A 2n : ... A nn A nj j1 A n ... a 2i A nj 0 j1 : : 0 n ... a ni A nj j1 n a ... 1i 0 A ... 0 ... A I So, A adj A A I A adj A adj A I A1 A A ( AA-1 A-1 A I ) 1 3 A1 ? Ex: A 3 7 7 3 adj A 3 1 7 3 7 3 adj A 3 1 2 2 A1 A 2 3 1 2 2 A 7 9 2 , Note:1. ( AB)1 B 1 A1 ( A1 B 1 ) 13 0 0 : A Is B1 A1 the inverse of AB ? ( M M 1 M 1 M I ) ( AB ) ( B 1 A1 ) A ( BB 1 ) A1 A A1 I ( B 1 A1 ) ( AB ) B 1 ( A1 A) B B B 1 I 1 1 Ex: B B 0 2 2 2 0 0 1 adj B 2 1 , , 1 2 B 1 0 0 1 2 B 1 1 2 1 1 3 7 1 1 3 1 4 4 1 AB ( AB) 7 4 17 7 17 17 3 4 4 0 1 2 B A 1 1 2 1 1 3 3 1 7 2 4 4 2 3 7 1 17 2 4 4 2 2. ( A)1 ( A1 ) 3 7 1 3 1 3 2 2 1 A (A ) Ex: A 3 1 3 7 3 7 2 2 3 3 7 7 2 2 2 2 1 1 A ( A ) 3 3 1 1 2 2 2 2 Note:1. For any pair of square matrices A and B , |AB|=|A|‧|B|. 2. If matrix B is obtained from A by adding a multiple of one row of A to another row of A, then 1 3 Ex: A 2 4 , 1 3 B 4 10 14 B A . A 4 6 2 , B 10 12 2 2 3 1 1 Ex: 3 2 1 2 4 2 5 2 1 0 5 8 0 2 4 0 1 11 3 1 1 0 1 1 7 3 4 5 (1) 2 3 1 8 4 7 3 1 11 4 47 13 (1) 0 18 1 11 5 4 235 234 1 Gauss-Jordan Method for Computing Inverses A → A-1 =? [ A I ] BA BI A-1A A-1I [ I A-1 ] Def:Elementary Row Operation (e.r.o.) . Note:Performing Elementary Row Operations to a matrix is equivalent to pre-multiply the matrix by another matrix. 1 0 1 1 1 1 1 2 1 1 -1 1 = , Ex: 1 1 = 1 1 0 3 1 1 3 3 0 1 2 1 1 Ex: 4 6 0 2 7 2 15 1 0 0 1 0 1 0 0 0 0 1 0 2 1 1 4 6 0 2 7 2 1 0 0 1 0 0 3 1 2 8 8 0 0 1 2 2 1 0 0 1 0 1 0 1 1 2 2 3 1 0 16 1 0 0 1 1 0 0 1 0 4 8 1 1 1 0 0 1 3 8 1 4 1 8 3 4 1 2 1 1 2 1 0 1 2 1 4 1 5 3 16 8 3 1 8 4 1 1 e.r.o. : 1. Multiply some row by a constant. 2. Add some multiple of one row to another row. Using Matrix Partition to find inverses: 1 3 Ex: A 1 6 2 0 4 4 3 1 5 7 A Suppose A= 11 A 21 AA-1=I 1 2 2 8 A12 A 22 AR I A 11 A 21 R 12 R 22 A12 R11 A 22 R 21 R12 I 0 R 22 0 I A11R11 A12 R 21 I A11R 12 A12 R 22 0 A 21R11 A 22 R 21 0 A 21R12 A 22 R 22 I R So , A R 11 R 21 -1 R , then A-1=R= 11 R 21 R12 [A11 A12 A 22 1A 21 ]1 R 22 A 22 1A 21R11 16 R11A12 A231 A 22 1[I A 21R12 ] 0 1 1 0 4 8 1 1 1 1 2 0 Gramer’s rule: Ex:5X1+7X2=2000 2000 7 460 2 X1 = 5 7 1 2 ; X1+2X2=460 5 2000 1 460 , X2 5 7 1 2 X1 b1 2000 5 7 A , X= , B= X b 460 1 2 2 2 AX=B A 1AX A 1B IX A 1B 1 X adj A B A X1 A11 A 21 X 2 X 1 A12 A 22 A Xn A1n 1 X i (A1i b1 A 2i b 2 A Ex:X1= 1 (A1i b1 A 2i b 2 A Gramer’s rule: X i A Bi A A n1 A n 2 A nn b1 b2 b n A ni b n ) A ni b n ) Where ABi is obtained from A by replacing the ith column of A by B. a11 a A 21 a n1 a1i a 2i a ni a1n a11 a a 2n , A Bi = 21 a nn a n1 a12 a 22 b1 b2 an2 bn 17 a1n a 2n A Bi b1A1i b 2 A 2i a nn b n A ni In AX=B, where A is n×n : If | A |≠0 , then “ unique solution ”. If | A |=0 , then either no solution or infinite # of solution ( depends on B ). Theorem:A -1 exists iff | A |≠0 Note:A -1= 1 adj A A , If | A |=0, A -1 does not exist sin gular. The Geometry of Vectors: Def:Given n vectors V1 ,V2 , … ,Vn , each containing m elements, and given n scalars α1, α2 , … , αn , then n V0=α1V1+α2V2+…+αnVn= i Vi is called the Linear Combination i 1 ( L.C ) of { V1,V2,…,Vn}. 1 3 Ex: V1 , V2 2 4 1 3 7 V0 2V1 3V2 2 3 2 4 16 If αi≧0, i=1,2, … ,n , then V0 is called a nonnegative L.C. of { V1,V2,…,Vn}. Ex:V0=2V1+4V2 n If αi≧0 , i=1,2, … ,n , and i 1, then V0 is called a Convex i 1 Combination (C.C.) of { V1,V2,…,Vn }. 18 1 3 2 3 Ex:V0= V1+ V2 a11 x1 a12 x 2 b1 a 21 x1 a 22 x 2 b 2 Ax B a A 11 a 21 a12 x b , x 1 , B 1 a 22 x 2 b2 V1 1 V2 2 V0 x A 1 A 2 1 B A 1 x1 A 2 x 2 B ( L.C. of columns of A ) x2 ( A i i th column of A ) 向量 常數 In general , n×n system of equations , Ax=B A1x1 A2 x 2 An x n B ( B is a L.C. of {A1 , A2 , … , An}) . C1 C Length of a vector C= 2 . C n C C 1 C C12 C 22 C 2 C C12 C22 C2n , n C i 1 C1 d1 C1 d1 C d C d 2 2 1 Distance between 2 vectors C and D , C-D 1 C n d n C n d n C D (C1 d1 ) 2 (C2 d 2 ) 2 (C n d n ) 2 n (C i 1 i di )2 C+D C C C-D C-D D D 19 2 i Angle between 2 vectors C and D: C D C D cos C D cos θ 1 - 1 , D , cos 2 1 Ex: C (1)(-1) (-2)(1) (1) (2) ( 1) (1) 2 2 2 2 CD C D 1 10 Area enclosed by 2 vectors: C d C 1 , D 1 C 2 d 2 C D 2 7 C1 C2 d1 C1d1 C 2 d 2 d2 7 2 38 Ex: C , D , Area 2 6 6 2 Area = | C D|. (Why?) (2.6,1.5) 1 0 , basis (基底) : 0 1 20 2 6 1 0 1 5 2.6 0 1.5 1 (natural basis) -2 1 0 3 (2) 0 3 1 1 3 Ex: basis? 2 4 -2 1 3 3 1 2 1 4 -2=α1+3α2 3=2α1+4α2 α1= 17 2 , α2= 7 2 => NO! 1 3 Ex: , basis? 2 6 2 1 3 3 1 2 2 6 Note: 1.The set of Vectors must be non-parallel in order to be a basis. 2. The set of Vectors must be Linearly independent in order to be a basis. 3.The set of Vectors must span the whole vector space in order to be a basis. 21 Linear Independence and Linear Dependence Def:The vectors V1 ,V2 , … ,Vn in m-dimensional space are linearly dependent ( L.D.) if there are n scalars α1, α2 , … , αn , Not all zero , such that α1V1+α2V2+…+αnVn=0. If the only set of αi’s such that α1V1+α2V2+…+αnVn=0 holds is α1=α2= … =αn=0, then V1,V2 , … ,Vn are Linearly Independence( L.I. ). 1 3 Ex: V1 , V2 2 4 3 2 0 1 3 0 1 2 1 2 4 0 21 4 2 0 1 2 0 V1 , V2 are L.I. 1 -2 Ex: V1 , V2 2 -4 2 2 0 1 2 0 1 2 1 2 4 0 21 4 2 0 choose 1 2 , 2 1 1 6 , 2 3 V1 , V2 are L.D. Theorem:Let V be an n-dimensional space and let S={X1,X2,…Xn} be a set of vectors in V. Then:(1) if S is L.I. , then S is a basis for V.(∵|S|=n) (2) if S spans V,….., S is L.I. 22 Theorem:If S={X1,X2,…Xn} is a basis for V , then any vector in V can be represented as one and only one L.C. of X1,X2,…Xn. Def:The rank of a Matrix A , R(A) or ρ(A) is the max. number of L.I. columns or rows in A . 1 2 3 Ex: A R(A) 2 2 1 1 1 2 3 Ex: B R(B) 1 ( 2 4 6 不能成為倍數) 1 2 2 Ex: C R(C) 2 2 7 4 1 2 3 Ex: A 2 5 6 R(A) 2 2 4 6 Note:(1) R(A)=R(A’) (2) If A is m×n , then ρ(A)≦min(m,n) Quadratic Forms: -important in minimization and maximization 2-Var 二次式: X1 X2 Let X= Q(X1,X2)=6 X12 +4X1X2+3 X 22 X1 6 2 , A= 2 3 Q(X , X ) X 'AX 1 2 a12 a A= 11 (a12 =a 21 ) Symmetric a 21 a 22 6 2 X1 X2 2 3 X 2 3-Var 三次式: 23 Q(X1,X2,X3)=3 X12 -4 X 22 +5 X 32 +2X1X2-6X1X3+8X2X3 X1 3 1 3 X X 2 , A= 1 -4 4 X3 3 4 5 Q(X1 , X 2 , X 3 ) X'AX Q(X1,X2)≧0 for all X1,X2 ? .……… ≦ ……………… Ex:2 X12 +3 X 22 ≧0 Q(X1,X2) = a11 X12 +2a12X1X2+a22 X 2 2 = a11 X12 +2a12X1X2+ a12 2 2 a 2 X 2 a 22 X 2 2 12 X 2 2 a11 a11 a12 a12 a11a 22 a12 2 2 X1X 2 ( X 2 ) ]+[ = a11[ X +2 ] X 22 a11 a11 a11 2 1 a1 1a 2-a2 The sign of Q(X1 ,X 2) depends on the signs of a 1 and 1 a11 2 1 2 . Note: (1) Let A=[A1,A2,…,An] , then , |A|≠0 if and only of A1,A2,…,An are L.I. (2) If A n is a Linear Combination of A 1,A 2,…,A n-1, then A 1,A 2,…,A n-1, A n are L.D. An is a L.C. of A 1,A 2,…,A n-1 A n 1A1 2 A 2 1A1 2 A 2 A1 ,A 2 , n 1A n 1 n 1A n 1 A n 0 ,A n-1 ,A n are L.D. 24 (3) The max. number of L.I. Vectors in m-dimensional space is m . 1 3 Ex: L.I 2 4 1 1 3 - 6 2 4 7 N o t L . I. 3 4 ; 5 1 2 1 6 - 1 7 1 6 Not L.I. Def:A set of vectors S={X1,X2,…Xn} in a Vector Space V is called a basis if (1) S spans V and (2) S is L.I. 1 0 0 1 0 1 2 0 Ex: S X1 , X2 = , X3 , X4 = 1 -1 2 0 0 2 1 1 The show (2): 1X1 2 X 2 3 X3 4 X 4 0 1 0 0 1 0 0 1 2 0 0 1 2 3 4 0 1 1 2 1 0 0 2 1 0 0 4 1 2 2 3 1 2 2 3 2 2 3 4 0 0 =0 0 1 2 3 4 0 The show (1): 25 S is L.I. a 1 0 0 1 b 0 1 2 0 Let X 0 1 2 3 4 c 1 1 2 0 d 0 2 1 1 Definitions: 1.If Q is always positive (Q>0) for all values of Xi’s (except when all Xi’s are zero) , then Q is positive definite (PD). 2……………….negative(Q<0)……………………………………………… ……….. negative definite (ND). 3……………… nonnegative (Q≧0)………………………………………... ……. positive semi-definite (PSD). 4.……………….nonpositive (Q≦0)………………………………………... …….. negative semi-definite (NSD). Def:Leading Principal Minors of a matrix A. The kth order leading principal minor of a matrix is the determinant of the matrix made up of the first k rows and columns. a11 a12 Ex: A a 21 a 22 a 31 a 32 a13 a 23 , n=3 a 33 a11 a12 A3 a 21 a 22 a 31 a 32 A1 a11 a11 A2 a11 a 21 a12 a11a 22 a 21a12 a 22 26 a13 a 23 = A a 33 A1 a1 1 1 1 2 4 Ex: 2 1 0 0 3 3 A2 1 2 1 4 5 2 1 1 2 A3 2 1 0 3 4 0 3 Def:Principal Minors of a matrix A. A principal minor of a matrix is the determinant of the matrix remained after a particular set of rows and the corresponding columns are removed. A kth-order principal minor of A is the determinant of the matrix remaining when all but k rows and corresponding columns are removed. a11 Ex: a 21 a 31 a12 a 22 a 32 a13 a 23 a 33 a1 1 a 21 a 3 1 a 1 2 a 1 3 a 2 2 a 2 3 a 3 2 a 3 3 (1,2) |A1(1)|=|a33|=a33 (1,3) |A1(2)|=|a22|=a22 (2,3) |A1(3)|=|a11|=a11 a (3) |A2(3)|= 11 a 21 a12 a 22 a11 |A3|= a 21 a 31 a (1) |A2(1)|= 22 a 32 a 23 a 33 a (2) |A2(2)|= 11 a 31 a13 a 33 a12 a 22 a 32 a13 a 23 =|A| a 33 Note:LPM PM Theorem: 1.A quadratic form, X’AX , is (a) PD if and only if all LPMs of A are positive ( |A1|>0 , |A2|>0 , …. , |An|>0 ). (b) ND if and only if the LPMs of A alternate in sign beginning negative. ( |A1|<0 , |A2|>0 , |A3|<0,…..) 27 A 2.A quadratic form, X’AX, is (a) PSD if and only if all PMs of A are nonnegative. ( all |A1|≧0 , all |A2|≧0 , … , all |An|≧0 ) (b) NSD if and only if all odd-order PMs of A are nonpositive. ( all |A1|≦0 , all |A3|≦0 , all |A5|≦0 , …………) and all even-order PMs of A are nonnegative. ( all |A2|≧0 , all |A4|≧0 , all |A6|≧0 , …………) Ex: Ex: 2X12 6X1X2 5X22 2X12 4X1X 2 6X1X 3 8X 2 X 3 X 2 2 4X 32 X1 X ' AX , X X 2 2 3 X 1 [ X 1 X 2 ] 3 5 X 2 X1 X2 2 2 3 X1 X 3 2 1 4 X 2 3 4 4 X 3 Ex1:X12+6X22 X1 1 0 X1 X2 0 6 X 2 A1 1 0 A2 1 0 60 0 6 all LPM>0 PD Ex2:-X12-6X22 X1 1 0 X1 X2 0 6 X 2 A1 | 1| 1 0 A2 1 0 60 0 6 , ,..... ND Note:If Q(X1,X2) is PD , then –Q(X1,X2) is ND. Ex3:X12-4X1X2+4X22+X32 PSD 28 PD NSD ND X1 X2 A1 1 1 2 0 X1 X 3 2 4 0 X 2 0 0 1 X 2 A2 1 2 0 2 4 Neither PD nor ND. A1 (1) 1 , A1 (2) 4 , A1 (3) 1 all 0 A 2 (1) 4 0 1 0 1 2 4 , A 2 (2) 1 , A 2 (3) 0 0 1 0 1 2 4 1 2 0 A 3 2 4 0 =0 0 0 1 all 0 all 0 PSD Ex4:-X12+4X1X2-4X22-X32 1 2 0 X1 X1 X 2 X3 2 4 0 X 2 0 0 1 X 2 A1 1 A2 0 Neither PD nor ND. A1 (1) 1 , A1 (2) 4 , A1 (3) 1 all 0 A 2 (1) 1 2 1 0 4 0 0 , A 2 (2) 1 , A 2 (3) 4 2 4 0 1 0 1 1 2 0 A 3 2 4 0 =0 all 0 NSD 0 0 1 Ex:X12-6X22 X1 1 0 X1 X2 0 6 X 2 A1 1 A2 1 0 6 0 6 A1 (1) 1 , A1 (2) 6 Neither PD nor ND Neither PSD nor NSD indefinite 29 all 0