Chapter 4 - MATRIX ALGEBRA. ... a 2j... a 2n. a i1 a i2... a ij... a in

Transcription

1 Chapter 4 - MATRIX ALGEBRA 4.1. Matrix Operations A a 11 a a 1j... a 1n a 21. a a 2j... a 2n. a i1 a i2... a ij... a in... a m1 a m2... a mj... a mn The entry in the ith row and the jth column of a matrix A is refered to as (A) ij. EXAMPLE: Algebra 2017/ A zero matrix is a matrix, written 0, whose entries are all zero. A square matrix has the same number of rows than columns. In general (m n), matrices are rectangular. The (main) diagonal of a matrix, or its diagonal entries, are the entries A diagonal matrix has all its nondiagonal entries equal to zero Algebra 2017/

2 A matrix is upper triangular if all its elements under the diagonal are zero A matrix is lower triangular if all its elements over the diagonal are zero The set of all possible matrices of dimension (m n) whose entries are real numbers is refered to as R m n The set of all possible matrices of dimension (m n) whose entries are complex numbers is refered to as C m n K 3 2 Algebra 2017/ OPERATIONS: Only for matrices with the same dimensions: Equality. Two matrices are equal if and only if their corresponding entries are equal Addition. A matrix whose entries are the sum of the corresponding entries of the matrices Algebra 2017/

3 Scalar Multiplication. A matrix whose entries are the corresponding entries of the matrix multiplied by the scalar PROPERTIES: Let A, B and C be matrices of K m n and λ, µ K: A + B B + A A+(B+C)(A+B)+C A + 0 A λ (A+B) λ A+λ B (λ+µ) A λ A+µ A λ (µ A) (λ µ) A Algebra 2017/ Matrix Multiplication K p K n K m One wonders: Does C exist C x A B x x K p? PROBLEM: What dimensions would C have? Algebra 2017/

4 If we write B [ b 1 b 2... b p and x x 1. x p, then: Bx x 1 b 1 + x 2 b x p b p A(Bx) Algebra 2017/ Let A be an (m n) matrix and let B be an (n p) matrix with columns b 1, b 2,..., b p. The matrix product of A by B is the (m p) matrix AB whose columns are Ab 1, Ab 2,..., Ab p. That is, AB A [ b 1 b 2... b p [ Ab 1 Ab 2... Ab p Warning: The dimensions of the matrices involved in a product must verify A B C ( ) ( ) ( ) Algebra 2017/

5 EXAMPLE: [ [ ( ) ( ) ( ) [ [ [ [ [ Algebra 2017/ Row-Column Rule for computing AB Consider A K m n, and B [ b 1... b p K n p such that (A) ik a ik, and (B) kj b kj. That is, AB [ Ab 1 Ab j Ab p {}}{ a 11 a a 1n 1.. b 1j a i1 a i2... a in b 2j.... i b. (AB) ij nj a m1 a m2... a mn (AB) ij a i1 a i2... a in m b 1j b 2j. b nj k a ik b kj Algebra 2017/

6 EXAMPLE: [ [ [ 1st row 3rd column ( 1, 3 ) entry [ [ [ 2nd row 1st column ( 2, 1 ) entry Algebra 2017/ PROBLEM: Find the 2nd row of AB. AB PROBLEM: Compute [ [ Algebra 2017/

7 PROPERTIES: Let A be an (m n) matrix, and B and C matrices of appropriate dimensions: A(BC) (AB)C A(B + C) AB + AC (B + C)A BA + CA µ (AB) (µ A) B A (µ B) µ K I m A A A I n where I k is the (k k) identity matrix 4.3 Algebra 2017/ WARNING: In general, AB BA EXPANSION AXIS X ROTATION 30 1st EXPANSION + 2nd ROTATION [ 2 0 B A AB 3 ROTATION 30 o EXPANSION AXIS X 1st ROTATION + 2nd EXPANSION [ A B BA Algebra 2017/

8 WARNING: In general, AB AC / B C ROTATION π/2 PROJECTION in X 1st ROTATION + 2nd PROJECTION B A AB REFLECTION x+y 0 PROJECTION in X 1st REFLECTION + 2nd PROJECTION C A AC Algebra 2017/ WARNING: In general, AB 0 / A 0 or B 0 PROJECTION in X PROJECTION in Y 1st X-PROJECTION + 2nd Y-PROJECTION B A AB WARNING: In general, A 2 0 / A 0 [ A [, A A 2 If two square matrices verify that AB BA, we say that A and B commute with each other. Algebra 2017/

9 The kth power of a matrix is defined: A k A A A A }{{} k times This only makes sense if A is a nonnegative integer. matrix and k is a For convenience, we define A 0 I. PROBLEM: Compute 4.4 Algebra 2017/ Transpose of a Matrix The transpose of an (m n) matrix A is the (n m) matrix A T whose columns are the rows of A. That is, EXAMPLE: B (A T ) ij (A) ji B T EXAMPLE: A symmetric matrix verifies A T A. An antisymmetric matrix verifies A T A. PROBLEM: Provide examples of (anti)symmetric matrices. Algebra 2017/

10 PROPERTIES: Let A and B be matrices of appropriate dimensions and µ K: (A T ) T A (µ A) T µ (A T ) (A + B) T A T + B T (AB) T B T A T Proof: Let be A K m n and B K n q ( (AB) T ) ij PROBLEM: Prove that (ABC) T C T B T A T. 4.7 Algebra 2017/ Conjugate Transpose of a Matrix The conjugate transpose of an (m n) matrix A is the (n m) matrix A, or A H, whose elements verify: (A ) ij (A) ji. EXAMPLE: 5 2 i B i B A [ a 1 a 2 a n A Algebra 2017/

11 PROPERTIES: Let A and B be matrices of appropriate dimensions and µ K: (A ) A (A + B) A + B (µ A) µ (A ) (AB) B A A A T if and only if A is a real matrix. A Hermitian matrix verifies A A. An antihermitian matrix verifies A A. PROBLEM: Provide examples of (anti)hermitian matrices. 4.8 Algebra 2017/ Inverse of a Matrix A square (n n) matrix A is invertible, or nonsingular, if there exists a matrix B such that AB I n A noninvertible or singular matrix has no inverse. 2 5 EXAMPLE: This matrix is invertible: A Because C verifies AC Algebra 2017/

12 [ cos φ sin φ EXAMPLE: This matrix is invertible: A sin φ cos φ A A 1 Thus, A 1 EXAMPLE: Matrix B has no inverse and is, therefore, a singular matrix: B 4.9 Algebra 2017/ Theorem 4.1. If A is an invertible (n n) matrix, then the equation Axb has the unique solution xa 1 b, b K n. Proof: That x A 1 b is a solution b can be checked by a mere substitution: As it has a solution b A must have a pivot in every row. A square No free variables Warning: Algebra 2017/

13 Theorem 4.2. Let A and B be (n n) matrices. Then: AB I BA I Proof: ( AB I BA I ) Suppose that BA X Let s define M I X [ m 1 m 2 m n. As That is, But now, Leading to Algebra 2017/ Theorem 4.3. If A is an invertible matrix, then A 1 is invertible and (A 1 ) 1 A. Proof: Theorem 4.4. If exists, the inverse of a matrix is unique. Proof: Let A be an invertible matrix, and B a matrix such that AB I (that is, B A 1 ). Suppose there exists C such that AC I (in other words, suppose that A has another inverse). Algebra 2017/

14 Theorem 4.5. If A is invertible, A T is also invertible and (A T ) 1 (A 1 ) T. Theorem 4.6. If A is invertible, A is also invertible and (A ) 1 (A 1 ). Proof: EXAMPLE: [ 1+i 1+2i 1 1 i [ 1 i 1 2i i then, [ 1 + i i 1 i 1 Algebra 2017/ Theorem 4.7. If A and B are invertible (n n) matrices, then AB is invertible and (AB) 1 B 1 A 1. Proof: EXAMPLE: Consider the linear transformations: A ROTATE B EXPAND. Then, AB (in this order!) and the inverse is (AB) PROBLEM: If A, B and C are nonsingular matrices of equal size, show that (ABC) 1 C 1 B 1 A 1. Algebra 2017/

15 An elementary matrix is one that is obtained by performing one elementary row operation on an identity matrix. EXAMPLE: E E E Notice: These matrices have a clear geometrical interpretation. They correspond to Algebra 2017/ Theorem 4.8. If an elementary row operation if performed on an (m n) matrix A, the resulting matrix can be written as EA, where E is the (m m) elementary matrix created by performing the same operation on I m. EXAMPLE: Consider the (3 2) matrix A a d b e c f I E 1 ( r 3 5 r 3 ) E 1 A a d b e c f Algebra 2017/

16 I E 2 ( r 2 r 3 ) E 2 A a d b e c f I E 3 ( r 2 r 2 4r 1 ) E 3 A a d b e c f A A A Algebra 2017/ Theorem 4.9. Every elementary matrix E is invertible and its inverse E 1 is the elementary matrix corresponding to the row operation that transforms E back into I. EXAMPLE: The matrix E 1 multiplies the 3rd row by five: E Its inverse E1 1 is the matrix that divides the 3rd row by five: E 1 1 Check: E 1 E 1 1 I 4.12 PROBLEM: Find the matrices E 1 2 and E 1 3. Algebra 2017/

17 Theorem An (n n) matrix A is invertible if and only if A is row equivalent to I n. In this case, any sequence of elementary row operations that transforms A into I n also transforms I n in A 1. Proof: A invertible Then, A 1 E p E p 1... E 2 E 1 and, in fact, 4.14 Algebra 2017/ An Algorithm for finding A 1 Construct the matrix [ A I Find its reduced echelon form. If this matrix has the form [ I B, then A 1 B. Otherwise, A does not have an inverse. EXAMPLE: }{{} }{{} }{{}}{{} A I Algebra 2017/

18 }{{}}{{} I }{{}}{{} }{{}}{{} A 1 Algebra 2017/ PROBLEM: If exists, find the inverse of the matrix C [ C I 4.16 Check: C C 1 Algebra 2017/

19 Theorem (The Square Matrix Theorem) If A K n n, the following statements are equivalent: 1. A is an invertible matrix. 2. There exists C K n n such that AC I n. 3. There exists D K n n such that DA I n. 4. A is row equivalent to I n. 5. A has n pivots. 6. The equation Ax 0 has only the trivial solution. 7. The columns/rows of A are linearly independent. 8. The equation Ax b has a (unique) solution b K n. 9. The columns/rows of A span K n. 10. The columns/rows of A form a basis of K n 11. A T is invertible. 12. A is invertible. 13. The linear transformation x Ax is bijective. 14. Col A Row A K n 15. dim Col A dim Row A n 16. rank A n 17. Nul A {0} 18. dim Nul A 0 A transformation T : K n K n is called invertible if there exists a transformation S : K n K n such that } S(T (x)) x x K n. T (S(x)) x The transformation S is called the inverse of T. Theorem Let T : K n K n be a linear transformation and A its canonical matrix. T is invertible if and only if A is nonsingular. In this case, S(x) A 1 x Algebra 2017/

20 4.3. Partitioned (or Block) Matrices EXAMPLE: A A where A 11, A 12, A 13 A 21, A 22, A 23 Algebra 2017/ EXAMPLE: Social web of 6 persons in 3 groups Adjacency Matrix M M 11 M 12 M 13 M 21 M 22 M 23 M 31 M 32 M 33 Algebra 2017/

21 EXAMPLE: Jefferson High School Algebra 2017/ EXAMPLE: Trade share matrix between countries Algebra 2017/

22 PROPERTIES: Addition: Matrices of equal size and identical partition can be summed block by block: A + B [ A11 A 12 A 13 A 21 A 22 A 23 [ B11 B + 12 B 13 B 21 B 22 B 23 Scalar Multiplication: [ A11 A λa λ 12 A 13 A 21 A 22 A 23 Algebra 2017/ Transpose of a matrix: [ A11 A A 12 A 13 A 21 A 22 A 23 A T A T 11 A T 21 A T 12 A T 22 A T 13 A T 23 Conjugate transpose of a matrix: A 11 A 21 A11 A A 12 A 13 A A A 21 A 22 A A 22 A 13 A 23 EXAMPLE: A A T Algebra 2017/

23 Multiplication of partitioned matrices: Two matrices A and B of respective dimensions (m n) and (n p) are conformable for block multiplication when the number of columns of each partition of A is equal to the number of rows of the corresponding partition of B. AB [ A11 A 12 A 21 A [ B11 B (Attention: ) Algebra 2017/ Concentrate on the dimensions of the blocks: (2 3) ( ) ( ) (3 5) (5 2) ( ) ( ) ( ) (2 3)(3 2) + ( )( ) ( )( ) + ( )( ) (2 2) + ( ) ( ) + ( ) ( ) ( ) [ ( ) Algebra 2017/

24 EXAMPLE: Let A be a block upper triangular matrix: A A11 A A 22 Assuming that A is invertible, A 11 is (p p) and A 22 is (q q), find a formula for A 1. Call B A 1. Partition B in such a way that we can write: A11 A AB 12 B11 B 12 I 0. 0 A 22 B 21 B 22 0 I The dimensions of the matrices involved are: (p p) ( ) ( ) ( ) ( ) (q q) ( ) ( ) [ ( ) ( ) ( ) ( ). Algebra 2017/ The equation can be written: [ I 0 0 I. Equating components, we obtain: (a) I (b) 0 (c) 0 (d) I We have to solve 4 matrix equations, which represent a linear system of (p + q) 2 equations with (p + q) 2 unknowns. Algebra 2017/

25 (d) (c) (a) (b) Obtaining, A 1. Algebra 2017/ Theorem A block diagonal matrix is invertible if and only if each of the diagonal blocks is invertible. Proof: The case of two blocks follows from the above result when A C C C nn Algebra 2017/

26 Theorem A diagonal matrix is invertible if and only if none of its diagonal elements is zero. 1 a a a nn 4.19 PROBLEM: Determine under what conditions the following matrix is invertible and, in that case, find its inverse: Im 0. A I n Algebra 2017/ Determinants Given an (m n) matrix A, we define the minor A ij as the ((m 1) (n 1)) matrix obtained by removing the ith row and the jth column of the matrix A. EXAMPLE: A Let A be an (n n) matrix whose entry (A) ij a ij. We define the determinant of A as det A A n ( 1) j+1 a 1j det A 1j j1 n a 1j C 1j, j1 where C ij ( 1) i+j det A ij is refered to as the ij cofactor of A. Algebra 2017/

27 Theorem The determinant of a square matrix A can be expressed as the cofactor expansion along any row of the matrix n n ( ) det A ( 1) k+j along the a kj det A kj a kj C kj j1 j1 kth row WARNING: Algebra 2017/ EXAMPLE: 1st row: det nd row: Algebra 2017/

28 Theorem If A is an (n n) triangular matrix, its determinant is the product of its diagonal entries. det a a a a 44 0 a 55 Algebra 2017/ Theorem Let A be an (n n) matrix. If we obtain a matrix B, By adding to a row of A the multiple of another row, det B det A. By multiplying one row of A by λ, det B λ det A. By interchanging two rows of A, det B det A. EXAMPLE: Algebra 2017/

29 Theorem Let A be a square matrix and U an echelon matrix obtained from A by adding multiples of rows and r row interchanges (but without multiplying any row by a scalar!). Then, 0 if A is not invertible det A ) ( 1) r if A is invertible ( product of the pivots 4.20 Proof: Note: This would add a new statement to theorem 4.11: 19. The determinant of A is nonzero. Algebra 2017/ WARNING: In general, Check theorem 4.17! A B / det A det B. WARNING: In general, det(a + B) det A + det B. EXAMPLE: If it was true, all determinants would be zero: ([ a b a 0 det det c d b ) c 0 0 d Algebra 2017/

30 Theorem If A and B are square matrices, det(ab) det A det B. Theorem If A is a square matrix, A T A and A A Proof: For elementary matrices, it s easy to see that E E T. If we obtain an echelon form of a matrix A: Leading to Now, as U is a triangular matrix, U T U and, consequently 4.23 Algebra 2017/