1 Special Types of SQUARE Matrices
Total Page:16
File Type:pdf, Size:1020Kb
1 Special Types of SQUARE Matrices Name Description Examples 2 3 Diagonal a = 0 for any i 6= j 4 0 0 0 0 ij 2 3 3 0 0 60 0 0 0 07 6 7 Notation: That is, all entries 40 1 05 or 60 0 11 0 07 6 7 Diag(a11; :::; ann) off the main diagonal 0 0 7 40 0 0 6 05 must be 0. Diago- 0 0 0 0 2 nal entries can be These matrices can be denoted by Diag(3; 1; 7) anything, including 0. and Diag(4; 0; 11; 6; 2), respectively. Identity (I ) A diagonal matrix n 21 0 03 whose main-diagonal 1 0 I = 0 1 0 or I = entries are all 1. 3 4 5 2 0 1 0 0 1 Scalar A diagonal matrix of 25 0 0 03 the form r(In) for 60 5 0 07 −3 0 5(I4) = 6 7 or −3(I2) = some scalar r. 40 0 5 05 0 −3 0 0 0 5 Upper Triangular A = [aij] with aij = 0 for all i > j 25 4 8 23 That is, entries 21 2 43 0 2 1 3 \below" the main 0 3 1 or 6 7 4 5 60 0 0 17 diagonal must be 0. 0 0 2 4 5 0 0 0 5 Entries on or \above" the main diagonal can be anything, including 0. Lower Triangular A = [aij] with aij = 0 for all i < j 24 0 0 0 03 2 3 That is, entries 9 0 0 61 3 0 0 07 6 7 \above" the main 44 2 05 or 60 4 5 0 07 6 7 diagonal must be 0. 1 3 8 41 3 5 9 05 Entries on or \below" 5 6 2 9 4 the main diagonal can be anything, including 0. Symmetric AT = A 21 2 3 4 5 3 2 3 4 12 3 62 0 7 8 9 7 (only applies or 6 7 to matrices with 412 1 85 or 63 7 2 6 107 3 8 5 64 8 6 3 1 7 real entries) aij = aji 4 5 5 9 10 1 7 Skew Symmetric AT = −A 2 0 7 6 −13 2 0 1 −33 (only applies or −7 0 0 2 −1 0 8 or 6 7 to matrices with 4 5 6−6 0 0 −47 3 8 0 4 5 real entries) aij = −aji 1 −2 4 0 NOTE: In the definitions of symmetric and skew symmetric matrices, we require that all entries be real numbers. For the other types of matrices mentioned, entires could (technically) be complex numbers, but we won't work with complex entries until later in the semester. Moreover, entries do not all have to be integers, even though the examples include only integer entries. This was merely for convenience and clarity. Question: Notice that in the examples I gave of skew symmetric matrices, the entries on the main diagonal were all 0. Does this always happen for skew symmetric matrices? If so, why? Question: What types of matrices can be both upper and lower triangular? Question: I claim that for any n, there is only one n × n square matrix which is both symmetric and skew symmetric. Am I right? If so, what is the unique matrix with this property? Question: Consider an n × n diagonal matrix A = Diag(a11; a22; :::; ann). What happens when you take powers of A? Compute A2 and A3. Can you make a guess at a formula for Ap for any positive integer p? Fact: If A is an m × n matrix, then ImA = A and AIn = A. Fact: (This is Exercise 1.5.29 in the book.) If A is an n × n square matrix, then we can show that A = S + K, where S is symmetric and K is skew symmetric. That is, A can be decomposed as the sum of a skew matrix and a skew symmetric matrix. Also, this decomposition is unique. 2 Powers of Matrices • A2 should mean AA. If A is size m × n, then AA = A2 only makes sense if n = m. In other words, we can only take powers of square matrices. • If A is a square matrix of size n × n and if p is a positive integer, then Ap = AA ··· A | {z } p times 0 • By convention, if A is size n × n, we say that A = In. • FACT: If p and q are nonnegative integers and A is a square matrix, then ApAq = Ap+q and (Ap)q = Apq • CAUTION: (AB)p and ApBp may not be (and usually are not) the same. (See Exercise 1.5.9) 3 Partitioned Matrices and Block Multiplication Please read about these topics in Section 1.5 of the textbook. It starts at the bottom of Pg. 43 and ends in the middle of Pg. 46. Then try out block multiplication by doing Exercise 1.5.56. Block multiplication helps computers complete matrix multiplication more quickly. However, when you are doing matrix multiplication by hand, block multiplication doesn't have major advantages..