Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Linear Algebra Grinshpan Patterns of matrix multiplication When the number of columns of a matrix A agrees with the number of rows of a matrix B, we may speak of the product matrix AB. The (i, j)-entry of AB is obtained by multiplying together ith row of A and jth column of B : b 1j .. = a b + . . . + a b . (AB)ij = ai1 . . . ain i1 1j in nj . bnj AB has as many rows as A and as many columns as B. The matrix product has many properties and patterns. Some are listed below. These hardly require memorization, but do require some recognition. A row times a column is a number, but a column times a row is a full xa xb x x y a b c = ya yb a b c y = ax + by + cz za zb z z grown matrix: xc yc zc Nonzero matrices may have zero product: 1 −1 1 1 0 0 0 1 0 1 0 0 = = −1 1 1 1 0 0 0 0 0 0 0 0 A zero ∗ ∗ ∗ ∗ 0 0 row of the ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ 0 A zero ∗ ∗ ∗ ∗ ∗ ∗ column ∗ ∗ ∗ ∗ ∗ ∗ first factor is ∗ ∗ ∗ ∗ = ∗ ∗ ∗ 0 0 inherited by the product: ∗ ∗ 0 of the second factor is inherited by the product: 0 ∗ ∗ 0 ∗ 0 ∗ = ∗ 0 ∗ 0 ∗ ∗ 0 ∗ The a 0 0 product of diagonal matrices 0 0 x 0 0 ax 0 b 0 0 y 0 = 0 by 0 c 0 0 z 0 0 is again a diagonal matrix: 0 0 cz The ∗ 0 0 product of upper (lower) triangular matrices is again upper (lower) triangular: ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ 0 ∗ ∗ = 0 ∗ ∗ 0 0 ∗ 0 0 ∗ 0 ∗ Multiplication a 0 0 1 0 b 0 1 0 0 c 1 by a diagonal 1 1 a 1 1 = b 1 1 c matrix on the left has a scaling effect on rows: a a b b c c Multiplication 1 1 1 a 1 1 1 0 1 1 1 0 by a diagonal 0 0 a b 0 = a 0 c a matrix on the right has a scaling effect on columns: b c b c b c Left multiplication by an identity matrix with switched rows switches rows: 0 1 0 a b c d e f 1 0 0 d e f = a b c 0 0 1 g h i g h i Right multiplication by an identity matrix with switched columns switches columns: b a c 0 1 0 a b c d e f 1 0 0 = e d f h g i 0 0 1 g h i Left multiplication by a shear matrix has the corresponding effect on rows: a+d b+e c+f 1 1 0 a b c 0 1 0 d e f = d e f add row 2 to row 1 g h i g h i 0 0 1 a b c 1 0 0 a b c 0 1 0 d e f = d e f subtract row 1 from row 3 −a + g −b + h −c + i g h i −1 0 1 Right multiplication by a shear matrix has the corresponding effect on columns: a b c 1 1 0 a a+b c d e f 0 1 0 = d d + e f add column 1 to column 2 g h i 0 0 1 g g+h i a b c 1 0 0 a + 2c b c d e f 0 1 0 = d + 2f e f add 2 columns 3 to column 1 g h i 2 0 1 g + 2i h i Column-row expansion If C1 , . . . , Cn are the columns of A and R1 , . . . , Rn are the rows of B, then AB = C1 R1 + . . . + Cn Rn . a b a b x y ax + bz ay + bt ax ay bz bt x y + z t = = = + c d c d z t cx + dz cy + dt cx cy dz dt