13
Matrix Operations: Determinant

Matrix Operations: Determinant

  • Upload
    others

  • View
    24

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Matrix Operations: Determinant

Matrix Operations: Determinant

Page 2: Matrix Operations: Determinant

Determinants

โ€ข Determinants are only applicable for square matrices.

โ€ข Determinant of the square matrix ๐ด is denoted as:

det(๐ด) or ๐ด

โ€ข Recall that the absolute value of the determinant of a 2 ร— 2 matrix is equal to the area of parallelogram of the rows of that matrix.

โ€ข Similarly, the absolute value of the determinant of a 3 ร— 3 matrix is equal to the volume of parallelepiped of the rows of that matrix.

โ€ข Therefore, the absolute value of the determinant of a ๐‘› ร— ๐‘› matrix is equal to the n-dimensional volume, constructed by the rows of that matrix.

Page 3: Matrix Operations: Determinant

Determinant of a 2 ร— 2 matrix

โ€ข Recall that:

๐ด =๐‘Ž11 ๐‘Ž12๐‘Ž21 ๐‘Ž22

, ๐ด =๐‘Ž11 ๐‘Ž12๐‘Ž21 ๐‘Ž22

= ๐‘Ž11๐‘Ž22 โˆ’ ๐‘Ž12๐‘Ž21.

๐’‚1

๐’‚2

๐’‚๐Ÿ๐’‚๐Ÿ

Page 4: Matrix Operations: Determinant

Determinant of a 3 ร— 3 matrix

โ€ข Also recall the determinant for a 3 ร—3 matrix:

โ€ข ๐‘… =

๐‘Ÿ11 ๐‘Ÿ12 ๐‘Ÿ13๐‘Ÿ21 ๐‘Ÿ22 ๐‘Ÿ23๐‘Ÿ31 ๐‘Ÿ32 ๐‘Ÿ33

โ€ข If the row vectors are linearlydependent, then the determinantis zero, and the matrix is NOT invertible.โ€ข Notice if the row vectors arelinearly dependent the volumewill be zero, as the vectors lie on a plane on a line.

๐’“1๐’“2๐’“3

Page 5: Matrix Operations: Determinant

Determinant of a 3 ร— 3 matrixโ€ข To compute the determinant of a 3 ร— 3 matrix,.

โ€ข The first element in the top row is multiplied with the determinant of the sub-matrix resulting from removing the (first) row and the (first) column corresponding to that element from the matrix.

โ€ข The negate of second element in the top row is multiplied with the determinant of the sub-matrix resulting from removing the (first) row and the (second) column corresponding to that element from the matrix.

โ€ข The third element in the top row is multiplied with the determinant of the sub-matrix resulting from removing the (first) row and the (third) column corresponding to that element from the matrix.

โ€ข ๐‘… =

๐‘Ÿ11 ๐‘Ÿ12 ๐‘Ÿ13๐‘Ÿ21 ๐‘Ÿ22 ๐‘Ÿ23๐‘Ÿ31 ๐‘Ÿ32 ๐‘Ÿ33

= ๐‘Ÿ11๐‘Ÿ22 ๐‘Ÿ23๐‘Ÿ32 ๐‘Ÿ33

โˆ’ ๐‘Ÿ12๐‘Ÿ21 ๐‘Ÿ23๐‘Ÿ31 ๐‘Ÿ33

+ ๐‘Ÿ13๐‘Ÿ21 ๐‘Ÿ22๐‘Ÿ31 ๐‘Ÿ32

=

๐‘Ÿ11 ๐‘Ÿ22๐‘Ÿ33 โˆ’ ๐‘Ÿ23๐‘Ÿ32 โˆ’ ๐‘Ÿ12 ๐‘Ÿ21๐‘Ÿ33 โˆ’ ๐‘Ÿ23๐‘Ÿ31 + ๐‘Ÿ13 ๐‘Ÿ21๐‘Ÿ32 โˆ’ ๐‘Ÿ22๐‘Ÿ31

Page 6: Matrix Operations: Determinant

Determinant of a 3 ร— 3 matrix / Cofactor

โ€ข In the determinant of a 3 ร— 3 matrix, we multiplied the first row elements in their corresponding cofactors.

โ€ข The cofactor of the element ๐‘–, ๐‘— of ๐‘› ร— ๐‘› matrix ๐ด is:๐ถ๐‘–๐‘— = (โˆ’1)๐‘–+๐‘—det๐‘€๐‘–๐‘—

โ€ข Where ๐‘€๐‘–๐‘— is submatrix after removing row ๐‘– and column ๐‘—.โ€ข Determinant of ๐ด is:

det๐ด = ๐‘Ž๐‘–1๐ถ๐‘–1 + ๐‘Ž๐‘–2๐ถ๐‘–2 +โ‹ฏ+ ๐‘Ž๐‘–๐‘›๐ถ๐‘–๐‘›โ€ข In the above formula the row ๐‘– could be any row of ๐ด and it is not

necessarily the first row.โ€ข In fact it need not be a row. It can be any column j. โ€ข (So in order to compute the determinant, it is always wise to choose the

row or a column that has most number of zeroes and compute the cofactor of only its non-zero elements.)

Page 7: Matrix Operations: Determinant

Determinant properties

โ€ข The determinant of identity matrix is 1.๐ผ = 1

โ€ข The determinant changes sign when two rows are exchanged.๐‘ ๐‘‘๐‘Ž ๐‘

= โˆ’๐‘Ž ๐‘๐‘ ๐‘‘

โ€ข The determinant is a linear function of each row separately.๐‘ก๐‘Ž ๐‘ก๐‘๐‘ ๐‘‘

= ๐‘ก๐‘Ž ๐‘๐‘ ๐‘‘

๐‘Ž + ๐‘Žโ€ฒ ๐‘ + ๐‘โ€ฒ

๐‘ ๐‘‘=

๐‘Ž ๐‘๐‘ ๐‘‘

+๐‘Žโ€ฒ ๐‘โ€ฒ

๐‘ ๐‘‘

Page 8: Matrix Operations: Determinant

Determinant properties

โ€ข If one row is a scalar multiple of another row then det(๐ด) = 0

๐‘Ž ๐‘๐‘ก๐‘Ž ๐‘ก๐‘

= 0๐‘Ž ๐‘ ๐‘๐‘‘ ๐‘’ ๐‘“๐‘ก๐‘Ž ๐‘ก๐‘ ๐‘ก๐‘

= 0

๐‘Ž ๐‘ ๐‘๐‘‘ ๐‘’ ๐‘“

๐‘Ž + ๐‘‘ ๐‘ + ๐‘’ ๐‘ + ๐‘“= 0,

๐‘Ž ๐‘ ๐‘๐‘‘ ๐‘’ ๐‘“

2๐‘Ž + ๐‘‘ 2๐‘ + ๐‘’ 2๐‘ + ๐‘“= 0

๐‘Ž ๐‘ ๐‘๐‘‘ ๐‘’ ๐‘“

2๐‘Ž + 5๐‘‘ 2๐‘ + 5๐‘’ 2๐‘ + 5๐‘“= 0

Page 9: Matrix Operations: Determinant

Determinant properties

โ€ข Row reduction does not change the determinant of ๐ด๐‘Ž ๐‘

๐‘ โˆ’ ๐›พ๐‘Ž ๐‘‘ โˆ’ ๐›พ๐‘=

๐‘Ž ๐‘๐‘ ๐‘‘

๐›พ is a non-zero scalar

โ€ข A matrix with a row of zeros has det(๐ด) = 0๐‘Ž ๐‘0 0

= 0

Page 10: Matrix Operations: Determinant

Determinant properties

โ€ข If ๐ด is a triangular then the determinant is the product of diagonal elements.

๐‘Ž ๐‘0 ๐‘‘

= ๐‘Ž๐‘‘,๐‘Ž 0๐‘ ๐‘‘

= ๐‘Ž๐‘‘

This is also applicable for diagonal matrices:๐‘Ž 0 00 ๐‘ 00 0 ๐‘

= ๐‘Ž๐‘๐‘

โ€ข If ๐ด is singular (columns or rows are linearly dependent) det(๐ด) = 0

โ€ข ๐ด๐ต = ๐ด ๐ต

โ€ข ๐ด๐‘‡ = ๐ด

Page 11: Matrix Operations: Determinant

Rank of Matrix

โ€ข Let ๐‘š = min ๐‘Ÿ๐‘œ๐‘ค, ๐‘๐‘œ๐‘™๐‘ข๐‘š๐‘›

โ€ข Rank of matrix is the size of the largest square sub-matrix with non-zero determinant.

โ€ข Matrix is full-ranked, if its rank = m.

โ€ข Matrix is rank-deficient, if its rank < m.

โ€ข It is not possible to have matrixโ€™s rank > m.

Page 12: Matrix Operations: Determinant

Sub-Matrix

โ€ข In order to find the rank of matrix we should find the largest quaresub-matrix with non-zero determinant.

โ€ข For making a sub-matrix we are allowed to remove rows or columns of a matrix

โ€ข Example: A is a 5 ร— 3 matrix

โ€ข Removing two rows of A๐‘Ÿ๐‘œ๐‘ค1๐‘Ÿ๐‘œ๐‘ค2๐‘Ÿ๐‘œ๐‘ค3๐‘Ÿ๐‘œ๐‘ค4๐‘Ÿ๐‘œ๐‘ค5

=๐‘Ÿ๐‘œ๐‘ค2๐‘Ÿ๐‘œ๐‘ค4๐‘Ÿ๐‘œ๐‘ค5

Page 13: Matrix Operations: Determinant

Matrix Rank

โ€ข Example: Find the rank of matrix A

๐ด =0 1 21 2 12 7 8

Row 1 and Row 2 of matrix A are linearly independent. However Row 3 is a linear combination of Row 1 and 2.

๐‘Ÿ๐‘œ๐‘ค3 = 3 ร— ๐‘Ÿ๐‘œ๐‘ค1 + 2 ร— ๐‘Ÿ๐‘œ๐‘ค2

So A only have two independent row vectors. Now let remove third row and first column of A then we have a 2 ร— 2 matrix which determinant is not zero.

1 22 1

โ‰  0

So rank of A is 2.