Académique Documents
Professionnel Documents
Culture Documents
Department of Mathematics
Indian Institute of Technology Guwahati
1 / 21
Eigenvalues and eigenvectors
Topics:
Determinant of Matrices
Eigenvalues and Eigenvectors
2 / 21
Determinant of a matrix
3 / 21
Determinant of a matrix
The determinant det(A) of 2 × 2 matrix A is defined by
a11 a12
det (A) = det = a11 a22 − a12 a21 .
a21 a22
3 / 21
Determinant of a matrix
The determinant det(A) of 2 × 2 matrix A is defined by
a11 a12
det (A) = det = a11 a22 − a12 a21 .
a21 a22
3 / 21
Determinant of a matrix
The determinant det(A) of 2 × 2 matrix A is defined by
a11 a12
det (A) = det = a11 a22 − a12 a21 .
a21 a22
3 / 21
Determinant of a matrix
The determinant det(A) of 2 × 2 matrix A is defined by
a11 a12
det (A) = det = a11 a22 − a12 a21 .
a21 a22
3 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ].
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A.
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
Example
1 5 0 0
2 0 8 0
det =
3 6 9 0
4 7 10 1
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
Example
1 5 0 0
0 8 0 2 8 0
" # " #
2 0 8 0
det = 1· det 6 9 0 −5· det 3 9 0 =
3 6 9 0
7 10 1 4 10 1
4 7 10 1
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
Example
1 5 0 0
0 8 0 2 8 0
" # " #
2 0 8 0
det = 1· det 6 9 0 −5· det 3 9 0 = −18.
3 6 9 0
7 10 1 4 10 1
4 7 10 1
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
Example
1 5 0 0
0 8 0 2 8 0
" # " #
2 0 8 0
det = 1· det 6 9 0 −5· det 3 9 0 = −18.
3 6 9 0
7 10 1 4 10 1
4 7 10 1
Theorem: The determinant of a diagonal, upper or lower triangular
matrix is the product of its diagonal entries.
4 / 21
Determinant of an n × n matrix
Definition: Let A = [aij ] ∈ Mn (F). The determinant of A is
defined by
det(A) = a11 , if n = 1, i.e., A = [a11 ]. For n ≥ 2:
det(A) = a11 det(A11 ) − a12 det(A12 ) + . . . + (−1)1+n a1n det(A1n ),
where Aij is the submatrix of A obtained by deleting the i-th row
and the j-th column of A. Note that det (A) ∈ F.
Example
1 5 0 0
0 8 0 2 8 0
" # " #
2 0 8 0
det = 1· det 6 9 0 −5· det 3 9 0 = −18.
3 6 9 0
7 10 1 4 10 1
4 7 10 1
Theorem: The determinant of a diagonal, upper or lower triangular
matrix is the product of its diagonal entries.
Proof: Use induction on the size of the matrix.
4 / 21
Properties of Determinant
Let A be a square matrix.
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
3 If A has two identical rows, then det(A) = 0.
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
3 If A has two identical rows, then det(A) = 0.
4 If B is obtained by multiplying a row of A by a scalar α, then
det(B) = α det(A).
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
3 If A has two identical rows, then det(A) = 0.
4 If B is obtained by multiplying a row of A by a scalar α, then
det(B) = α det(A).
5 If the matrices A, B and C are identical except that one of the
rows of C is the sum of the corresponding rows of A and B,
then det(C ) = det(A) + det(B).
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
3 If A has two identical rows, then det(A) = 0.
4 If B is obtained by multiplying a row of A by a scalar α, then
det(B) = α det(A).
5 If the matrices A, B and C are identical except that one of the
rows of C is the sum of the corresponding rows of A and B,
then det(C ) = det(A) + det(B).
6 If B is obtained by adding a multiple of one row of A to
another row, then det(B) = det(A).
5 / 21
Properties of Determinant
Let A be a square matrix.
1 If B is obtained by interchanging two rows of A, then
det(B) = −det(A).
2 If A has a zero row, then det(A) = 0.
3 If A has two identical rows, then det(A) = 0.
4 If B is obtained by multiplying a row of A by a scalar α, then
det(B) = α det(A).
5 If the matrices A, B and C are identical except that one of the
rows of C is the sum of the corresponding rows of A and B,
then det(C ) = det(A) + det(B).
6 If B is obtained by adding a multiple of one row of A to
another row, then det(B) = det(A).
7 det (αA) = αn det A.
5 / 21
Expansion of det along rows and columns
6 / 21
Expansion of det along rows and columns
6 / 21
Expansion of det along rows and columns
6 / 21
Expansion of det along rows and columns
6 / 21
Expansion of det along rows and columns
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij =
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij = − 1, det Ei (α) =
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij = − 1, det Ei (α) = α, det Eij (α) =
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij = − 1, det Ei (α) = α, det Eij (α) = 1.
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij = − 1, det Ei (α) = α, det Eij (α) = 1.
For any elementary matrix E , det (E ) = det (E T ).
For any elementary matrix E , det (EA) = det (E ) det (A).
6 / 21
Expansion of det along rows and columns
Theorem:
Determinants of elementary matrices:
det Eij = − 1, det Ei (α) = α, det Eij (α) = 1.
For any elementary matrix E , det (E ) = det (E T ).
For any elementary matrix E , det (EA) = det (E ) det (A).
6 / 21
Two important properties
Suppose A ∈ Mn (F).
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei .
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
Definition A matrix A ∈ Mn (F) is said to be singular or
non-singular according as
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
Definition A matrix A ∈ Mn (F) is said to be singular or
non-singular according as det(A) = 0 or det(A) 6= 0.
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
Definition A matrix A ∈ Mn (F) is said to be singular or
non-singular according as det(A) = 0 or det(A) 6= 0.
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
Definition A matrix A ∈ Mn (F) is said to be singular or
non-singular according as det(A) = 0 or det(A) 6= 0.
Theorem:
(a) For any matrix A ∈ Mn (F), det (A) = det (AT ).
(b) For any A, B ∈ Mn (F), det (AB) = det (A) det (B).
7 / 21
Two important properties
Suppose A ∈ Mn (F). Then A = Ek · · · E1 · rref(A) for some
elementary Ei . We have
det (A) = det (Ek ) · · · det (E1 ) det (rref(A)).
Thus,
det (A) 6= 0 iff det (rref(A)) 6= 0 iff rref(A) = In iff A is invertible.
Definition A matrix A ∈ Mn (F) is said to be singular or
non-singular according as det(A) = 0 or det(A) 6= 0.
Theorem:
(a) For any matrix A ∈ Mn (F), det (A) = det (AT ).
(b) For any A, B ∈ Mn (F), det (AB) = det (A) det (B).
7 / 21
Eigenvalues and Eigenvectors
8 / 21
Eigenvalues and Eigenvectors
8 / 21
Eigenvalues and Eigenvectors
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B =
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B = [T (v1 )]B , . . . , [T (vn )]B =
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B = [T (v1 )]B , . . . , [T (vn )]B = diag{λ1 , . . . , λn }.
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B = [T (v1 )]B , . . . , [T (vn )]B = diag{λ1 , . . . , λn }.
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B = [T (v1 )]B , . . . , [T (vn )]B = diag{λ1 , . . . , λn }.
8 / 21
Eigenvalues and Eigenvectors
T (α1 v1 + · · · + αn vn ) = λ1 α1 v1 + · · · + λn αn vn .
Remark
[T vi ]B = λi [vi ]B .
[T ]B = [T (v1 )]B , . . . , [T (vn )]B = diag{λ1 , . . . , λn }.
8 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F).
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn ,
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
Thus it is enough to study eigenvales/eigenvectors of square
matrices.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
Thus it is enough to study eigenvales/eigenvectors of square
matrices.
" #
1 3
Example: Consider the matrix A = .
3 1
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
Thus it is enough to study eigenvales/eigenvectors of square
matrices.
" #
1 3
Example: Consider the matrix A = . Then 4 is an
3 1
eigenvalue of A with [1, 1]> as a corresponding eigenvector.
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
Thus it is enough to study eigenvales/eigenvectors of square
matrices.
" #
1 3
Example: Consider the matrix A = . Then 4 is an
3 1
eigenvalue of A with [1, 1]> as a corresponding eigenvector.
" #
cos θ − sin θ
Next, consider the rotation matrix A = . Then
sin θ cos θ
Av 6= λv for any λ ∈ R and 0 6= v ∈ R2 .
9 / 21
Eigenvalues and eigenvectors
Definition: Let A ∈ Mn (F). If Ax = λx for some λ ∈ F and
0 6= x ∈ Fn , then λ is called an eigenvalue of A and x an
eigenvector of A corresponding to λ.
Fact: Suppose dim(V) = n, and T : V → V is an LT. Then
T (v) = λv iff [T (v)]B = λ[v]B for any ordered basis B of V.
Thus it is enough to study eigenvales/eigenvectors of square
matrices.
" #
1 3
Example: Consider the matrix A = . Then 4 is an
3 1
eigenvalue of A with [1, 1]> as a corresponding eigenvector.
" #
cos θ − sin θ
Next, consider the rotation matrix A = . Then
sin θ cos θ
Av 6= λv for any λ ∈ R and 0 6= v ∈ R2 .
Henceforth, we consider F = C. 9 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C).
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
10 / 21
Eigenspace
Definition: Let λ be an eigenvalue of a matrix A ∈ Mn (C). Then
Eλ = {x | Ax = λx}
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces?
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
det (A − λI ) =
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0
3 1−λ
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
E4 = null(A − 4I ) =
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
−3 3
E4 = null(A − 4I ) = x: x=0 =
3 −3
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
−3 3
E4 = null(A − 4I ) = x: x=0 = span([1, 1]> ),
3 −3
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
−3 3
E4 = null(A − 4I ) = x: x=0 = span([1, 1]> ),
3 −3
E(−2) = null(A + 2I ) =
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
−3 3
E4 = null(A − 4I ) = x: x=0 = span([1, 1]> ),
3 −3
3 3
E(−2) = null(A + 2I ) = x: x=0 =
3 3
11 / 21
Facts:
The eigenvalues of A are the zeros of the characteristic
polynomial det(A − λI ) of A. Hence A has at most n
eigenvalues.
The eigenvalues of a triangular matrix are its diagonal entries.
A is invertible iff 0 is not an eigenvalue of A.
1 3
Example: What are the eigenvalues of A = and their
3 1
eigenspaces? We have
1−λ 3
det (A − λI ) = det =0 ⇔ λ = 4, −2,
3 1−λ
−3 3
E4 = null(A − 4I ) = x: x=0 = span([1, 1]> ),
3 −3
3 3
E(−2) = null(A + 2I ) = x: x=0 = span([1, −1]> ),
3 3
11 / 21
Computation of eigenvalues and bases for eigenspaces
12 / 21
Computation of eigenvalues and bases for eigenspaces
12 / 21
Computation of eigenvalues and bases for eigenspaces
12 / 21
Computation of eigenvalues and bases for eigenspaces
Example
A real matrix may have complex eigenvalues and complex
eigenvectors.
12 / 21
Computation of eigenvalues and bases for eigenspaces
Example
A real matrix may have complex
eigenvalues
and complex
0 1
eigenvectors. Consider A = .
−1 0
12 / 21
Computation of eigenvalues and bases for eigenspaces
Example
A real matrix may have complex
eigenvalues
and complex
0 1
eigenvectors. Consider A = . The char. poly. of A is
−1 0
λ2 + 1 and so the eigenvalues are ± ι̇.
12 / 21
Computation of eigenvalues and bases for eigenspaces
Example
A real matrix may have complex
eigenvalues
and complex
0 1
eigenvectors. Consider A = . The char. poly. of A is
−1 0
λ2 + 1 and so the eigenvalues are ± ι̇. Corresponding to the
eigenvalues ± ι̇, it has eigenvectors [1, ι̇]> and [ι̇, 1]> , respectively.
12 / 21
Computation of eigenvalues and bases for eigenspaces
Example
A real matrix may have complex
eigenvalues
and complex
0 1
eigenvectors. Consider A = . The char. poly. of A is
−1 0
λ2 + 1 and so the eigenvalues are ± ι̇. Corresponding to the
eigenvalues ± ι̇, it has eigenvectors [1, ι̇]> and [ι̇, 1]> , respectively.
The eigenspaces Eι̇ and E(−ι̇) are one dimensional.
12 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
13 / 21
Algebraic and geometric multiplicities
Definition: Let λ be an eigenvalue of A ∈ Mn (C).
The algebraic multiplicity of λ is the multiplicity of λ as a
root of the characteristic polynomial of A.
The geometric multiplicity of λ is the dimension of Eλ .
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A,
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi .
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI.
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not.
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 .
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
= (λk − λ1 )α1 x1 + · · · + (λk − λk−1 )αk−1 xk−1 .
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
= (λk − λ1 )α1 x1 + · · · + (λk − λk−1 )αk−1 xk−1 .
Since {x1 , . . . , xk−1 } is LI,
(λk − λ1 )α1 = · · · = (λk − λk−1 )αk−1 = 0,
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
= (λk − λ1 )α1 x1 + · · · + (λk − λk−1 )αk−1 xk−1 .
Since {x1 , . . . , xk−1 } is LI,
(λk − λ1 )α1 = · · · = (λk − λk−1 )αk−1 = 0,
i.e., α1 = · · · = αk−1 = 0
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
= (λk − λ1 )α1 x1 + · · · + (λk − λk−1 )αk−1 xk−1 .
Since {x1 , . . . , xk−1 } is LI,
(λk − λ1 )α1 = · · · = (λk − λk−1 )αk−1 = 0,
i.e., α1 = · · · = αk−1 = 0 (as λi are distinct),
14 / 21
Linearly independent eigenvectors
Theorem: Eigenvectors corresponding to distinct eigenvalues are
linearly independent.
Proof. Suppose that λ1 , . . . , λm are distinct eigenvalues of A, xi
is an eigenvector of A corresponding to λi . Suppose, if possible, xi
are not LI. Then there is k > 1 such that {x1 , . . . , xk−1 } is LI, but
{x1 , . . . , xk } is not. Let xk = α1 x1 + · · · + αk−1 xk−1 . Then,
Axk = λ1 α1 x1 + · · · + λk−1 αk−1 xk−1 .
Therefore, 0 = λk xk − Axk
= (λk − λ1 )α1 x1 + · · · + (λk − λk−1 )αk−1 xk−1 .
Since {x1 , . . . , xk−1 } is LI,
(λk − λ1 )α1 = · · · = (λk − λk−1 )αk−1 = 0,
i.e., α1 = · · · = αk−1 = 0 (as λi are distinct), i.e. xk = 0, a
contradiction.
14 / 21
Cayley-Hamilton Theorem
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Cayley-Hamilton Theorem
Let p(λ) be the characteristic polynomial of a matrix A. Then
p(A) = 0, the zero matrix.
15 / 21
Exercise
Let A be a matrix with eigenvalue λ and corresponding
eigenvector x.
For any positive integer n, show that λn is an eigenvalue of An
with corresponding eigenvector x.
If A is invertible, then show that λ1 is an eigenvalue of A−1
with corresponding eigenvector x.
If A is invertible then show that for any integer n, λ−n is an
eigenvalue of A−n with corresponding eigenvector x.
Let v1 , v2 , . . . , vm be eigenvectors of a matrix A with
corresponding eigenvalues λ1 , λ2 , . . . , λm , respectively. Let
x = c1 v1 + c2 v2 + . . . + cm vm . Show that for any positive
integer k,
16 / 21
Google Search Engine
The task of extracting information from all the web pages available
on the Internet is performed by search engines.
17 / 21
Google Search Engine
The task of extracting information from all the web pages available
on the Internet is performed by search engines.
17 / 21
Google Search Engine
The task of extracting information from all the web pages available
on the Internet is performed by search engines.
17 / 21
Google Search Engine
The task of extracting information from all the web pages available
on the Internet is performed by search engines.
17 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
18 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
The web is an example of a directed graph. Let all the web pages
be ordered as P1 , . . . , Pn . Link from Pi to Pj represents an arrow.
Google assigns rank to a page based on its in-links and out-links.
18 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
The web is an example of a directed graph. Let all the web pages
be ordered as P1 , . . . , Pn . Link from Pi to Pj represents an arrow.
Google assigns rank to a page based on its in-links and out-links.
18 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
The web is an example of a directed graph. Let all the web pages
be ordered as P1 , . . . , Pn . Link from Pi to Pj represents an arrow.
Google assigns rank to a page based on its in-links and out-links.
18 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
The web is an example of a directed graph. Let all the web pages
be ordered as P1 , . . . , Pn . Link from Pi to Pj represents an arrow.
Google assigns rank to a page based on its in-links and out-links.
18 / 21
PageRank Algorithm
Google PageRank algorithm assigns ranks to all the web pages and
is formulated as a matrix eigenvalue problem.
The web is an example of a directed graph. Let all the web pages
be ordered as P1 , . . . , Pn . Link from Pi to Pj represents an arrow.
Google assigns rank to a page based on its in-links and out-links.
18 / 21
PageRank Algorithm (eigenvalue problem)
Let xj ≥ 0 be the rank of page Pj . Then xj > xk =⇒ Pj is more
important than page Pk .
19 / 21
PageRank Algorithm (eigenvalue problem)
Let xj ≥ 0 be the rank of page Pj . Then xj > xk =⇒ Pj is more
important than page Pk .
If page Pj contains nj out-links, one of which links to page Pk ,
then page Pk ’s score is boosted by xj /nj .
19 / 21
PageRank Algorithm (eigenvalue problem)
Let xj ≥ 0 be the rank of page Pj . Then xj > xk =⇒ Pj is more
important than page Pk .
If page Pj contains nj out-links, one of which links to page Pk ,
then page Pk ’s score is boosted by xj /nj .
19 / 21
PageRank Algorithm (eigenvalue problem)
Let xj ≥ 0 be the rank of page Pj . Then xj > xk =⇒ Pj is more
important than page Pk .
If page Pj contains nj out-links, one of which links to page Pk ,
then page Pk ’s score is boosted by xj /nj .
Eigenvalue problem Hv = v ,
19 / 21
The Google matrix
The hyperlink matrix is modified to obtain the Google matrix
G := (1 − m)H + mS,
20 / 21
The Google matrix
The hyperlink matrix is modified to obtain the Google matrix
G := (1 − m)H + mS,
20 / 21
The Google matrix
The hyperlink matrix is modified to obtain the Google matrix
G := (1 − m)H + mS,
20 / 21
The Google matrix
The hyperlink matrix is modified to obtain the Google matrix
G := (1 − m)H + mS,
20 / 21
The Google matrix
The hyperlink matrix is modified to obtain the Google matrix
G := (1 − m)H + mS,
20 / 21
References
21 / 21
References
21 / 21
References
***
21 / 21