A Topological Approach to Left Eigenvalues of Quaternionic Matrices

A Topological Approach to Left Eigenvalues of Quaternionic Matrices

A Topological Approach to Left Eigenvalues of Quaternionic Matrices M.J. Pereira-S´aezy (joint work with E. Mac´ıas-Virg´os∗) yDepartamento de Econom´ıaAplicada II, Universidade da Coru~na ∗Departamento de Xeometr´ıae Topolox´ıa, Universidade de Santiago de Compostela Sevilla, September 9-14, 2012 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 2 / 69 Left eigenvalues 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 3 / 69 Left eigenvalues Notion 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 4 / 69 Left eigenvalues Notion Notion Definition Let A 2 Mn×n(H), a quaternion λ 2 H is said to be a left eigenvalue of the matrix A if Av = λv for some vector v 2 Hn; v 6= 0: Equivalently, A − λI is a singular matrix. For a matrix A 2 Mn×n(H), we denote σl (A) its left spectrum, i.e. the set of left eigenvalues. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 5 / 69 Huang and So [Linear Algebra Appl., 2001] proved that a 2 × 2 matrix may have one, two or infinite left eigenvalues. A different proof was presented by EMV and MJPS [Electron. J. Linear Algebra, 2009]. Both proofs are algebraic in nature and seemingly difficult to generalize for n > 2. Left eigenvalues Notion About the existence of left eigenvalues: Wood [Bull. Lond. Math. Soc., 1985] proved that any n × n quaternionic matrix A has at least one left eigenvalue. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 6 / 69 A different proof was presented by EMV and MJPS [Electron. J. Linear Algebra, 2009]. Both proofs are algebraic in nature and seemingly difficult to generalize for n > 2. Left eigenvalues Notion About the existence of left eigenvalues: Wood [Bull. Lond. Math. Soc., 1985] proved that any n × n quaternionic matrix A has at least one left eigenvalue. Huang and So [Linear Algebra Appl., 2001] proved that a 2 × 2 matrix may have one, two or infinite left eigenvalues. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 6 / 69 Both proofs are algebraic in nature and seemingly difficult to generalize for n > 2. Left eigenvalues Notion About the existence of left eigenvalues: Wood [Bull. Lond. Math. Soc., 1985] proved that any n × n quaternionic matrix A has at least one left eigenvalue. Huang and So [Linear Algebra Appl., 2001] proved that a 2 × 2 matrix may have one, two or infinite left eigenvalues. A different proof was presented by EMV and MJPS [Electron. J. Linear Algebra, 2009]. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 6 / 69 Left eigenvalues Notion About the existence of left eigenvalues: Wood [Bull. Lond. Math. Soc., 1985] proved that any n × n quaternionic matrix A has at least one left eigenvalue. Huang and So [Linear Algebra Appl., 2001] proved that a 2 × 2 matrix may have one, two or infinite left eigenvalues. A different proof was presented by EMV and MJPS [Electron. J. Linear Algebra, 2009]. Both proofs are algebraic in nature and seemingly difficult to generalize for n > 2. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 6 / 69 * λ 2 H is on σl (A) iff the matrix A − λI is not invertible. Left eigenvalues Notion { There is no systematic study of left eigenvalues for n > 2. { In this talk we try a topological approach that could be generalized to other dimensions. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 7 / 69 Left eigenvalues Notion { There is no systematic study of left eigenvalues for n > 2. { In this talk we try a topological approach that could be generalized to other dimensions. * λ 2 H is on σl (A) iff the matrix A − λI is not invertible. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 7 / 69 Left eigenvalues Study's determinant 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 8 / 69 Left eigenvalues Study's determinant Study's determinant It is possible to generalize to the quaternions the norm j det j (that is, with real values) of the complex determinant. Definition Let the quaternionic matrix A 2 Mn×n(H) be decomposed as A = X + jY with X ; Y 2 Mn×n(C). We shall call Study's determinant of A the non-negative real number Sdet(A) := (det c(A))1=2; X −Y where c(A) is the complex matrix c(A) = 2 M ( ): Y X 2n×2n C M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 9 / 69 Left eigenvalues Study's determinant Proposition Study's determinant Sdet is the only functional that verifies the following properties : 1 Sdet(AB) = Sdet(A) · Sdet(B); 2 if A is a complex matrix then Sdet(A) = j det(A)j. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 10 / 69 Left eigenvalues Study's determinant The following immediate consequences are very useful for computations. Corollary 1 Sdet(A) > 0 if and only if the matrix A is invertible; −1 2 let A and B = PAP be similar matrices, then Sdet(A) = Sdet(B); 3 Sdet(A) does not change when a (right) multiple of one column is added to another column. 4 Sdet(A) does not change when a (left) multiple of one row is added to another row; 5 Sdet(A) does not change when two columns (or two rows) are permuted. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 11 / 69 Left eigenvalues Study's determinant We shall need the following property too Proposition For any matrix formed by two boxes A; B of order m and n respectively, it holds that A 0 Sdet = Sdet(A) · Sdet(B): ∗ B M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 12 / 69 Left eigenvalues Characteristic map 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 13 / 69 Proposition The spectrum σl (A) is compact. Left eigenvalues Characteristic map Characteristic map Definition The map µ : H ! H is a characteristic map of the matrix A 2 Mn×n(H) if, up to a constant, its norm verifies jµ(λ)j = Sdet(A − λI) for all λ 2 H. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 14 / 69 Left eigenvalues Characteristic map Characteristic map Definition The map µ : H ! H is a characteristic map of the matrix A 2 Mn×n(H) if, up to a constant, its norm verifies jµ(λ)j = Sdet(A − λI) for all λ 2 H. Proposition The spectrum σl (A) is compact. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 14 / 69 Left eigenvalues Characteristic map For example, let D = diag(q1;:::; qn) be a diagonal matrix. Then µ(λ) = (q1 − λ) ··· (qn − λ) is a characteristic map for D. For a triangular matrix it is the same. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 15 / 69 Spectrum of 2 × 2 matrices 1 Left eigenvalues Notion Study's determinant Characteristic map 2 Spectrum of 2 × 2 matrices Topological degree Linearization Sylvester equation Classification of left spectra 3 3 × 3 matrices Polynomial case Rational case Topological study of the 3 × 3 case M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 16 / 69 Spectrum of 2 × 2 matrices Remark The left spectrum is not invariant by similarity. However, if P is an invertible real matrix then Sdet(A − λI) = Sdet(PAP−1 − λI), hence A and PAP−1 have the same characteristic maps. M.J. Pereira-S´aez(UDC) Left Eigenvalues of Quaternionic Matrices Sevilla, September 9-14, 2012 17 / 69 if bc = 0 (triangular matrix), σl (A) = fa; dg; µ(λ) = (d − λ)(a − λ): we study now the bc 6= 0 case.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    93 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us