Mat-1.3656 Seminar on
numerical analysis and computational science
Monday, Sept 16, 2013, room M233 at 14.15, Eirola & Stenberg
Luca Dieci, School of Mathematics, Georgia Institute of Technology, Atlanta
Numerical methods to locate coalescing eigenvalues of matrices depending on several parameters
Abstract:
There are well established techniques in numerical linear algebra to
compute eigenvalues
of Hermitian matrices. However, when a matrix depends on parameters, it
is important to
locate parameter values where the eigenvalues coalesce (come together),
and this is a
rather delicate task. In this talk, we discuss when and how one can
numerically detect
and approximate the points in parameters' space where such coalescings
occur. Theoretical
and algorithmic aspects and examples will be presented.
[Joint work with A.Papini, A.Pugliese.]