arXiv ScienceSearch

arXiv · 1108.6030

Dynamics of the symmetric eigenvalue problem with shift strategies

Abstract

A common algorithm for the computation of eigenvalues of real symmetric tridiagonal matrices is the iteration of certain special maps $F_σ$ called shifted $QR$ steps. Such maps preserve spectrum and a natural common domain is ${\cal T}_Λ$, the manifold of real symmetric tridiagonal matrices conjugate to the diagonal matrix $Λ$. More precisely, a (generic) shift $s \in \RR$ defines a map $F_s: {\cal T}_Λ\to {\cal T}_Λ$. A strategy $σ: {\cal T}_Λ\to \RR$ specifies the shift to be applied at $T$ so that $F_σ(T) = F_{σ(T)}(T)$. Good shift strategies should lead to fast deflation: some off-diagonal coordinate tends to zero, allowing for reducing of the problem to submatrices. For topological reasons, continuous shift strategies do not obtain fast deflation; many standard strategies are indeed discontinuous. Practical implementation only gives rise systematically to bottom deflation, convergence to zero of the lowest off-diagonal entry $b(T)$. For most shift strategies, convergence to zero of $b(T)$ is cubic, $|b(F_σ(T))| = Θ(|b(T)|^k)$ for $k = 3$. The existence of arithmetic progressions in the spectrum of $T$ sometimes implies instead quadratic convergence, $k = 2$. The complete integrability of the Toda lattice and the dynamics at non-smooth points are central to our discussion. The text does not assume knowledge of numerical linear algebra.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ricardo S. Leite, Nicolau C. Saldanha, Carlos Tomei. 2011-08-30. Dynamics of the symmetric eigenvalue problem with shift strategies. https://doi.org/10.1093/imrn%2Frns186

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Monotonicity, global symplectification and the stability of Dry Ten Martini Problem

We prove that, for every irrational frequency and every analytic Type I potential, each supercritical spectral energy satisfying the gap-labelling condition is an endpoint of an open spectral gap. This establishes the conjecture of Ge--Jitomirskaya--You \cite{GJY,You} in the supercritical regime. Consequently, the ``all gaps open'' property of the supercritical almost Mathieu operator persists under sufficiently small analytic perturbations. The main ingredient is a global symplectification of the center bundle that preserves quantitative monotonicity. This allows us to study gap opening through the center dynamics of the dual long-range operator, which has no natural Schrödinger form. We first establish the result for trigonometric polynomial potentials and then pass to general analytic potentials by controlling the dependence on the truncation dimension. The proof combines a discrete Hellmann--Feynman identity, dimension-free Aubry duality in weighted analytic norms, and a quantitative cone argument based on pre-monotonicity. These estimates ensure that the gaps survive in the analytic limit. Our results establish analytic stability of the Dry Ten Martini Problem in the supercritical regime and give a partial answer to a question of M. Shamis on the persistence of periodic spectral gaps.

math.DS

Asymmetry of a class of Mellin transforms via bounded solutions

We introduce a family of parametrized non-homogeneous linear complex differential equations on $[1,\infty)$, depending on a complex parameter $s$ in the critical strip. We identify sufficient conditions on the non-homogeneous term that induce a structural asymmetry between the solutions corresponding to the parameters $s$ and $1-s$. More precisely, if both solutions with initial value $1$ are bounded on $[1,\infty)$, then necessarily $\Re(s)=\tfrac12$. The initial condition associated with the unique bounded solution corresponding to a parameter $s$ represents a zero of the Mellin transform associated with the non-homogeneous term at the point $s$.

math.DS

Self-similar Delone sets and Pisot numbers

We consider Delone point patterns with self-similarity. Under mild conditions, the similarity factor is a Pisot number if and only if the pattern is uniformly discrete. The classical case is a Meyer set $Λ$ with $Λ\supset θΛ$ for some $θ>1,$ for which $θ$ must be a Pisot number or a Salem number. When $Λ$ contains several similar copies of itself, the case of a Salem number drops out for $θ<2.$ On the other hand, strictly self-similar patterns with a Pisot factor must be Meyer sets. Various examples are given.

math.DS