arXiv subjects

Zhonggang Zeng

Publications and source records attributed to Zhonggang Zeng.

10 recordsLinked to original sources

arXiv2024

A Newton's Iteration Converges Quadratically to Nonisolated Solutions Too

The textbook Newton's iteration is practically inapplicable on solutions of nonlinear systems with singular Jacobians. By a simple modification, a novel extension of Newton's iteration regains its local quadratic convergence toward nonisolated solutions that are semiregular as properly defined regardless of whether the system is square, underdetermined or overdetermined while Jacobians can be rank-deficient. Furthermore, the iteration serves as a regularization mechanism for computing singular solutions from empirical data. When a system is perturbed, its nonisolated solutions can be altered substantially or even disappear. The iteration still locally converges to a stationary point that approximates a singular solution of the underlying system with an error bound in the same order of the data accuracy. Geometrically, the iteration approximately approaches the nearest point on the solution manifold. The method simplifies the modeling of nonlinear systems by permitting nonisolated solutions and enables a wide range of applications in algebraic computation.

math.NA↗

arXiv2023

Computing multiple roots of inexact polynomials

We present a combination of two algorithms that accurately calculate multiple roots of general polynomials. Algorithm I transforms the singular root-finding into a regular nonlinear least squares problem on a pejorative manifold, and calculates multiple roots simultaneously from a given multiplicity structure and initial root approximations. To fulfill the input requirement of Algorithm I, we develop a numerical GCD-finder containing a successive singular value updating and an iterative GCD refinement as the main engine of Algorithm II that calculates the multiplicity structure and the initial root approximation. While limitations of our algorithm exist in identifying the multiplicity structure in certain situations, the combined method calculates multiple roots with high accuracy and consistency in practice without using multiprecision arithmetic even if the coefficients are inexact. This is perhaps the first blackbox-type root-finder with such capabilities. To measure the sensitivity of the multiple roots, a structure-preserving condition number is proposed and error bounds are established. According to our computational experiments and error analysis, a polynomial being ill-conditioned in the conventional sense can be well conditioned with the multiplicity structure being preserved, and its multiple roots can be computed with high accuracy.

math.NA↗

arXiv2021

Multiple zeros of nonlinear systems

As an attempt to bridge between numerical analysis and algebraic geometry, this paper formulates the multiplicity for the general nonlinear system at an isolated zero, presents an algorithm for computing the multiplicity structure, proposes a depth-deflation method for accurate computation of multiple zeros, and introduces the basic algebraic theory of the multiplicity. Furthermore, this paper elaborates and proves some fundamental properties of the multiplicity, including local finiteness, consistency, perturbation invariance, and depth-deflatability. As a justification of this formulation, the multiplicity is proved to be consistent with the multiplicity defined in algebraic geometry for the special case of polynomial systems. The proposed algorithms can accurately compute the multiplicity and the multiple zeros using floating point arithmetic even if the nonlinear system is perturbed.

math.NA↗

arXiv2021

The numerical factorization of polynomials

Polynomial factorization in conventional sense is an ill-posed problem due to its discontinuity with respect to coefficient perturbations, making it a challenge for numerical computation using empirical data. As a regularization, this paper formulates the notion of numerical factorization based on the geometry of polynomial spaces and the stratification of factorization manifolds. Furthermore, this paper establishes the existence, uniqueness, Lipschitz continuity, condition number, and convergence of the numerical factorization to the underlying exact factorization, leading to a robust and efficient algorithm with a MATLAB implementation capable of accurate polynomial factorizations using floating point arithmetic even if the coefficients are perturbed.

math.NA↗

arXiv2021

The numerical greatest common divisor of univariate polynomials

This paper presents a regularization theory for numerical computation of polynomial greatest common divisors and a convergence analysis, along with a detailed description of a blackbox-type algorithm. The root of the ill-posedness in conventional GCD computation is identified by its geometry where polynomials form differentiable manifolds entangled in a stratification structure. With a proper regularization, the numerical GCD is proved to be strongly well-posed. Most importantly, the numerical GCD solves the problem of finding the GCD accurately using floating point arithmetic even if the data are perturbed. A sensitivity measurement, error bounds at each computing stage, and the overall convergence are established rigorously. The computing results of selected test examples show that the algorithm and software appear to be robust and accurate.

math.NA↗

arXiv2021

Sensitivity and computation of a defective eigenvalue

A defective eigenvalue is well documented to be hypersensitive to data perturbations and round-off? errors, making it a formidable challenge in numerical computation particularly when the matrix is known through approximate data. This paper establishes a finitely bounded sensitivity of a defective eigenvalue with respect to perturbations that preserve the geometric multiplicity and the smallest Jordan block size. Based on this perturbation theory, numerical computation of a defective eigenvalue is regularized as a well-posed least squares problem so that it can be accurately carried out using floating point arithmetic even if the matrix is perturbed.

math.NA↗

arXiv2021

A numerical method for computing the Jordan Canonical Form

The Jordan Canonical Form of a matrix is highly sensitive to perturbations, and its numerical computation remains a formidable challenge. This paper presents a regularization theory that establishes a well-posed least squares problem of finding the nearest staircase decomposition in the matrix bundle of the highest codimension. A two-staged algorithm is developed for computing the numerical Jordan Canonical Form. At the first stage, the method calculates the Jordan structure of the matrix and an initial approximation to the multiple eigenvalues. The staircase decomposition is then constructed by an iterative algorithm at the second stage. As a result, the numerical Jordan Canonical decomposition along with multiple eigenvalues can be computed with high accuracy even if the underlying matrix is perturbed.

math.NA↗

arXiv2021

On the sensitivity of singular and ill-Conditioned linear systems

Solving a singular linear system for an individual vector solution is an ill-posed problem with a condition number infinity. From an alternative perspective, however, the general solution of a singular system is of a bounded sensitivity as a unique element in an affine Grassmannian. If a singular linear system is given through empirical data that are sufficiently accurate with a tight error bound, a properly formulated general numerical solution uniquely exists in the same affine Grassmannian, enjoys Lipschitz continuity and approximates the underlying exact solution with an accuracy in the same order as the data. Furthermore, any backward accurate numerical solution vector is an accurate approximation to one of the solutions of the underlying singular system.

math.NA↗

arXiv2021

Singular algebraic equations with empirical data

Singular equations with rank-deficient Jacobians arise frequently in algebraic computing applications. As shown in case studies in this paper, direct and intuitive modeling of algebraic problems often results in nonisolated singular solutions. The challenges become formidable when the problems need to be solved from empirical data of limited accuracy. A newly discovered low-rank Newton's iteration emerges as an effective regularization mechanism that enables solving singular equations accurately with an error bound in the same order as the data error. This paper elaborates applications of new methods on solving singular algebraic equations such as singular linear systems, polynomial GCD and factorizations as well as matrix defective eigenvalue problems.

math.NA↗

arXiv2021

Geometric modeling and regularization of algebraic problems

Discontinuity with respect to data perturbations is common in algebraic computation where solutions are often highly sensitive. Such problems can be modeled as solving systems of equations at given data parameters. By appending auxiliary equations, the models can be formulated to satisfy four easily verifiable conditions so that the data form complex analytic manifolds on which the solutions maintain their structures and the Lipschitz continuity. When such a problem is given with empirical data, solving the system becomes a least squares problem whose solution uniquely exists and enjoys Lipschitz continuity as long as the data point is in a tubular neighborhood of the manifold. As a result, the singular problem is regularized as a well-posed computational problem.

math.NA↗