Least-squares approximations in geometric buildup for solving distance geometry problems (Q637550): Difference between revisions
From MaRDI portal
Set profile property. |
Set OpenAlex properties. |
||
Property / full work available at URL | |||
Property / full work available at URL: https://doi.org/10.1007/s10957-011-9806-6 / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W1965397948 / rank | |||
Normal rank |
Revision as of 19:48, 19 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Least-squares approximations in geometric buildup for solving distance geometry problems |
scientific article |
Statements
Least-squares approximations in geometric buildup for solving distance geometry problems (English)
0 references
6 September 2011
0 references
This article further investigates some theoretical and computational issues of the geometric buildup algorithm proposed by \textit{A. Sit, Z. Wu} and \textit{X. Yuan} [Bull. Math. Biol. 71, No. 8, 1914--1933 (2009; Zbl 1180.92029)] for the solution of the distance geometry problem with sparse and inexact distances. This problem is derived from the protein modeling problem, where \(n\) atoms with their distances \(d_{ij}\) under the Euclidean norm are given and we seek their coordinates. The algorithm repeatedly uses a least-squares approximation to determine the position of an undetermined atom, using the distances from this atom to a set of previously determined ones. The least-squares approximation, obtained from the singular value decomposition of a distance-induced matrix, can find the best possible position for the atom, even if the distances have small errors, as they usually do in practice, and therefore make the geometric buildup algorithm more stable than its previous versions that relied on linear system solvers. In this paper, the authors estimate its numerical errors and prove some of its key mathematical properties. They also present some numerical results with varying some of the parameters in the algorithm and show how they may be used to improve its performance and computational accuracy.
0 references
biomolecular modeling
0 references
distance geometry
0 references
nonlinear least squares
0 references
singular value decomposition
0 references