SIAM Journal on Control · 1971 · 439 citations · 9 references
Natural SciencesMathematicsAlgorithmsComputational MechanicsOptimizationFinite Difference MethodsMatrix ComputationNumerical AnalysisNumerical Linear AlgebraQuadratic ProgrammingFinite Difference AnalysisFree-boundary ProblemsJournal BearingsQuadratic Programming Relaxations
The problem involves a real symmetric positive definite matrix A and vector b, arising in free‑boundary journal bearing simulations where A is a finite‑difference matrix. The goal is to find nonnegative vectors x and y satisfying A x = b + y, xᵀ y = 0, and x, y ≥ 0. We modify systematic over‑relaxation to solve this constrained quadratic system, especially effective when A is a finite‑difference matrix.
Let ${\bf A}$ be a real symmetric positive definite $n \times n$ matrix and ${\bf b}$ a real column n-vector. We consider the following problem: Find real column n-vectors ${\bf x}$ and ${\bf y}$ such that \[{\bf A}{\bf x} = {\bf b} + {\bf y},\]\[\{\bf x}^T {\bf y} = 0,\quad {\bf x} \geqq 0,\quad {\bf y} \geqq 0.\] ] Problems of this type occur when the method of Christopherson is used to solve free boundary problems for journal bearings. In such cases, ${\bf A}$ is a “finite difference" matrix. We present a method for solving the above problem which is a modification of systematic over-relaxation. This method is particularly suitable when ${\bf A}$ is a finite difference matrix.
9
James H. Bramble, Richard S. Varga · Mathematics of Computation · 1963 · 4.2K citations
Nonlinear and Dynamic Programming
A. Wouk, G. Hadley · Technometrics · 1965 · 331 citations