Numerically stable methods for quadratic programming

Numerically stable algorithms for quadratic programming are discussed. A new algorithm is described for indefinite quadratic programming which utilizes methods for updating positivedefinite factorizations only. Consequently all the updating procedures required are common to algorithms for linearly-constrained optimization. The new algorithm can be used for the positive-definite case without loss of efficiency.     

The computation of Lagrange-multiplier estimates for constrained minimization

Almost all efficient algorithms for constrained optimization require the repeated computation of Lagrange-multiplier estimates. In this paper we consider the difficulties in providing accurate estimates and what tests can be made in order to check the validity of the estimates obtained. A variety of formulae for the estimation of Lagrange multipliers are derived and their respective merits discussed. Finally the role of Lagrange multipliers within optimization algorithms is discussed and in addition to other results, it is shown that some algorithms are particularly sensitive to errors in the estimates.     

Linearly constrained minimax optimization

We present an algorithm for nonlinear minimax optimization subject to linear equality and inequality constraints which requires first order partial derivatives. The algorithm is based on successive linear approximations to the functions defining the problem. The resulting linear subproblems are solved in the minimax sense subject to the linear constraints. This ensures a feasible-point algorithm. Further, we introduce local bounds on the solutions of the linear subproblems, the bounds being adjusted automatically, depending on the quality of the linear approximations. It is proved that the algorithm will always converge to the set of stationary points of the problem, a stationary point being defined in terms of the generalized gradients of the minimax objective function. It is further proved that, under mild regularity conditions, the algorithm is identical to a quadratically convergent Newton iteration in its final stages. We demonstrate the performance of the algorithm by solving a number of numerical examples with up to 50 variables, 163 functions, and 25 constraints. We have also implemented a version of the algorithm which is particularly suited for the solution of restricted approximation problems.This work has been supported by the Danish Natural Science Research Council, Grant No. 511-6874.     

The inverse problem of stationary covariance generation

The paper considers the problem of passing from a stationary covariance, or spectral matrix, associated with the output of a constant linear finite-dimensional system excited by white noise to the set of all possible systems of minimum possible dimension which will generate this covariance. The problem, originally posed by R. E. Kalman in 1965, is solved by identifying each possible system with the solution of a quadratic matrix inequality; an algorithm for the solution of the inequality is also presented.Work supported by the Australian Research Grants Committee. University of New castle Technical Report EE-6816.     

A monotone streamline upwind method for quadratic finite elements

A direct streamline upwind method has been developed for convection-dominated flow problems utilizing quadratic elements. The approach presented retains the curve-sidedness feature offered by these elements. This facilitates the use of boundary conforming elements in domains that possess extreme curvature such as turbomachinery bladed components, for which the method is particularly suited. Three test cases are solved to evaluate the stability and diffusive characteristics of the numerical solution. The results presented clearly demonstrate that the proposed method does not exhibit any non-physical spatial oscillations, nor does it suffer from the traditional problem of excessive numerical diffusion.     

On the mean value of

Let χ be the Dirichlet character modulo q3 and L(s,χ) denote the corresponding Dirichlet L-function. The mean value of is studied and a few asymptotic formulae are given. Hybrid mean value of , general Kloosterman sums and general quadratic Gauss sums are considered.     

Stability of a quadratic Jensen type functional equation in the spaces of generalized functions

Making use of the fundamental solution of the heat equation we find the solution and prove the stability theorem of the quadratic Jensen type functional equation

     

Calculus students’ understanding of the vertex of the quadratic function in relation to the concept of derivative

The purpose of this study was to gain insight into 30, first year calculus students’ understanding of the relationship between the concept of vertex of a quadratic function and the concept of the derivative. APOS (action-process-object-schema) theory was applied as a guiding framework of analysis on student written work, think-aloud and follow up group interviews. Students’ personal meanings of the vertex, including misconceptions, were explored, along with students’ understanding to solve problems pertaining to the derivative of a quadratic function. Results give evidence of students’ weak schema of the vertex, lack of connection between different problem types and the importance of linguistics in relation to levels of APOS theory. A preliminary genetic decomposition was developed based on the results. Future research is suggested as a continuation to improve student understanding of the relationship between the vertex of quadratic functions and the derivative.