Truncated Newton method for sparse unconstrained optimization using automatic differentiation

When solving large complex optimization problems, the user is faced with three major problems. These are (i) the cost in human time in obtaining accurate expressions for the derivatives involved; (ii) the need to store second derivative information; and (iii), of lessening importance, the time taken to solve the problem on the computer. For many problems, a significant part of the latter can be attributed to solving Newton-like equations. In the algorithm described, the equations are solved using a conjugate direction method that only needs the Hessian at the current point when it is multiplied by a trial vector. In this paper, we present a method that finds this product using automatic differentiation while only requiring vector storage. The method takes advantage of any sparsity in the Hessian matrix and computes exact derivatives. It avoids the complexity of symbolic differentiation, the inaccuracy of numerical differentiation, the labor of finding analytic derivatives, and the need for matrix store. When far from a minimum, an accurate solution to the Newton equations is not justified, so an approximate solution is obtained by using a version of Dembo and Steihaug's truncated Newton algorithm (Ref. 1).This paper was presented at the SIAM National Meeting, Boston, Massachusetts, 1986.     

The Third-Order Singular-Perturbation Expansion of the Planar Cold-Fluid Magnetron Equations

The singular-perturbation expansion of the plasma cold-fluid equations for crossed fields in a planar geometry is presented. The general expansion is carried out to third order. Various instabilities that occur in the first, second, and third orders are discussed.     

On the analysis of a dynamic evolutionary algorithm

Evolutionary algorithms are applied as problem-independent optimization algorithms. They are quite efficient in many situations. However, it is difficult to analyze even the behavior of simple variants of evolutionary algorithms like the (1+1) EA on rather simple functions. Nevertheless, only the analysis of the expected run time and the success probability within a given number of steps can guide the choice of the free parameters of the algorithms. Here static (1+1) EAs with a fixed mutation probability are compared with dynamic (1+1) EAs with a simple schedule for the variation of the mutation probability. The dynamic variant is first analyzed for functions typically chosen as example-functions for evolutionary algorithms. Afterwards, it is shown that it can be essential to choose the suitable variant of the (1+1) EA. More precisely, functions are presented where each static (1+1) EA has exponential expected run time while the dynamic variant has polynomial expected run time. For other functions it is shown that the dynamic (1+1) EA has exponential expected run time while a static (1+1) EA with a good choice of the mutation probability has polynomial run time with overwhelming probability.     

Simple, expedient methods for the determination of water and electrolyte contents of cellulose solvent systems

The concentrations of water, W, and electrolytes present in solutions of LiCl in N,N-dimethylacetamide, LiCl/DMAc, and of tetrabutylammonium fluoride. x-hydrate in DMSO, TBAF.xW/DMSO can be accurately and expediently determined by three independent methods, UV–vis, FTIR and EMF measurement. The first relies on the use of solvatochromic probes whose spectra are sensitive to solution composition. It is applicable to W/LiCl/DMAc solutions but not to TBAF.xW/DMSO, because the charge-transfer complex bands of the probes are suppressed by strong interactions with the latter electrolyte. Integration of ν_OH band of water may be employed in order to determine [W], hence [electrolyte] by weight difference. EMF measurement uses ion-selective electrodes in order to determine [electrolyte], hence [W] by weight difference. Results of the latter method were in excellent agreement with those of FTIR. The reason for the failure of Karl Fischer titration is addressed, and the relevance of the results obtained to functionalization of cellulose under homogenous solution conditions is briefly commented on.     

3D-TOFSIMS characterization of black spots in polymer light emitting diodes

The occurrence and formation of black spots areas in PolyLED devices has been studied by time-of-flight SIMS (TOFSIMS). The composition, shape and position of the black spots is visualised by three-dimensional (3D)-TOFSIMS depth-profiling. It has been established that the formation of non-emissive spots is due to the growth of aluminium oxide clusters at the AlBa/polymer interface. Electron injection in the black spots is lost by the resulting local increase of the resistivity of the cathode.     

A model for adatom structures

A model concerning adatom structures is proposed. Attractive nearest neighbour interactions, which may be of electronic nature lead to 2-dimensional condensation. Every pair bond causes and elastic dipole. The elastic dipoles interact via substrate strains with an anisotropics ^–3 power law. Different types of adatoms or sites are permitted and many-body effects result, from the assumptions. Electric dipole interactions of adatoms are included for comparison. The model is applied to the W(110) surface and compared with superstructures experimentally found in the W(110)-0 system. It is found that there is still lack for an additional next-nearest neighbour interaction.     

Timing and detuning studies of a synchronously pumped mode-locked dye laser

The timing relationship between a synchronously pumped mode-locked dye laser and its pump laser is explored experimentally and theoretically. The experimental data is in quantitative agreement with numerical solutions of a semiclassical model that includes no free parameters. The data also agrees qualitatively with an approximate, analytic distillation of this model which is presented here for the first time. Our study has a direct bearing on the jitter and the long term stability of these important ultrashort pulse lasers.     

An applied cell mapping method for optimal control problems

From the application point of view, a series of modifications are proposed for the cell mapping method discussed in Ref. 1 for the optimal control analysis of dynamical systems. The cell order around the target set is rearranged. A set of common discriminate principles is used for the selection of the optimal one among competing control strategies of the same cost. Inequality constraints of the system are taken into account. The number of elements in the set of allowable time intervals is not prescribed, but left open. These modifications seem to make the cell mapping method more efficient for analyzing feedback systems and for obtaining their global optimal control information. The algorithms presented in this paper could broaden the application of the cell mapping approach of Ref. 1 to a wider class of engineering problems.