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.     

Acoustic field in a closed layered shell with resonant systems

The acoustic field inside a shell excited by a spatially inhomogeneous harmonic pressure field is studied. The shell is assumed to have a finite length, a set of orthogonal stiffening ribs, two ends bounding the acoustic volume, and a sound-insulating structure, which includes layers of sound-insulating material, resonant elements, and an interior panel. The shell is considered to be orthotropic with boundary conditions corresponding to a free support. For the acoustic field in the closed volume, analytical expressions are derived with allowance for the elastoacoustic interaction of the shell with the sound-insulating layers and with the medium both inside the shell (with arbitrary impedance values at the ends) and outside it. These expressions are used to investigate the effect of different types of resonant systems on the sound field inside the shell.     

Specific features of photoisomerization of provitamin D 3 in a nematic liquid crystal

Photoisomerization of provitamin D ₃ (7-dehydrocholesterol) in a nematic liquid crystal (ZLI-1695, Merck) is investigated in detail by UV absorption spectroscopy. It is found that dissolution of chiral molecules of provitamin D ₃ induces the cholesteric phase in a nematic. The spectral kinetics of photoisomerization in this phase changes significantly from that in an ethanol solution. A sharp nonmonotonic dependence of the increase in accumulation of trans isomer tachysterol in a liquid crystal matrix with a decrease in the induced cholesteric pitch from 2200 to 25 μm is revealed.     

Effect of hydrogen on the photoelectronic properties of GaAs/InGaAs quantum-well heterostructures with an island palladium layer on the surface

The effect of hydrogen on the photoluminescence and planar conductivity of GaAs/InGaAs quantum-well heterostructures with an island Pd layer at the anodically oxidized surface was studied. Unlike continuous deposited Pd layers, island layers do not cause the formation of defects in the GaAs surface region and yet the Pd layer maintains high catalytic activity with respect to hydrogen. It is found that the thermal treatment of such a structure in a hydrogen atmosphere causes atomic-hydrogen passivation of the defects in quantum wells. Studies of the characteristics of planar photoresistors with an island Pd layer acting as hydrogen sensors show that their hydrogen detectivity is approximately two orders of magnitude higher than that of diode structures with continuous Pd layers.     

Polynomial-time computation of the degree of a dominant morphism in zero characteristic. II

Consider a projective algebraic variety W which is an irreducible component of the set of all common zeros of a family of homogeneous polynomials of degrees less than d in n + 1 variables over a field of zero characteristic. Consider a dominant rational morphism from W to W′ given by homogeneous polynomials of degree d′. We suggest algorithms for constructing objects in general position related to this morphism. They generalize some algorithms from the first part of the paper to the case dim W > dim W′. These algorithms are deterministic and polynomial in (dd′)n and the size of the input. Bibliography: 12 titles. __________ Translated from Zapiski Nauchnykh Seminarov POMI, Vol. 325, 2005, pp. 181–224.     

Modelling recruitment training in mathematical human resource planning

This paper deals with mathematical human resource planning; more specifically, it suggests a new model for a manpower‐planning system. In general, we study a k‐classed hierarchical system where the workforce demand at each time period is satisfied through internal mobility and recruitment. The motivation for this work is based on various European Union incentives, which promote regional or local government assistance programs that could be exploited by firms not only for hiring and training newcomers, but also to improve the skills and knowledge of their existing personnel. In this respect, in our augmented mobility model we establish a new ‘training/standby’ class, which serves as a manpower inventory position for potential recruits. This class, which may very well be internal or external to the system, is incorporated into the framework of a non‐homogeneous Markov chain model. Furthermore, cost objectives are employed using the goal‐programming approach, under different operating assumptions, in order to minimize the operational cost in the presence of system's constraints and regulations. Copyright © 2002 John Wiley & Sons, Ltd.     

The Markov power moment problem in problems of controllability and frequency extinguishing for the wave equation on a half-axis

In this paper necessary and sufficient conditions of null-controllability and approximate null-controllability are obtained for the wave equation on a half-axis. Controls solving these problems are found explicitly. Moreover, bang-bang controls solving the approximate null-controllability problem are constructed with the aid of solutions of a frequency extinguishing problem in the restricted band (−a,a) for this equation and the Markov power moment problem.