Adaptive Dynamic Method Research Articles

Measure dynamics on a one-dimensional continuous trait space: theoretical foundations for adaptive dynamics

The measure dynamics approach to modelling single-species coevolution with a one-dimensional trait space is developed and compared to more traditional methods of adaptive dynamics and the Maximum Principle. It is assumed that individual fitness results from pairwise interactions together with a background fitness that depends only on total population size. When fitness functions are quadratic in the real variables parameterizing the one-dimensional traits of interacting individuals, the following results are derived. It is shown that among monomorphisms (i.e. measures supported on a single trait value), the continuously stable strategy (CSS) characterize those that are Lyapunov stable and attract all initial measures supported in an interval containing this trait value. In the cases where adaptive dynamics predicts evolutionary branching, convergence to a dimorphism is established. Extensions of these results to general fitness functions and/or multi-dimensional trait space are discussed.

Distributed interface bits: dynamic dialogue composition from ambient computing resources

This paper discusses a particular issue in the context of disappearing computing, namely, user mobility. Mobile users may carry with them a variety of wireless gadgets while being immersed in a physical environment encompassing numerous computing devices. In such a situation, it is most likely that the number and type of devices may dynamically vary during interactions. The Voyager development framework supports the implementation of ambient dialogues, i.e., dynamically distributed user Interfaces, which exploit, on-the-fly, the wireless devices available at a given point in time. This paper describes the Voyager implementation, focusing on: device discovery and registry architecture, device-embedded software implementation, ambient dialogue style and corresponding software toolkit development, and a method for dynamic interface adaptation, ensuring dialogue state persistence. Additionally, this paper presents two ambient dialogue applications developed using Voyager, namely, a game and a navigator.

Improving the Convergence of the Backpropagation Algorithm Using Local Adaptive Techniques

Since the presentation of the backpropagation algorithm, a vast variety of improvements of the technique for training a feed forward neural networks have been proposed. This article focuses on two classes of acceleration techniques, one is known as Local Adaptive Techniques that are based on weightspecific only, such as the temporal behavior of the partial derivative of the current weight. The other, known as Dynamic Adaptation Methods, which dynamically adapts the momentum factors, , and learning rate, , with respect to the iteration number or gradient. Some of most popular learning algorithms are described. These techniques have been implemented and tested on several problems and measured in terms of gradient and error function evaluation, and percentage of success. Numerical evidence shows that these techniques improve the convergence of the Backpropagation algorithm. Keywords— Backpropagation, Dynamic Adaptation Methods, Local Adaptive Techniques, Neural networks.

Method and apparatus for dynamic adaptation of a large vocabulary speech recognition system and for use of constraints from a database in a large vocabulary speech recognition system

The vocabulary of a large-vocabulary speech recognition system is structured to effectuate the rapid and efficient edition of words to a lexical network storing the vocabulary. The lexical network is structured to include Phonetic Constraint Nodes which organize the inter-word phonetic information in the network, and Word Class Nodes which organize the syntactic/semantic information in the network. Network fragments, corresponding to phoneme pronunciations and labeled to specify permitted interconnections, are precompiled to facilitate the rapid generation of pronunciation for new words and thereby enhance the rapid addition of words to the vocabulary even during speech recognition. Different language models and different vocabularies for different portions of a discourse are invoked dependent in part on the discourse history.

Dynamic adaptation method for numerical solution of axisymmetric Stefan problems

Abstract Application of the dynamic adaptation method for the numerical solution of multidimensional axisymmetric Stefan problems with explicit tracking of interfaces is presented. The dynamic adaptation method is based on the idea of transition of the physical coordinate system to the non‐stationary coordinate system. The results of computational experiments for modelling the action of high energy fluxes on metals are given.

DYNAMIC ADAPTATION METHOD FOR NUMERICAL SOLUTION OF AXISYMMETRIC STEFAN PROBLEMS

Application of the dynamic adaptation method for the numerical solution of multidimensional axisymmetric Stefan problems with explicit tracking of interfaces is presented. The dynamic adaptation method is based on the idea of transition of the physical coordinate system to the non‐stationary coordinate system. The results of computational experiments for modelling the action of high energy fluxes on metals are given.

Adaptive Dynamics of Parabolic Replicators

The growth of many artificial replicators is approximately parabolic (sub-exponential) in solution, due to the self-inhibition through duplex formation by the association of single-stranded molecules. This type of growth implies “survival of everybody” under a selection constraint. Parabolic growth requires high enough concentration so that the single strands can find one another. The selective outcome is more complicated when spontaneous decay of molecules is also taken into account. When double strands decompose at a slower rate than single strands, coexistence or survival of the fittest becomes a quantitative issue. Here we investigate the evolution of parabolic replicators by the methods of adaptive dynamics. Directional selection for higher replication rate in general results in a “parabolic quasi-species”, due to the fact that the fittest template is followed by a moving shadow of inferior templates that owe their presence to parabolic growth. Under the assumption of cross-hybridisation between non-identical templates molecular coexistence disappears when such pairing is sufficiently non-selective, because replicators do not inhibit themselves more than they limit the others. At intermediate specificity of pairing adaptive branching of the population becomes feasible, due to the fact that distant enough sequences are able to escape from cross-limitation by other sub-populations.

On the evolutionary coexistence of parasite strains

Classical models of parasite competition show that coexistence is impossible if different strains give complete cross-immunity. However, parasite coexistence is possible if some of the model assumptions are changed. For instance, coexistence is impossible if density-dependence operates only in hosts' fertility, but surprisingly becomes possible if hosts' mortaility is density-dependent. Parasite strains can also coexist if a host already infected with one strain may become infected by another strain (superinfection). I examine here if these reasons for coexistence carry over to evolutionary timescales: in other words, suppose that potentially a continuum of parasite strains may arise by mutations; will evolution arrive at a halt? in that case, will only one or several strains persist? The paradigm and methods of adaptive dynamics are used in this study. It is found, under reasonably general assumptions, that a unique evolutionarily stable state for virulence, α *, exist for both models. However, the pattern of the invasibility plots depends on the shape of the trade-off (between virulence and transmissibility, or superinfection rates) functions, and on the host demography. In many cases, the state α * is evolutionarily stable only with respect to small mutations, not to larger ones; hence, evolutionary dynamics will bring virulence to α * only if mutations are sufficiently small; for larger mutations, evolutionary dynamics are more complex and still mainly unresolved.

Solution of the Multi-interface Stefan Problem by the Method of Dynamic Adaptation

AbstractThe application of the dynamic adaptation method to the numerical solution of multidimensional Stefan problems with explicit tracking of interfaces is considered. The dynamic adaptation method is based on the idea of transition to any non-stationary coordinate system. The results of the numerical experiment on modelling the pulsed action of high-energy fluxes on metals are presented.

Solution of two‐dimensional multi‐interface Stefan problem by the method of dynamic adaptation

Abstract In the present work a method of numerical solution of multi‐interface two‐dimensional Stefan problem with explicit tracking of the interfaces in the domains of arbitrary form is considered. The method is based on the idea of dynamic adaptation of the calculated grid by means of transition to an arbitrary non‐stationary coordinate system. The coordinate system transformation is controlled by the solution. The method is described by using the example of the problem that is typical for treatment of materials with concentrated energy fluxes.

3D simulations of turbulent deflagrations using dynamic grid adaptation

AbstractWe present a CFD code for simulating turbulent deflagrations in gas mixtures for studies of explosion accidents. A dynamic grid adaptation method is used to increase the accuracy of the calculation in regions of interest, such as at flame and shock fronts. The numerical simulations compared with experimental data and is shown to yield strongly improved results at at modest cost in computational time.

SIMULATION OF LASER MELTING AND EVAPORATION OF SUPERCONDUCTING CERAMICS

The peculiarities of pulsed laser melting and evaporation of the superconducting ceramics are analyzed by means of numerical simulation. The appearance of the overheated metastable states in solid and liquid phases is shown as a result of the phase front dynamics and volume nature of laser energy release. A method of dynamic adaptation for the multifront Stefan problem is proposed.

Use of a Dynamic Grid Adaptation in the Asymmetric Weighted Residual Method

A dynamic grid adaptive method has been developed for use with the asymmetric weighted residual method. The method automatically adapts the number and position of the spatial mesh points as the sol...