Random Search Approach Research Articles

WITHDRAWN: A new random search approach to find the global minimum

This article has been withdrawn consistent with Elsevier Policy on Article Withdrawal ( http://www.elsevier.com/locate/withdrawalpolicy ). The Publisher apologizes for any inconvenience this may cause.

Spatiotemporal variations in vertical gravity gradients at the Campi Flegrei caldera (Italy): a case for source multiplicity during unrest?

SUMMARY We present an evaluation of residual vertical gravity‐height change gradients obtained from gravimetric and elevation data between 1982 and 2000 at the Campi Flegrei caldera (CFc) in Italy. Spatial and temporal variations in the gradients are indicative of multiple causative sources during unrest, in particular for ground subsidence from 1988 onwards. Supported by results obtained from time-series inversion for the period 1988‐2000 using a random search approach of a purely elastic earth model and a genetic algorithm accounting for elastic-gravitational effects, we propose a centre of dilatation undergoing predominantly pressure changes yet negligible mass changes as the dominant cause for caldera deflation. Mass fluctuations in randomly active secondary sources along the periphery of the CFc can be best explained by dynamic changes along the caldera boundary (ring) faults.

A simple method to find a robust output feedback controller by random search approach

This paper presents a simple but effective method for finding a robust output feedback controller via a random search algorithm. The convergence of this algorithm can be guaranteed. Moreover, the probability to find a solution as well as the number of random trials can be estimated. The robustness of the closed-loop system is improved by the minimization of a given cost function reflecting the performance of the controller for a set of plants. Simulation studies are used to demonstrate the effectiveness of the proposed method.

Synchronisation issue of supply chains: a six-sigma constrained random search approach

Supply Chains (SC) can be viewed as a network of facilities or companies in which an order will flow through various business processes and ultimately reach the customer on time. In such networks, end-to-end material and information flows are optimally and collaboratively managed to create value for the customer and other stakeholders. The delivery performance of such networks can be optimised by pushing the workflow through the system in a way that there is a high probability of final products reaching the end user in specified delivery window. This paper addresses the problem of synchronisation of the business processes in a supply chain with the objective of minimising the total cost involved. A mathematical formulation of the problem is developed to relate the costs involved and the tolerances gained. A Genetic Algorithm (GA)-based heuristic is proposed to achieve the near optimal solution of the problem.

Dynamic responses of protein homeostatic regulatory mechanisms to perturbations from steady state

Nineteen hypothetical protein homeostatic regulatory mechanisms were constructed and analysed in terms of the rate at which they recovered from a perturbation in the steady-state concentration of any component. Systems were constructed to symbolize transcription/translation processes of the average protein from Escherichia coli (1000 copies of protein P along with 1 gene G per cell). In some model systems, G catalysed the synthesis of P directly, while in others G catalysed the synthesis of mRNA (called M), and M catalysed the synthesis of P in a subsequent step. Recovery rates for each regulatory mechanism were obtained by generating the corresponding system of differential equations, linearizing the system about the steady state, and determining eigenvalues of the associated coefficient matrix. The optimal rate of recovery for a given mechanism, R D , was determined by combining random and gradient search approaches to find rate constants for which the system recovered fastest. Regulatory elements that improved dynamic regulation were identified. These consisted of negative feedback relationships that involved P binding to either G (to shut off the synthesis of P) or M (to stimulate its degradation). Regulation improved as increasing numbers of P's bound to either G or M; however, the binding to M was more effective. In other mechanisms PP dimers bound G. Dimer-binding mechanisms were roughly twice as effective in terms of regulation as those that bound P monomers. The effect of linking two regulatory “modules” was also investigated. Linking had no effect on R D , but optimal rate constants for the linked system were similar to those of the unlinked modules, suggesting that it may be feasible to construct regulatory networks by linking individual modules of this type.

Combining Random and Systematic Search Heuristic Procedures for Solving Spatially Constrained Forest Management Scheduling Models

Abstract This article analyzes three heuristics for solving forest management scheduling models encompassing both timber production and landscape structure objectives. Emphasis is on developing an efficient solution approach that may address complex temporal and spatial interactions of forest management scheduling decisions. Two random search approaches, simulated annealing and evolution programs, are compared with a new heuristic, sequential quenching and tempering, that combines random and systematic search techniques. The three heuristics were applied to four large eucalyptus forest management problems. The test forests encompassed 300 to 900 stands. The number of management alternatives ranged from 33,000 to 220,000. Model building encompassed the generation of binary decision variables for all the problems considered. In order to address economic and ecological management objectives, all the models included timber volume flow constraints, minimum and maximum clearcut opening constraints and constraints on the minimum number of old forest patches with minimum area requirements. All constraints were defined over a temporal horizon extending to thirty 1 yr periods. Results from over 1,300 test computer runs are discussed for application to these large problems. Results show that the new strategy can be compared favorably to the random search approaches. They suggest that in order to find feasible solutions to such a complex problem, random search may be combined with a systematic search component within a heuristic procedure. FOR. SCI. 48(3):608–621.

A note on a new greedy-solution representation and a new greedy parallelizable heuristic for the traveling salesman problem

A new approach is presented to the traveling salesman problem (TSP) relying on a novel greedy representation of the solution space and leading to a different definition of neighborhood structures required in many local and random search approaches. Accordingly, a parallelizable search strategy is proposed based upon local search with random restarts that exploits the characteristics of the representation. Preliminary experimental results on several sets of test problems, among which very well-known benchmarks, show that the representation developed, matched with the search strategy proposed, attains high quality near-optimal solutions in moderate execution times.

Adaptive Random Search Approach to Identification of Neural Network Model

Adaptive Random Search Approach to Identification of Neural Network Model

Forest management models and combinatorial algorithms: analysis of state of the art

Linear Programming and Mixed Integer Linear Programs have been used for forest planning since the 60's to support decision making on forest harvesting and management. In particular, during the last two decades of forest management there has been an increased interest in spatial issues. Further, new environmental concerns, such as resource sustainability and wildlife protection, impose that increased attention be paid to activities carried out on the ground. Road building needed for access also requires spatial definiton. As a result, more complex models must be used. We discuss the issues which have led to the combinatorial nature of some main forest management problems and the solution algorithms that have been proposed for these problems, including local search heuristics, random search approaches, strengthening of mixed integer model formulations and Lagrangian relaxation. In this survey, we discuss which of the proposed approaches have been used succesfully, the advantages and shortcomings of each and what are still open research problems.

Training Neural Networks by Adaptive Random Search Techniques

A relatively simple stochastic optimization procedure based on the adaptive random search algorithm is presented to train artificial neural networks of the type encountered in applied mechanics applications. After discussing some essential features of the algorithm that influence its search efficiency, a procedure is outlined for replacing the back-propagation training approach by the new method in order to train networks involving high-dimensional parameter vectors. The method is successfully used in conjunction with a multilayer network involving a parameter vector of very high dimension. It is shown that the adaptive random search approach shifts the training effort from the user to the computer by exchanging additional computer search effort for easier training tasks on the part of the user. Extensive simulation studies are presented to provide statistically significant results related to the characteristics of the stochastic training approach. Guidelines are provided for applying the method to generic neural network training episodes.

A Controlled Random Search Technique Incorporating the Simulated Annealing Concept for Solving Integer and Mixed Integer Global Optimization Problems

In this paper, a computational algorithm, named RST2ANU algorithm, has been developed for solving integer and mixed integer global optimization problems. This algorithm, which primarily is based on the original controlled random search approach of Price [22i], incorporates a simulated annealing type acceptance criterion in its working so that not only downhill moves but also occasional uphill moves can be accepted. In its working it employs a special truncation procedure which not only ensures that the integer restrictions imposed on the decision variables are satisfied, but also creates greater possibilities for the search leading to a global optimal solution. The reliability and efficiency of the proposed RST2ANU algorithm has been demonstrated on thirty integer and mixed integer optimization problems taken from the literature. The performance of the algorithm has been compared with the performance of the corresponding purely controlled random search based algorithm as well as the standard simulated annealing algorithm. The performance of the method on mathematical models of three realistic problems has also been demonstrated.

Application of global optimization to particle identification using light scattering

Numerical methods of solving the inverse light scattering problem for spheres are presented. The methods are based on two stochastic global optimization techniques: Deep's random search and the multilevel single-linkage clustering analysis due to Rinnooy Kan and Timmer. Computational examples show that the radius and the refractive index of spheres comparable with or larger than the wavelength of light can be recovered from multiangle scattering data. While the random search approach is faster, the clustering analysis is shown to be more reliable. A general discussion of the clustering method is also given.

Genetics of food allergy

Genetics of human diseases has passed through three main historical phases: (1) studies of formal genetics aimed at investigating mechanisms of inheritance using Mendelian models in population, family and twin surveys; (2) studies of associations between HLA antigens and diseases; and (3) direct mapping of genes through candidate gene or random genomic search approach. Very few data in each of these three phases are available for food allergy and intolerance. They are mostly confined to formal genetics studies of allergy in general and to HLA association studies in celiac diseases. The main reason for paucity of data in this important area of investigation is represented by the heterogeneity of the clinical entities grouped under the label of food allergy at the level of: (1) the clinical phenotype, because of the many diseases and end-organs interested; (2) pathophysiological variables involved, since several immunological and non-immunological mechanisms can be invoked in different cases of food allergy and intolerance; and (3) the foods or their absorbed metabolites which induce symptoms. However, progress made in genetics of allergy can be in part extrapolated to the limited number of cases where an IgE-mediated mechanism has been demonstrated. The review of studies based on a more punctual definition of the allergic phenotype and of the candidate genes (regions) of allergy: (1) suggests that allergen recognition and specific IgE response, total IgE (and IgG4) polyclonal activation, up-regulation of inflammatory cells (mast cells and eosinophils mainly) and hyper-responsiveness of end-organs are possibly regulated by different genetic and environmental factors; and (2) calls particular attention on the following genomic regions: 5q31.1–33, 6p21.3, 11q13,14. Research on genetics of food allergy and intolerance is highly recommended because of its high prevalence and of the potential applicative value of results for preventive measures of dietary control in subjects at risk. Since `food allergy and intolerance' does not represent an useful phenotype for genetic studies because of its heterogeneity, adequate strategies of gene mapping should be designed in study groups selected for defined variables with a well established role in the pathogenesis of the different clinical expressions of this common condition.

Conformational Distribution of a Tetrapeptide in Solution Using a Combined Random Search and Continuum Dielectric Approach

Conformational Distribution of a Tetrapeptide in Solution Using a Combined Random Search and Continuum Dielectric Approach

Multi-stage Monte Carlo and non-linear test problems

The multi-stage Monte Carlo approach is similar to the regular Monte Carlo or random search approach that is frequently used on difficult multivariate problems. However, with multi-stage, the regular Monte Carlo search is considered to be stage one. This is followed by several more stages in an ever narrowing and re-positioning region following the best answers so far, hopefully on a trail to the optimal or a close approximation. The element of risk is certainly present in multi-stage Monte Carlo. Therefore to compare it with other algorithms for effectiveness, six test problems from the literature are selected and ‘ solved ’ with multistage. The results with multi-stage and other approaches are similar for the first five problems. However, in the last test problem, multi-stage finds many lower values than the stated ‘ minimum ’ and fourteen of them are presented here along with computer programs and a discussion of multi-stage Monte Carlo

Optimization for pattern classification using biased random search techniques

We demonstrate how optimization problems arise in the field of pattern classification, in particular in using piecewise-linear classification and classification based on an optimal linear separator. We motivate the need in this area for a general purpose optimization approach. We discuss ALOPEX, a biased random search approach, from the point of view of this need. While ALOPEX itself failed to fulfil our need, a newly-introduced generalization of it (iterated ALOPEX) was found to be appropriate for the optimization problems of our particular concern. We conclude the paper with a brief critical evaluation of this approach as compared to our original aims.

Computational algorithms based on random search for solving global optimization problems

Computational algorithms based on random search approach for solving unconstrained and con-strained global optimization problems have been proposed and tested on a number of test problems taken from literature. The results show that the proposed algorithms can be used to solve moderately sized unconstrained, inequality constrained and a class of equality constrained global optimization problems.

Optimization Design of Multivariable Control in Multimachine Power Systems

A new algorithm and model for the optimization design of multivariable control in multimachine power systems are presented in this paper. This method exploits the modified random search approach to directly optimize the output feedback gains, without solving Riccati equation. Using this approach, we can design the optimal control with all state feedback of the entire system, or the suboptimal control with partial state feedback of the entire system, or the optimal decentralized control with any state feedback of the local individual machine in the system. It has been employed for the parameters design of an optimal excitation control-ler (OEC) installed at a power station in N.W. power system of China. Re-sults show that this method can effectively search out the optimal para-meters of the controller, hence much improve dynamic performance and highly enhance the steady stability limit of the system. A comparative study of the OEC (from solving Riccati equation) and the decentralized local excitation control (DLEC) has been completed; and it is shown that both dynamic performances are very close. This means that the DLEC is a practical, valuable and perspective way for the power system control.

A general-purpose global optimizer: Implimentation and applications

This paper, written from a user stand-point, advocates the Adaptive Random Search strategy as an efficient tool for global optimization. First is presented a brief overview of the various types of methods available in the literature for global optimization, and practical advantages of the random search approach are advanced. Some modifications, which were found to improve the efficiency and versatility of the method, and a detailed description of the practical implementation of the resulting algorithm are presented. The routine is used first to treat seven test-cases from the literature for comparison purposes. Then two examples are treated related to automatic control theory. The first one is a parameter estimation problem. the second one a control problem. Finally a practical application of the method to automated registration in medical nuclear imagery is presented.