Comparison of genetic algorithms, random restart and two-opt switching for solving large location-allocation problems

Houck, Christopher R.; Joines, Jeffrey A.; Kay, Michael G.

doi:10.1016/0305-0548(95)00063-1

Cited by 148 publications

(76 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Therefore, traditional search techniques based on local linearization, which use characteristics of the problem to determine the next sampling point (e.g., gradients, Hessians, linearity, and continuity), are computationally efficient but fail to identify the best-fit solution when the starting model is too far from the global optimal solution. On the other hand, stochastic search techniques (e.g., genetic algorithms and simulated annealing) have been shown to efficiently identify promising regions in the search space, but their convergence rate has been shown to be poor in localized searches (Bersini and Renders, 1994;Houck et al, 1996).…”

Section: Stochastic Seismogram Inversion Algorithm With Deterministicmentioning

confidence: 99%

Site Amplification and Attenuation via Downhole Array Seismogram Inversion: A Comparative Study of the 2003 Miyagi-Oki Aftershock Sequence

Assimaki¹,

Wei²,

Steidl³

et al. 2008

Bulletin of the Seismological Society of America

View full text Add to dashboard Cite

Weak-motion geotechnical array recordings at 38 stations of the Japanese strong-motion network KiK-Net from the 2003 M w 7:0 Miyagi-Oki aftershock sequence are used here to quantify the amplification and attenuation effects of nearsurface formations to incident seismic motion. Initially, a seismic waveform optimization algorithm is implemented for the evaluation of high-resolution, low-strain velocity (V s ), attenuation (Q s ), and density (ρ) profiles at the sites of interest. Based on the inversion results, V s versus Q s correlations are developed, and scattering versus intrinsic attenuation effects are accounted for in their physical interpretation. Surfaceto-downhole traditional spectral ratios (SSR), cross-spectral ratios (c-SSR), and horizontal-to-vertical (H/V) site-response estimates are next evaluated and compared, while their effectiveness is assessed as a function of the site conditions classified on the basis of the weighted average V s of the upper 30 m (V s30 ) of the formations. Single and reference-station site-response estimates are successively compared to surface-to-rock outcrop amplification spectra and are evaluated by deconvolution of the downhole records based on the inversion results; comparison of the observed SSR and estimated surface-to-rock outcrop amplification spectra illustrates the effects of destructive interference of downgoing waves at the downhole instrument level as a function of the site class. Site amplification factors are successively computed in reference to the National Earthquake Hazards Reduction Program (NEHRP) B-C boundary site conditions (V s30 760 m=sec), and results are compared to published values developed on the basis of strong-motion data and site-response analyses. Finally, weak-motion SSR estimates are compared to the mainshock spectra, and conclusions are drawn for the implications of soil nonlinearity in the near surface. Results presented in this article suggest that currently employed site classification criteria need to be reevaluated to ensure intraclass consistency in the assessment of amplification potentials and nonlinearity susceptibility of near-surficial soil formations.

show abstract

Section: Stochastic Seismogram Inversion Algorithm With Deterministicmentioning

confidence: 99%

Site Amplification and Attenuation via Downhole Array Seismogram Inversion: A Comparative Study of the 2003 Miyagi-Oki Aftershock Sequence

Assimaki¹,

Wei²,

Steidl³

et al. 2008

Bulletin of the Seismological Society of America

View full text Add to dashboard Cite

show abstract

“…For example, the various versions of variable neighborhood search in the above comparative study use Cooper's algorithm in the local search step. The initial population in the genetic algorithm of Houck et al (1996) is obtained by repeating Cooper's algorithm from random starting points until an adequate number of local minima is found, and after the crossover operation, the new solution is improved (mutation operation) using the Cooper algorithm. For a further update on metaheuristic-based methods for solving the continuous p-median problem, see Brimberg et al (2008a).…”

Section: The Continuous P-median Problemmentioning

confidence: 99%

Hub Location Problems: The Location of Interacting Facilities

Kara

Taner

2011

International Series in Operations Research &Amp; Management Science

View full text Add to dashboard Cite

), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. PrefaceThis book is the final result of a number of incidents that occurred to us while discussing location issues with colleagues at conferences. Frequently we attended presentations, in which the authors quoted the well-known references that helped to make the discipline what it is today. Upon further inquiry, though, it turned out that some of these colleagues had never actually read the original papers. We then discussed among ourselves which contributions could be credited with shaping the field. And, lo and behold, we found that we, too, had neglected to read some of the papers that form the foundation of our science. Whether it was laziness or other things that got in the way, it had become clear that something had to be done. Our first thought was to collect the original contributions (once we could agree on what they were) and reprint them. When discussing this possibility with a publisher, we immediately ran into a roadblock in the form of copyright. While this appeared to have stopped our enthusiastic effort dead in its tracks, we kept on collecting and reading what we considered original contributions.This went on until we met Camille Price, who suggested that, rather than reprinting the original contributions, we should invite some of the leaders in our field and ask them to describe the original contribution, explain and interpret it, and comment on the impact that it had to the field. This was, of course, an excellent idea, and the response by our colleagues to our pertinent requests was equally enthusiastic. What you hold in your hands is the result of this effort.In other words, the purpose of this book is to provide easy access to the main contributions to location theory. The book is organized as follows. The introductory chapter provides an overview of some of the many facets of location analysis. This is followed by contributions in the main three fields of inquiry: minisum, minimax, and covering problems. The next chapters are part of an ever-growing list of nonstandard location models: models including competitive components, those that locate undesirable facilities, those with probabilistic features, and those that allow interactions between facilities. The following chapters discuss solution techniques: after a discussion of exact and heuristic techniques, we devote an entire chapter to Weiszfeld's method, and another to Lagrangean techniques. The last chapters of this book deal with the spheres of influence that the facilities generate and that attract viii customers to them, an...

show abstract

“…In particular, a nonlinear Gauss-Newton scheme is employed, forcing the active parental generation to convergence to local minima or maxima prior to mutation, crossover, and reproduction. This technique, referred to as the hill-climbing method of local optimization, has been shown to significantly enhance the performance of genetic algorithms (Stoffa and Sen, 1991;Houck et al, 1996). The objective function of the nonlinear least-squares optimization is defined in the frequency domain, as the energy error between the model and data vectors:…”

Section: CMmentioning

confidence: 99%

“…This optimization technique comprises a genetic algorithm in the wavelet domain coupled to a nonlinear least-squares fit in the frequency domain, and it has been shown to improve the computational efficiency of the former while avoiding the pitfalls of using local linearization techniques such as the latter (Houck et al, 1996). The parameters estimated are stepwise variations of the shear-wave velocity, attenuation, and density with depth for horizontally layered media with predefined layer thickness.…”

Section: Introductionmentioning

confidence: 99%

Quantifying Nonlinearity Susceptibility via Site-Response Modeling Uncertainty at Three Sites in the Los Angeles Basin

Assimaki¹,

Li²,

Steidl³

et al. 2008

Bulletin of the Seismological Society of America

View full text Add to dashboard Cite

The effects of near-surface soil stratigraphy on the amplitude and frequency content of ground motion are accounted for in most modern U.S. seismic design codes for building structures as a function of the soil conditions prevailing in the area of interest. Nonetheless, currently employed site-classification criteria do not adequately describe the nonlinearity susceptibility of soil formations, which prohibits the development of standardized procedures for the computationally efficient integration of nonlinear ground response analyses in broadband ground-motion simulations. In turn, the lack of a unified methodology for nonlinear site-response analyses affects both the prediction accuracy of site-specific ground-motion intensity measures and the evaluation of site-amplification factors when broadband simulations are used for the development of hybrid attenuation relations. In this article, we introduce a set of criteria for quantification of the nonlinearity susceptibility of soil profiles based on the site conditions and incident ground-motion characteristics, and we implement them to identify the least complex ground response prediction methodology required for the simulation of nonlinear site effects at three sites in the Los Angeles basin. The criteria are developed on the basis of a comprehensive nonlinear site-response modeling uncertainty analysis, which includes both detailed soil profile descriptions and statistical adequacy of ground-motion time histories. Approximate and incremental nonlinear models are implemented, and the limited site-response observations are initially compared to the ensemble site-response estimates. A suite of synthetic ground motions for rupture scenarios of weak, medium, and large magnitude events (M 3:5-7:5) is next generated, parametric studies are conducted for each fixed magnitude scenario by varying the source-to-site distance, and the variability introduced in ground-motion predictions is quantified for each nonlinear site-response methodology. A frequency index is developed to describe the frequency content of incident ground motion relative to the resonant frequencies of the soil profile, and this index is used in conjunction with the rock-outcrop acceleration peak amplitude (PGA RO ) to identify the site conditions and ground-motion characteristics where incremental nonlinear analyses should be employed in lieu of approximate methodologies. We show that the proposed intensity-frequency representation of ground motion may be implemented to describe the nonlinearity susceptibility of soil formations in broadband simulations by accounting both for the magnitude-distance-orientation characteristics of seismic motion and the profile stiffness characteristics. The synthetic ground-motion predictions are next used for the development of site-amplification factors for the alternative site-response methodologies, and the results are compared to published site factors of attenuation relations. For the site conditions investigated, currently established amplification fac...

show abstract

Comparison of genetic algorithms, random restart and two-opt switching for solving large location-allocation problems

Cited by 148 publications

References 4 publications

Site Amplification and Attenuation via Downhole Array Seismogram Inversion: A Comparative Study of the 2003 Miyagi-Oki Aftershock Sequence

Site Amplification and Attenuation via Downhole Array Seismogram Inversion: A Comparative Study of the 2003 Miyagi-Oki Aftershock Sequence

Hub Location Problems: The Location of Interacting Facilities

Quantifying Nonlinearity Susceptibility via Site-Response Modeling Uncertainty at Three Sites in the Los Angeles Basin

Contact Info

Product

Resources

About