Conventional Simulated Annealing Research Articles

Segmented Simulated Annealing Applied to Sliding Mode Controller Design

This paper develops an optimisation technique based on Simulated Annealing (SA). It outlines conventional SA theory and then describes a variation which has been called Segmented Simulated Annealing (SSA). SSA is a method which selects the best results from numerous conventional SA runs where each SA starts at a different point in the search space. Conventional SA and SSA are compared by applying these techniques to the optimisation of two decoupled sliding mode controllers for a submarine simulation. These results are also compared with hand-tuned responses, which show that these optimisation techniques are more convenient to use than conventional hand-tuning methods.

Application of Very Fast Simulated Annealing to the Determination of the Crustal Structure Beneath Tibet

SUMMARY The simulated annealing method has recently been applied to several multiparameter optimization problems, including those of geophysical inversion. A new variant of simulated annealing, called very fast simulated annealing (VFSA), overcomes some of the drawbacks of a conventional simulated annealing; it has been found to be a practical tool for geophysical inversion (Sen & Stoffa 1995). The method is particularly useful for non-linear problems with multiple parameters where a grid-search method is prohibitively expensive. Here, we have applied VFSA to retrieve the crustal structure beneath seismic stations in Tibet using teleseismic body-waveform data. Our approach is to compare the radial-component records with generalized ray synthetics directly in the time domain. For any given crustal structure, we have formulated a direct relationship between the radial- and the vertical-component signal, so that we can generate synthetic radial-component data from recorded vertical-component seismograms (Zhao & Frohlich 1996). We have tested the VFSA inversion algorithm using synthetics and confirmed that it works very well, i.e. it finds solutions very close to the true solution. Using the algorithm, we have determined the crustal structures beneath 11 stations in Tibet. From south to north, the total crustal thickness is quite uniform. Our tectonic interpretation of these crustal models is that they may represent crust from the Eurasian plate injected beneath the crust of the converging Indian plate. Certain features of the models are consistent with the presence of a small convection cell or plume beneath north-central Tibet, as suggested by Molnar (1990).

On mapping production systems onto multiprocessors

In this paper, we consider the problem of mapping rules in a production system onto nodes in a multiprocessor system. Since this problem is NP-hard, we present an efficient heuristic algorithm based on simulated annealing to solve this mapping problem. Compared to conventional simulated annealing algorithms to solve this problem, where moves or exchanges are used exclusively as perturbation functions, we use a combination of moves and exchanges to achieve consistently better solutions in a reasonable amount of computation time. Further, by restricting the random acceptance of configurations (mappings) during the annealing process, we achieve a faster convergence. We demonstrate the effectiveness of our algorithm by comparing its performance with that of another recently proposed algorithm for mapping rule based systems onto multiprocessors.

Cooling-rate dependence of the ground-state energy using microcanonical simulated annealing.

We have studied the cooling-rate dependence of approaching the ground state in several spin glass models within the microcanonical simulated annealing scheme. The final state energies of the two-dimensional (2D) and 3D Gaussian models and the 3D \ifmmode\pm\else\textpm\fi{}J model were found to depend on the cooling rate \ensuremath{\Delta}E/\ensuremath{\tau}, where \ensuremath{\Delta}E is the energy decrement in the microcanonincal simulation and \ensuremath{\tau} is the number of Monte Carlo steps per spin at each annealing stage. In particular the ground-state energy dependence on \ensuremath{\tau} for all models studied shows a stretched exponential form for a wide range of the values \ensuremath{\tau}. This behavior is very different from that of the conventional simulated annealing algorithms. \textcopyright{} 1996 The American Physical Society.

The annealing evolution algorithm as function optimizer

In this paper an annealing evolution algorithm is applied to the optimization of continuous problems. The algorithm uses an evolutionary strategy to guide the search in simulated annealing. Furthermore, we discuss an implementation of the algorithm and compare its performance with the conventional simulated annealing algorithm and the parallel genetic algorithm. The performance evaluation is carried out for a standard set of test functions from the literature. Here a breakthrough can be reported. The annealing evolution algorithm is able to find the global minimum of Rastrigin's function of dimension 500 on the VAX 8600 machine.

Molecular conformation on the CM-5 by parallel two-level simulated annealing

In this paper, we propose a new kind of simulated annealing algorithm calledtwo-level simulated annealing for solving certain class of hard combinatorial optimization problems. This two-level simulated annealing algorithm is less likely to get stuck at a non-global minimizer than conventional simulated annealing algorithms. We also propose a parallel version of our two-level simulated annealing algorithm and discuss its efficiency. This new technique is then applied to the Molecular Conformation problem in 3 dimensional Euclidean space. Extensive computational results on Thinking Machines CM-5 are presented. With the full Lennard-Jones potential function, we were able to get satisfactory results for problems for cluster sizes as large as 100,000. A peak rate of over 0.8 giga flop per second in 64-bit operations was sustained on a partition with 512 processing elements. To the best of our knowledge, ground states of Lennard-Jones clusters of size as large as these have never been reported before.

THE ANNEALING EVOLUTION ALGORITHM

ABSTRACT An annealing evolution algorithm is applied to the optimization of continuous problems. The algorithm uses an evolutionary strategy to guide the search in simulated annealing. Furthermore, we discuss an implementation of the algorithm and compare its performance with the conventional simulated annealing algorithm and the parallel genetic algorithm. The performance evaluation is carried out for a standard set of test functions from the literatures. Here a breakthrough can be reported. The annealing evolution algorithm is able to find the global minimum of Rastrigin's function of dimension 500 on the VAX8600 machine.

Multiresolution texture segmentation with application to diagnostic ultrasound images

A multiresolution texture segmentation (MTS) approach to image segmentation that addresses the issues of texture characterization, image resolution, and time to complete the segmentation is presented. The approach generalizes the conventional simulated annealing method to a multiresolution framework and minimizes an energy function that is dependent on the resolution of the size of the texture blocks in an image. A rigorous experimental procedure is also proposed to demonstrate the advantages of the proposed MTS approach on the accuracy of the segmentation, the efficiency of the algorithm, and the use of varying features at different resolution. Semireal images, created by sampling a series of diagnostic ultrasound images of an ovary in vitro, were tested to produce statistical measures on the performance of the approach. The ultrasound images themselves were then segmented to determine if the approach can achieve accurate results for the intended ultrasound application. Experimental results suggest that the MTS approach converges faster and produces better segmentation results than the single-level approach.

Extraction of the parameters of equivalent circuits of microwave transistors using tree annealing

The problem of extracting a physically based equivalent circuit model for a heterostructure bipolar transistor (HBT) from S-parameter measurements is solved with a new formulation of simulated annealing. The physical model necessary for an accurate representation of the HBT leads to an extraction problem with many local minima. A satisfactory minimum can be found by conventional gradient-based techniques only with considerable expert guidance. The proposed algorithm finds equivalent circuits as good as those from conventional techniques but without human intervention. It is more efficient than conventional stochastic simulated annealing because it accumulates a probability density of good equivalent circuits which it subsequently uses to refine its statistical search for the best equivalent circuit.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Relational Matching Using a Neural Network

AbstractThis paper proposes an algorithm for relational matching, which is important in pattern recognition, by using the Hopfield model as one of the neural network models. First, the objective function for relational matching is derived and the computer simulation of the neural networking for relational matching according to this objective function is conducted by use of a simple model. Next, the local minima in connection with the Hopfield model is avoided by using triangular noise and its effectiveness is verified in the network for relational matching. This method differs from the conventional simulated annealing method and, by adding noise to the network, the network will reach only one stationary state for which results will be derived by this method.

Nonconvex optimization by fast simulated annealing

Recent advances in the solution of nonconvex optimization problems use simulated annealing techniques that are considerably faster than exhaustive global search techniques. This letter presents a simulated annealing technique, which is t/log (t) times faster than conventional simulated annealing, and applies it to a multisensor location and tracking problem.