Wyniki wyszukiwania - BazTech

1

Comparative Study of PSO and CMA-ES Algorithms on Black-box Optimization Benchmarks

Szynkiewicz P.

Journal of Telecommunications and Information Technology

|

2018

|

nr 4

5--17

EN

Numerous practical engineering applications can be formulated as non-convex, non-smooth, multi-modal and ill-conditioned optimization problems. Classical, deterministic algorithms require an enormous computational effort, which tends to fail as the problem size and its complexity increase, which is often the case. On the other hand, stochastic, biologically-inspired techniques, designed for global optimum calculation, frequently prove successful when applied to real life computational problems. While the area of bio-inspired algorithms (BIAs) is still relatively young, it is undergoing continuous, rapid development. Selection and tuning of the appropriate optimization solver for a particular task can be challenging and requires expert knowledge of the methods to be considered. Comparing the performance of viable candidates against a deﬁned test bed environment can help in solving such dilemmas. This paper presents the benchmark results of two biologically inspired algorithms: covariance matrix adaptation evolution strategy (CMA-ES) and two variants of particle swarm optimization (PSO). COCO (COmparing Continuous Optimizers) – a platform for systematic and sound comparisons of real-parameter global optimization solvers was used to evaluate the performance of CMA-ES and PSO methods. Particular attention was paid to the effciency and scalability of both techniques.

2

An adaptive multi-spline refinement algorithm in simulation based sailboat trajectory optimization using onboard multi-core computer systems

Dębski R.

International Journal of Applied Mathematics and Computer Science

|

2016

|

Vol. 26, no. 2

351--365

EN

A new dynamic programming based parallel algorithm adapted to on-board heterogeneous computers for simulation based trajectory optimization is studied in the context of “high-performance sailing”. The algorithm uses a new discrete space of continuously differentiable functions called the multi-splines as its search space representation. A basic version of the algorithm is presented in detail (pseudo-code, time and space complexity, search space auto-adaptation properties). Possible extensions of the basic algorithm are also described. The presented experimental results show that contemporary heterogeneous on-board computers can be effectively used for solving simulation based trajectory optimization problems. These computers can be considered micro high performance computing (HPC) platforms—they offer high performance while remaining energy and cost efficient. The simulation based approach can potentially give highly accurate results since the mathematical model that the simulator is built upon may be as complex as required. The approach described is applicable to many trajectory optimization problems due to its black-box represented performance measure and use of OpenCL.

3

Simulation-based sailboat trajectory optimization using on-board heterogeneous computers

Dębski R.

Computer Science

|

2016

|

Vol. 17 (4)

461--481

EN

A dynamic programming-based algorithm adapted to on-board heterogeneous computers for simulation-based trajectory optimization was studied in the context of high-performance sailing. The algorithm can efficiently utilize all OpenCL-capable devices, starting the computation (if necessary, in single precision) on a GPU and finalizing it (if necessary, in double-precision) with the use of a CPU. The serial and parallel versions of the algorithm are presented in detail. Possible extensions of the basic algorithm are also described. The experimental results show that contemporary heterogeneous on-board/mobile computers can be treated as micro HPC platforms. They offer high performance (the OpenCL-capable GPU was found to accelerate the optimization routine 41 fold) while remaining energy and cost efficient. The simulation-based approach has the potential to give very accurate results, as the mathematical model upon which the simulator is based may be as complex as required. The black-box represented performance measure and the use of OpenCL make the presented approach applicable to many trajectory optimization problems.

4

Gradient-Based Algorithms in the Brachistochrone Problem Having a Black-Box Represented Mathematical Model

Dębski R.

Journal of Telecommunications and Information Technology

|

2014

|

nr 1

32--40

EN

Trajectory optimization problems with black-box represented objective functions are often solved with the use of some meta-heuristic algorithms. The aim of this paper is to show that gradient-based algorithms, when applied correctly, can be effective for such problems as well. One of the key aspects of successful application is choosing, in the search space, a basis appropriate for the problem. In an experiment to demonstrate this, three simple adaptations of gradient-based algorithms were executed in the forty-dimensional search space to solve the brachistochrone problem having a blackbox represented mathematical model. This experiment was repeated for two different bases spanning the search space. The best of the algorithms, despite its very basic implementation, needed only about 100 iterations to find very accurate solutions. 100 iterations means about 2000 objective functional evaluations (simulations). This corresponds to about 20 iterations of a typical evolutionary algorithm, e.g. ES(μ,l ).

5

Benchmarking Procedures for Continuous Optimization Algorithms

Opara K., Arabas J.

Journal of Telecommunications and Information Technology

|

2011

|

nr 4

73-80

EN

Reliable comparison of optimization algorithms requires the use of specialized benchmarking procedures. This paper highlights motivations which influence their structure, discusses evaluation criteria of algorithms, typical ways of presenting and interpreting results as well as related statistical procedures. Discussions are based on examples from CEC and BBOB benchmarks. Moreover, attention is drawn to these features of comparison procedures, which make them susceptible to manipulation. In particular, novel application of the weak axiom of revealed preferences to the field of benchmarking shows why it may be misleading to assess algorithms on basis of their ranks for each of test problems. Additionally, an idea is presented of developing massively parallel implementation of benchmarks. Not only would this provide faster computation but also open the door to improving reliability of benchmarking procedures and promoting research into parallel implementations of optimization algorithms.