Automated, Parallel Optimization of Stochastic Functions Using a Modified Simplex Algorithm

This paper proposes a framework and new parallel algorithm for optimization of stochastic functions based on a downhill simplex algorithm. The function to be optimized is assumed to be subject to random noise, the variance of which decreases with sampling time, this is the situation expected for man...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2010 Sixth IEEE International Conference on E-Science Workshops s. 98 - 103
Hlavní autoři: Chahal, D, Stuart, S J, Goasguen, S, Trout, C J
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.12.2010
Témata:
Computational modeling
Noise
Noise measurement
Object oriented modeling
Optimization
parallel optimizaton
Servers
simplex
Stochastic processes
water model
ISBN:9781424489886, 1424489881
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This paper proposes a framework and new parallel algorithm for optimization of stochastic functions based on a downhill simplex algorithm. The function to be optimized is assumed to be subject to random noise, the variance of which decreases with sampling time, this is the situation expected for many real-world and simulation applications where results are obtained from sampling, and contain experimental error or random noise. The proposed optimization method is found to be comparable to previous stochastic optimization algorithms. The new framework is based on a master-worker architecture where each worker runs a parallel program. The parallel implementation allows the sampling to proceed independently on multiple processors, and is demonstrated to scale well to over 100 vertices. It is highly suitable for clusters with an ever increasing number of cores per node. The new method has been applied successfully to the reparameterization of the TIP4P water model, achieving thermodynamic and structural results for liquid water that are as good as or better than the original model, with the advantage of a fully automated parameterization process.
ISBN:9781424489886
1424489881
DOI:10.1109/eScienceW.2010.25