Font Size: a A A

Study On Structure Optimization Of Pt-Pd Alloy Nanoparticles With CUDA-based Parallel Particle Swarm Optimization Algorithm

Posted on:2018-03-07Degree:MasterType:Thesis
Country:ChinaCandidate:Z P LiFull Text:PDF
GTID:2381330515953663Subject:Systems Engineering
Abstract/Summary:PDF Full Text Request
The structure research of Pt-Pd based nanoparticles has been a hot issue in material fields because of it ' s extraordinary catalytic and optical potential.There are lots of intelligent algorithm involved in studying the stable structure of Pt-Pd based nano-particles,such as particle swarm optimization(PSO),differential evolutionary algorithm.Most of them perform well in optimization with small atomic scale.However,their computational efficiency decrease with the increasing particles size as computation greatly improve.Thus,in this paper,we propose a CUDA based parallel PSO to improve the performance of traditional PSO.In our simulations,for the above Pt-Pd based particles,the Q-SC potential is adopted to describe the interactions between atoms.The Q-SC potential provides a proper basic property model for nanocrystals,leading to an accurate prediction of the crystals' structure.For the problem,parameters are optimized to describe the surface,cohesive,and vacancy formation energies and other parameters for better simulating the metallic nanocrystal structures.Moreover,we propose a CUDA based parallel PSO(GPSO)for calculating the stable structure of Pt-Pd cluster.In the algorithm,two dimensional coding method of population thread block is designed which could be adopt in other works.And combined with the hardware characteristics of GPU,the GPSO could select optimal threads for computing while ensuring the best operation efficiency.And the data access speed of GPSO is improved by using the combined access technology.To validate the proposed GPSO,we simulate the stable structure of Pt-Pd particles for different size.And we adopt three index for validation:equal number of iterations?equal scale precision and equal scale termination condition.Moreover,the convergence and stability of the algorithm are examined.Simulation results show that,our proposed GPSO is effective and stable to optimize the structure of alloy nanoparticles.The research results have practical significance and application value to improve the efficiency of structure study of nanoparticles.
Keywords/Search Tags:parallel PSO, nanoparticles, accelerating ratio
PDF Full Text Request
Related items