模体厚度对MC程序计算速度与效率的影响研究

基于蒙特卡罗(MC)程序MCNP4c, DOSXYZnrc, DOSRZnrc 和DPM， 考察了模体厚度变化对MC程序计算速度和MC效率的影响， 发现模体的厚度变化对不同MC程序速度提升的影响差异较大， 其中DOSXYZnrc 提升最多， DOSRZnrc提升最小。 从MC效率方面来讲， DPM和DOSRZnrc具有明显的优势。 就4个程序而言， 薄模体均表现出一定的MC 效率优势。 从MC 效率提升角度来看， MCNP4c， DOSXYZnrc和DOSRZnrc表现出一定的相似性， 而DPM相对低些。 也对DOSRZnrc异常的统计不确定性特征、 速度提升进行了研究， 发现其内部植入的光子强迫方差减小技巧在降低体元统计不确定性的同时， 也削弱了薄模体的速度优势。 但是由于显著影响了方差， 所以总体来讲， 仍然提高了DOSRZnrc薄模体的MC 模拟效率。

Effect of Phantom Thickness Variation on Speed and Efficiency for Different MC Codes

The effect of the phantom thickness on the simulation speed and efficiency of Monte Carlo (MC) codes, MCNP4c, DOSXYZnrc, DOSRZnrc and DPM, is investigated. The result shows that the increases of the simulation speed for different MC codes are distinct difference with dicreasing the phantom thickness, in which DOSXYZnrc is the most one, and DOSRZnrc is the least one. In the viewpoint of MC efficiency, DPM and DOSRZnrc have the remarkable advantages over others; and the efficiency for the thinner phantom shows the consistent advantage over the thicker one for these MC codes. In the viewpoint of MC efficiency increase, MCNP4c, DOSXYZnrc and DOSRZnrc show the same comparability, but DPM shows relatively low characteristics. The strange characteristics of DOSRZnrc in statistical uncertainty, speed increase are further investigated to find that it is due to the photon forcing VRT (variance reduction technique) implemented in it. This VRT reduces the statistical uncertainty, but it also weakens the speed advantage of the thinner phantom. However it is due to the remarkably under estimate of the variance on the interest region, the MC efficiency of DOSRZnrc is still increasing for the thinner phantom.