使用神经网络优化 Woodcock 粒子追踪方法
Optimization of the Woodcock Particle Tracking Method Using Neural Network
AI总结:
本文借鉴 PINN,用神经网络表示并优化位置相关的 Woodcock 接受率 $q(x)$,以 FOM 为损失函数,在一维高斯吸收截面输运问题中显著提升 FOM。
摘要:
Woodcock 追踪算法中的接受率被推广为任意的位置相关变量 $q(x)$。本文使用神经网络优化 $q(x)$,并将 FOM 值用作损失函数。该思想源于物理信息神经网络(PINN),即用神经网络表示微分方程的解。在这里,神经网络 $q(x)$ 应当求解使 FOM 最优化的泛函方程。对于一个具有高斯吸收截面的一维输运问题,我们观察到,与常数 $q$ 情形以及原始 Woodcock 方法相比,FOM 值有显著提升。将神经网络 Woodcock(NNW)方法推广到三维体素模型仍有待探索。
英文摘要:
The acceptance rate in Woodcock tracking algorithm is generalized to an arbitrary position-dependent variable $q(x)$. A neural network is used to optimize $q(x)$, and the FOM value is used as the loss function. This idea comes from physics informed neural network(PINN), where a neural network is used to represent the solution of differential equations. Here the neural network $q(x)$ should solve the functional equations that optimize FOM. For a 1d transmission problem with Gaussian absorption cross section, we observe a significant improvement of the FOM value compared to the constant $q$ case and the original Woodcock method. Generalizations of the neural network Woodcock(NNW) method to 3d voxel models are waiting to be explored.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。