【摘要】 计算机模拟可用于可视化并计算分子相互作用体积,从而辅助催化剂的高效设计。但现有交互体积计算方法体素遍历速度慢,体积计算精度低。针对这些问题,本文对现有方法进行改进以实现高效准确的体素遍历和交互体积计算。首先,使用改进的行进立方体算法迭代遍历获取所有边界体素并进行线性插值,生成分子等值面;其次,对等值面进行切片并将每层切割点分为两组,使用包围盒算法和射线法识别每层切片的相交区域;最后,累加相交区域面积与切片层高的乘积获取相互作用体积。实验结果表明,在三组分子等值面点云数据样本中,改进后方法均能提高体素遍历速度与体积计算精度。其中,体素平均遍历速度提高4.7倍,交互体积的平均相对误差为0.19%,低于现有方法的1.73%。更多还原

【Abstract】 Computer simulation can visualize and compute molecular interacted volumes,aiding the efficient design of catalysts.However,existing method for interacted volume calculation suffers from slow voxel traversal and low accuracy in volume computation.To address these issues,this work improves existing method to achieve efficient and accurate voxel traversal and interacted volume computation.First,an improved marching cubes algorithm is applied to iteratively traverse boundary voxels and perform linear interpolation to generate molecular isosurfaces.Then,the isosurfaces are sliced,with intersection points at each layer grouped into two sets,and intersected regions in each slice are identified using bounding-box and ray-casting algorithms.Finally,the interacted volume is obtained by summing the products of intersected areas and slice thickness across slices.Experimental results on three molecular isosurface point cloud datasets show that the improved method enhances voxel traversal speed and calculation accuracy,achieving a 4.7-fold speedup in voxel traversal and an average relative error of 0.19%in interacted volumes,compared with 1.73%for the existing method.更多还原