发表机构
Trinity University; Tulane University; Centre College（三一大学; 杜兰大学; 中心学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对细菌水动力学模拟中近场精度与计算效率的矛盾，提出基于正则化Stokeslet的降阶力模型，通过约束最小二乘校准和主成分分析减少力点，在自由空间和平面边界保持主导流动结构，降低计算成本。
AI 中文摘要
准确解析许多游动细菌产生的近场流动，同时保持计算效率，仍然是一个挑战。基于Stokeslet的模型使用多个力点可以捕捉详细的近场水动力学，但计算成本高昂。其他方法包括远场和低阶模型，它们使用较少的点但牺牲了近场精度。我们引入了一个基于正则化Stokeslet方法的降阶框架，该框架在显著减少力点数量的同时保留了重要的近场流动特征。该方法将高保真模型的力分布替换为稀疏点集上的力，这些力的强度通过约束最小二乘校准到单个细菌的高保真速度场来确定。校准后的力产生的近似保留了主导的参考流动结构，同时满足自推进游动的无力和无扭矩条件。然后使用主成分分析仅用几个主导模态来表示校准力的相位相关变化。通过将相应的模态系数拟合为鞭毛相位的连续函数，可以在鞭毛周期的任何相位近似降阶力。该框架在自由空间和无滑移平面边界附近均得到应用，保留了主导流动结构并捕捉了壁面引起的周围流体重定向。优化的细胞体正则化参数对壁面距离的依赖性较弱，剩余的速度差异主要集中在细胞体附近。通过大幅减少力点数量，降阶模型降低了速度场评估的计算成本和内存需求，使得多游泳者流场的大域模拟更加实用。
英文摘要
Accurately resolving the near-field flow generated by many swimming bacteria while retaining computational efficiency remains challenging. Stokeslet-based models with many force points can capture detailed near-field hydrodynamics but are computationally expensive. Other approaches include far-field and low-order models that use fewer points but sacrifice near-field accuracy. We introduce a reduced-order framework based on the method of regularized Stokeslets that preserves important near-field flow features while using substantially fewer force points. The method replaces the force distribution of a high-fidelity model with forces on a sparse set of points whose strengths are determined by constrained least-squares calibration to the high-fidelity velocity field of a single bacterium. The calibrated forces yield an approximation that preserves the dominant reference-flow structure while satisfying the force-free and torque-free conditions of self-propelled swimming. Principal component analysis is then used to represent the phase-dependent variation of the calibrated forces using only a few dominant modes. By fitting the corresponding modal coefficients as continuous functions of flagellar phase, the reduced-order forces can be approximated at any phase of the flagellar cycle. The framework is applied in both free space and near a no-slip planar boundary, where it preserves the dominant flow structures and captures the wall-induced redirection of the surrounding fluid. The optimized cell-body regularization parameter depends weakly on wall distance, and the remaining velocity discrepancy is concentrated primarily near the cell body. By substantially reducing the number of force points, the reduced-order model lowers both the computational cost and memory requirements of velocity-field evaluation, making large-domain simulations of multi-swimmer flow fields more practical.
