我们提出了一种求解流体力学反问题的高效计算方法。我们研究的反问题以可纳入数据项与正则化项的确定性损失函数形式构建。我们引入了一种多重网格分解技术,用于加速参数定义在网格上的优化问题的梯度类方法收敛速度。我们将该多重网格技术整合到ODIL(Optimizing a DIscrete Loss,优化离散损失)框架中。多分辨率ODIL(mODIL)将原有方法的收敛速度提升了一个数量级,且能更好地避免陷入局部极小值。此外,mODIL支持使用自动微分计算损失函数的梯度,从而降低了框架的实现难度。我们在多种反问题与流场重构问题上验证了mODIL的性能:包括Burgers方程的解重构、由温度测量值反推电导率,以及三维情况下由尾流速度测量值反推物体形状。我们还将该方法与当前流行的相关方法Physics-Informed Neural Networks(PINNs,物理信息神经网络)进行了对比研究。结果表明,在包含简单偏微分方程、顶盖驱动方腔问题的基准测试中,mODIL的计算成本比PINNs低3到5个数量级。我们的研究结果表明,mODIL是求解流体力学反问题的一种高效、快速且性能稳定的方法。
英文摘要:
We present a potent computational method for the solution of inverse problems in fluid mechanics. We consider inverse problems formulated in terms of a deterministic loss function that can accommodate data and regularization terms. We introduce a multigrid decomposition technique that accelerates the convergence of gradient-based methods for optimization problems with parameters on a grid. We incorporate this multigrid technique to the ODIL (Optimizing a DIscrete Loss) framework. The multiresolution ODIL (mODIL) accelerates by an order of magnitude the original formalism and improves the avoidance of local minima. Moreover, mODIL accommodates the use of automatic differentiation for calculating the gradients of the loss function, thus facilitating the implementation of the framework. We demonstrate the capabilities of mODIL on a variety of inverse and flow reconstruction problems: solution reconstruction for the Burgers equation, inferring conductivity from temperature measurements, and inferring the body shape from wake velocity measurements in three dimensions. We also provide a comparative study with the related, popular Physics-Informed Neural Networks (PINNs) method. We demonstrate that mODIL has three to five orders of magnitude lower computational cost than PINNs in benchmark problems including simple PDEs and lid-driven cavity problems. Our results suggest that mODIL is a very potent, fast and consistent method for solving inverse problems in fluid mechanics.
Automatic differentiation approach for reconstructing spectral functions with neural networks
Frankfurt Institute for Advanced Studies(法兰克福高等研究院);Xidian-FIAS International Joint Research Center(西电-法兰克福高等研究院国际联合研究中心);McGill University(麦吉尔大学);Stony Brook University(石溪大学)
FANTASY: User-Friendly Symplectic Geodesic Integrator for Arbitrary Metrics with Automatic Differentiation
我们提出了FANTASY(Finally A Numerical Trajectory Algorithm both Straightforward and sYmplectic),这是一个用Python编写的用户友好的开源辛测地线积分器。FANTASY被设计为“开箱即用”,除了度量和测地线的初始条件外,不需要用户提供任何东西。FANTASY使用自动微分有效地计算高达机器精度的导数,允许在任意空间(时间)中积分测地线,而无需用户手动输入Christoffel符号或任何其他度量导数。此外,FANTASY利用了一种将相空间加倍的哈密顿积分方案,其中粒子相空间的两个副本一起演化。这种技术允许一个集成方案,这是明确的和辛的,即使当哈密尔顿是不可分的。FANTASY预建有二阶和四阶格式,并且很容易扩展到高阶格式。FANTASY还包括一个自动雅可比计算器,允许自动完成坐标变换。
TaylUR, an arbitrary-order diagonal automatic differentiation package for Fortran 95
我们提出了TaylUR,Fortran 95模块自动计算数值的复值函数的衍生物w.r.t.多个变量,每个变量的顺序任意,但不包括混合衍生物。算术运算符和Fortran内部函数被重载,以正确地作用于定义类型为“taylor”的对象,该类型将函数及其前几个导数w.r.t.用户自定义的独立变量。乘积和复合函数的导数使用莱布尼茨规则和Faa di Bruno公式计算。TaylUR大量使用运算符重载和其他面向对象的Fortran 95特性。