arXiv subjects

Andrew Ning

Publications and source records attributed to Andrew Ning.

3 recordsLinked to original sources

arXiv2026

Accelerating the Simulation of Ordinary Differential Equations Through Physics-Preserving Neural Networks

Numerical simulation of ordinary differential equations (ODEs) can be challenging when the system exhibits high accelerations and rapidly changing dynamics. Under these conditions the ODE solver often needs to take very small time steps in order to resolve the solution accurately, resulting in increased computational cost. In order to accelerate the simulation of these ODEs we present a novel methodology that uses a pseudo-invertible neural network to map system states into a high-dimensional latent-space. The network is then trained so that the dynamics in this learned latent space are slow, and can be simulated with relatively few function calls. Unlike existing neural methods, the latent dynamic equations are not learned from trajectory data, but derived from the original system equations and the chain rule. This allows the method to generalize better than existing approaches because the derived equations are correct by construction. In this work, we derive latent state equations of motion for any general ODE, and describe the loss function used to enforce slow time evolution of the latent states. We then apply this technique to multiple example ODEs and show that these problems can be solved with $3$x to $20$x fewer function calls for the same accuracy when simulating in the learned latent space. This reduction in cost could decrease computational demands for scientific simulations across engineering and physics applications.

math.NA↗

arXiv2023

Automating Steady and Unsteady Adjoints: Efficiently Utilizing Implicit and Algorithmic Differentiation

Algorithmic differentiation (AD) has become increasingly capable and straightforward to use. However, AD is inefficient when applied directly to solvers, a feature of most engineering analyses. We can leverage implicit differentiation to define a general AD rule, making adjoints automatic. Furthermore, we can leverage the structure of differential equations to automate unsteady adjoints in a memory efficient way. We also derive a technique to speed up explicit differential equation solvers, which have no iterative solver to exploit. All of these techniques are demonstrated on problems of various sizes, showing order of magnitude speed-ups with minimal code changes. Thus, we can enable users to easily compute accurate derivatives across complex analyses with internal solvers, or in other words, automate adjoints using a combination of AD and implicit differentiation.

math.OC↗

arXiv2022

Reviving the Vortex Particle Method: A Stable Formulation for Meshless Large Eddy Simulation

The vortex particle method (VPM) is a mesh-free approach to computational fluid dynamics (CFD) solving the Navier-Stokes equations in their velocity-vorticity form. The VPM uses a Lagrangian scheme, which not only avoids the hurdles of mesh generation, but it also conserves vortical structures over long distances with minimal numerical dissipation while being orders of magnitude faster than conventional mesh-based CFD. However, VPM is known to be numerically unstable when vortical structures break down close to the turbulent regime. In this study, we reformulate the VPM as a large eddy simulation (LES) in a scheme that is numerically stable, without increasing its computational cost. A new set of VPM governing equations are derived from the LES-filtered Navier-Stokes equations. The new equations reinforce conservation of mass and angular momentum by reshaping the vortex elements subject to vortex stretching. In addition to the VPM reformulation, a new anisotropic dynamic model of subfilter-scale (SFS) vortex stretching is developed. This SFS model is well suited for turbulent flows with coherent vortical structures where the predominant cascade mechanism is vortex stretching. Advection, viscous diffusion, and vortex stretching are validated through simulation of isolated and leapfrogging vortex rings. Mean and fluctuating components of turbulent flow are validated through simulation of a turbulent round jet, where Reynolds stresses are resolved directly and compared to experimental measurements. Finally, the computational efficiency of the scheme is showcased in the simulation of an aircraft rotor in hover, showing our meshless LES to be 100x faster than a mesh-based LES with similar fidelity, while being 10x faster than a low-fidelity unsteady Reynolds-average Navier-Stokes simulation and 1000x faster than a high-fidelity detached-eddy simulation.

physics.flu-dyn↗