Question about the external force term in LBM physics guidance

[WanqiYuan]

Hi！

Thanks for the excellent work!

I have a question regarding the physics guidance with LBM. My current understanding is that, for a sampled time step (T) during one optimization iteration, (F_w(T)) provides a wind force field defined on the grid at that time step. Then, LBM performs an iterative rollout from the initial state t=0 to t = T, and the resulting physical quantities are used to compute the physics loss. Please correct me if this interpretation is wrong.

I noticed that the LBM formulation includes an external force term, (F_oi). Since the wind field changes continuously over both space and time, I would expect this force term to be spatially and temporally varying rather than a constant global force.

Could you clarify how this external force term is handled in your system? For example, is it directly derived from the optimized wind field (F_w), predicted separately, or simplified in some way during the LBM rollout? My concern is that this F_oi is a very complec force which needs a proper define and without a proper external force term, the LBM rollout may not be well-defined, while setting it as a constant force may not accurately represent the dynamic wind field.

Thanks!

[huahuo359]

Hi,

Thanks for your question.

One clarification is that (F_{oi}) denotes an external force applied to the wind field, rather than the force from the object to the wind. In our experiments, this term is set to zero. Therefore, it is neither predicted separately nor derived from the optimized wind field (F_w).

We include (F_{oi}) only as part of the general LBM formulation. More details of the LBM solver can be found in High-Order Moment-Encoded Kinetic Simulation of Turbulent Flows. (https://dl.acm.org/doi/10.1145/3618341)