Hey Kau. One hypothesis we have with Miguel about this is that having different encoding values…
Francisco Ingham

That is very plausible, thanks! I am a fluid dynamicist so it is normal (expected) for dynamical ‘features’ to have multiple time scales. And it does makes sense to me now that the embedding space, which is still pretty high dimensional, should have similar structure. Both of your posts and subsequent discussions have therefore helped greatly, so kudos!