In matrix form (state space form) you often define the system such that you try to adjust the system in a square form,in which the considered variable of interest might be a rate or a gradient(vectors and their components), where you can further break it down using several approaches with relatively lower complexity but ofc things such as how much of coupling is involved(between the variables) can then further complicate the task depending on how "non-linear" the interaction seems to be,but we can kind of approximate this non linearity around some points if some conditions are met.
brunton's online videos, lectures, learning material are enormously helpful , thank you.
This guy is legend ofcourse
Absolutely amazing video! Subscribe to his RUclips channel. It has a lot of great playlists.
❤
I have a question for the Lorentz system: why not include d^2x/dt^2 terms in machine learning??
They’re redundant. Any second order equation can be written as two first order equations by introducing a new variable y=dx/dt.
In matrix form (state space form) you often define the system such that you try to adjust the system in a square form,in which the considered variable of interest might be a rate or a gradient(vectors and their components), where you can further break it down using several approaches with relatively lower complexity but ofc things such as how much of coupling is involved(between the variables) can then further complicate the task depending on how "non-linear" the interaction seems to be,but we can kind of approximate this non linearity around some points if some conditions are met.
Isn’t this Steve Brunton teaching control theory?
we need to merge phycology ,philosophy ,neuroscience, biology and physics to make an A.G.I