Koopman Observable Subspaces & Finite Linear Representations of Nonlinear Dynamics for Control

แชร์
ฝัง
  • เผยแพร่เมื่อ 3 ม.ค. 2025

ความคิดเห็น • 16

  • @kutsjuice
    @kutsjuice 2 ปีที่แล้ว +8

    Tons of thanks for the very great lectures!
    It is very impressive and helpful!

    • @Eigensteve
      @Eigensteve  2 ปีที่แล้ว

      Awesome, glad you like them!

  • @lukistrela
    @lukistrela 3 ปีที่แล้ว

    Dear Steve! Great playlist on Koopman analysis. A little difficult to find the right order to watch videos. Thanks!

  • @AmanThakur-v8u
    @AmanThakur-v8u ปีที่แล้ว

    Thank you for good videos on Koopman Operators can you suggest me some research paper to study more about truncation of koopman observables.Where I can find more material on this one.

  • @johnfinn9495
    @johnfinn9495 2 ปีที่แล้ว +1

    Steve, Great series. I have a question. All this Koopman technology is about taking a nonlinear system and transforming it into a linear equation of higher order. But this is exactly what you can do to a Riccati equation, and Riccati equations occur everywhere throughout control theory. Are the results of a Riccati transformation a special case of a Koopman invariant subspace?
    PS Met and talked with Nathan Kutz a few months ago in Santa Fe. Send my regards. -- John Finn

    • @Eigensteve
      @Eigensteve  2 ปีที่แล้ว

      That is a great question! I just recently heard about this possible connection from a collaborator, so we started looking into it a little more. Short answer is I'm not sure, but definitely a good lead to find a connection.
      That is great that you met Nathan -- I'll say hi!

    • @johnfinn9495
      @johnfinn9495 2 ปีที่แล้ว +1

      @@Eigensteve Steve, I came to Koopman methods from the more standard adjoint methods for various applications. I have agreed to give a seminar (or two!) on Koopman methods at WSU in October, and I may need to talk with you or Nathan about some issues. Would you be available?
      Also, it must be the case that Riccati equations (including matrix Riccati) must be a \emph{very} special case of NL equations that can be transformed to a finite dimensional linear system. (E.g. cannot have limit cycles, chaotic attractor, etc.) I'm not sure of the relation between Riccati equations and your example with \dot{x_2}=\mu(x_2-x_1^2).
      - John

    • @Eigensteve
      @Eigensteve  2 ปีที่แล้ว

      @@johnfinn9495 Sounds very interesting, and would be available to chat sometime. Next month is crazy, but maybe after that.

  • @hkfnchsmh
    @hkfnchsmh 6 หลายเดือนก่อน

    thanks for this education. one tip though: the red graphs are very hard to see on the black background.

  • @theshank4936
    @theshank4936 3 ปีที่แล้ว

    Is the optimal control, a local optimum or a global optimum? In a linear system, LQR controller gives the global optimum, but is that the same for a koopan linearized system?

  • @goodgame7474
    @goodgame7474 2 ปีที่แล้ว

    Can we apply koopman operator theory to systems with inputs ?

  • @paulisaac3489
    @paulisaac3489 4 ปีที่แล้ว +8

    You have math I've never heard of before, love you in a totally non gay way. KNOWLEDGE!

  • @chaosun689
    @chaosun689 ปีที่แล้ว

    cool, you can write reversely??!

  • @Ericlau1126
    @Ericlau1126 4 ปีที่แล้ว +1

    reversed writing skill. Crazy!

    • @nndei
      @nndei 3 ปีที่แล้ว +2

      I don't know if he does that, actually. He could be left handed and writing normally on a transparent board, then he just flips the video.

    • @navsquid32
      @navsquid32 3 ปีที่แล้ว

      Wat