sparse jurdjevic quinn stabilization
play

Sparse Jurdjevic-Quinn stabilization Francesco Rossi - Universit - PowerPoint PPT Presentation

Oct 30, 2023 •279 likes •1.4k views

Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Sparse Jurdjevic-Quinn stabilization Francesco Rossi - Universit dAix-Marseille, France http://www.LSIS.org/frossi/

  1. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations The simplest example on R 2 � ˙ � 1 � 0 � � � x = u 1 g 1 + u 2 g 2 = u 1 + u 2 y ˙ 0 1 Exercise One can stabilize the system to (0 , 0) by choosing the functional V = x 2 + y 2 . f = 0 , hence L f V = 0 ; L g 1 V = L g 2 V = 0 in (0 , 0) only; Jurdjevic-Quinn gives u ( x, y ) = ( − 2 x, − 2 y ) , What happens when adding the component-wise sparsity constraint? 9 / 40

  2. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Hybrid systems to prevent chattering The component-wise sparsity constraint itself produces switching rules, that naturally introduce chattering phenomena. We present three possible solutions: Strategies 1 and 2 are based on time sampling; Strategy 3 is based on hysteresis. 10 / 40

  3. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Hybrid systems to prevent chattering The component-wise sparsity constraint itself produces switching rules, that naturally introduce chattering phenomena. We present three possible solutions: Strategies 1 and 2 are based on time sampling; Strategy 3 is based on hysteresis. 10 / 40

  4. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations 11 / 40

  5. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations 11 / 40

  6. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Fix the sampling time τ > 0 . For the initial state x 0 ∈ R n , consider the trajectory x ( t ) of (S) with the time-varying feedback control u ( t, x ) defined as follows: on each time interval [( km + i ) τ, ( km + i + 1) τ ) , apply the feedback control u ( t, x ) = − L g i V ( x ( t )) e i . Remark This is a kind of averaged control. 12 / 40

  7. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Fix the sampling time τ > 0 . For the initial state x 0 ∈ R n , consider the trajectory x ( t ) of (S) with the time-varying feedback control u ( t, x ) defined as follows: on each time interval [( km + i ) τ, ( km + i + 1) τ ) , apply the feedback control u ( t, x ) = − L g i V ( x ( t )) e i . Remark This is a kind of averaged control. 12 / 40

  8. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Theorem (Caponigro-Piccoli-Rossi-Trélat, submitted) Let Z = { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } , and Ω be the largest subset of Z invariant for ˙ x = f ( x ) . If Ω is locally attractive , then, for each initial state x 0 ∈ R n , there exist τ 1 > 0 such that, for every τ ∈ (0 , τ 1 ) , Strategy 1 with sampling time τ stabilizes the control system (S) to Ω . Remark Local attractiveness is necessary, see counterexamples. 13 / 40

  9. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Theorem (Caponigro-Piccoli-Rossi-Trélat, submitted) Let Z = { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } , and Ω be the largest subset of Z invariant for ˙ x = f ( x ) . If Ω is locally attractive , then, for each initial state x 0 ∈ R n , there exist τ 1 > 0 such that, for every τ ∈ (0 , τ 1 ) , Strategy 1 with sampling time τ stabilizes the control system (S) to Ω . Remark Local attractiveness is necessary, see counterexamples. 13 / 40

  10. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback τ = 1 14 / 40

  11. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback τ = 0 . 1 14 / 40

  12. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback τ = 0 . 01 14 / 40

  13. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Strategy 1 is far from being “optimal” in some sense; 14 / 40

  14. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1: Sparse time-varying feedback Strategy 1 is far from being “optimal” in some sense; It is useful on long-range, not locally. 14 / 40

  15. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations 15 / 40

  16. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations 15 / 40

  17. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sampled sparse feedback Fix a sampling time τ > 0 . In x 0 , choose the smallest integer i ∈ { 1 , . . . , m } such that | L g i V ( x 0 ) | ≥ | L g j V ( x 0 ) | , ∀ j � = i. Then consider the sampling solution associated with u and the sampling time τ until x 1 = x ( τ ) . Compute the new control in x 1 , and interatively for each x n . Remark The control is chosen as a “steepest descent” for V , then it is kept for the whole time interval [0 , τ ] . 16 / 40

  18. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sampled sparse feedback Fix a sampling time τ > 0 . In x 0 , choose the smallest integer i ∈ { 1 , . . . , m } such that | L g i V ( x 0 ) | ≥ | L g j V ( x 0 ) | , ∀ j � = i. Then consider the sampling solution associated with u and the sampling time τ until x 1 = x ( τ ) . Compute the new control in x 1 , and interatively for each x n . Remark The control is chosen as a “steepest descent” for V , then it is kept for the whole time interval [0 , τ ] . 16 / 40

  19. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sampled sparse feedback Theorem (CPRT) If { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } = { ¯ x } x ∈ R n , then, for every initial state x 0 ∈ R n , there exists for some ¯ τ 2 > 0 such that, for every τ ∈ (0 , τ 2 ) , Strategy 2 with sampling time τ practically asymptotically stabilizes (S) to ¯ x . Remark The functional V ( t ) is not strictly decreasing, one can have problems for Z not reduced to a point. 17 / 40

  20. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sampled sparse feedback Theorem (CPRT) If { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } = { ¯ x } x ∈ R n , then, for every initial state x 0 ∈ R n , there exists for some ¯ τ 2 > 0 such that, for every τ ∈ (0 , τ 2 ) , Strategy 2 with sampling time τ practically asymptotically stabilizes (S) to ¯ x . Remark The functional V ( t ) is not strictly decreasing, one can have problems for Z not reduced to a point. 17 / 40

  21. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sparse time-varying feedback τ = 1 18 / 40

  22. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sparse time-varying feedback τ = 0 . 5 18 / 40

  23. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sparse time-varying feedback τ = 0 . 01 18 / 40

  24. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2: Sparse time-varying feedback Strategy 2 is efficient along axes. It is not efficient around the switching set. 18 / 40

  25. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Fix ε ∈ (0 , 1) and apply the following algorithm: Initialize: n = 0 and t 0 = 0 . While t n < + ∞ apply Step n : At time t n choose i = 1 , . . . , m the smallest integer such that | L g i V ( x ( t n )) | ≥ | L g j V ( x ( t n )) | , for every j � = i. If | L g i V ( x ( t n )) | ≥ 2 t − 1 n , keep the control u = − L g i V ( x ) until | L g i V ( x ( t )) | ≤ t − 1 or | L g i V x ( t )) | ≤ (1 − ε ) | L g j V ( x ( t )) | for some j = 1 , . . . , m. If | L g i V ( x ( t n )) | < 2 t − 1 n , keep the control u = 0 until | L g j V ( x ( t )) | ≥ 4 t − 1 , for some j = 1 , . . . , m. Remark The control is chosen as a “steepest descent” for V , then continuously evaluated along time. 19 / 40

  26. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Fix ε ∈ (0 , 1) and apply the following algorithm: Initialize: n = 0 and t 0 = 0 . While t n < + ∞ apply Step n : At time t n choose i = 1 , . . . , m the smallest integer such that | L g i V ( x ( t n )) | ≥ | L g j V ( x ( t n )) | , for every j � = i. If | L g i V ( x ( t n )) | ≥ 2 t − 1 n , keep the control u = − L g i V ( x ) until | L g i V ( x ( t )) | ≤ t − 1 or | L g i V x ( t )) | ≤ (1 − ε ) | L g j V ( x ( t )) | for some j = 1 , . . . , m. If | L g i V ( x ( t n )) | < 2 t − 1 n , keep the control u = 0 until | L g j V ( x ( t )) | ≥ 4 t − 1 , for some j = 1 , . . . , m. Remark The control is chosen as a “steepest descent” for V , then continuously evaluated along time. 19 / 40

  27. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Fix ε ∈ (0 , 1) and apply the following algorithm: Initialize: n = 0 and t 0 = 0 . While t n < + ∞ apply Step n : At time t n choose i = 1 , . . . , m the smallest integer such that | L g i V ( x ( t n )) | ≥ | L g j V ( x ( t n )) | , for every j � = i. If | L g i V ( x ( t n )) | ≥ 2 t − 1 n , keep the control u = − L g i V ( x ) until | L g i V ( x ( t )) | ≤ t − 1 or | L g i V x ( t )) | ≤ (1 − ε ) | L g j V ( x ( t )) | for some j = 1 , . . . , m. If | L g i V ( x ( t n )) | < 2 t − 1 n , keep the control u = 0 until | L g j V ( x ( t )) | ≥ 4 t − 1 , for some j = 1 , . . . , m. Remark The control is chosen as a “steepest descent” for V , then continuously evaluated along time. 19 / 40

  28. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Fix ε ∈ (0 , 1) and apply the following algorithm: Initialize: n = 0 and t 0 = 0 . While t n < + ∞ apply Step n : At time t n choose i = 1 , . . . , m the smallest integer such that | L g i V ( x ( t n )) | ≥ | L g j V ( x ( t n )) | , for every j � = i. If | L g i V ( x ( t n )) | ≥ 2 t − 1 n , keep the control u = − L g i V ( x ) until | L g i V ( x ( t )) | ≤ t − 1 or | L g i V x ( t )) | ≤ (1 − ε ) | L g j V ( x ( t )) | for some j = 1 , . . . , m. If | L g i V ( x ( t n )) | < 2 t − 1 n , keep the control u = 0 until | L g j V ( x ( t )) | ≥ 4 t − 1 , for some j = 1 , . . . , m. Remark The control is chosen as a “steepest descent” for V , then continuously evaluated along time. 19 / 40

  29. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Theorem (CPRT) For any ε ∈ (0 , 1) , Strategy 3 asymptotically stabilizes (S) to Ω , the largest invariant subset of { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } . Remark The strategy is more robust, since it works for any ε . One needs a continuous evaluation of the control. 20 / 40

  30. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Theorem (CPRT) For any ε ∈ (0 , 1) , Strategy 3 asymptotically stabilizes (S) to Ω , the largest invariant subset of { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } . Remark The strategy is more robust, since it works for any ε . One needs a continuous evaluation of the control. 20 / 40

  31. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Theorem (CPRT) For any ε ∈ (0 , 1) , Strategy 3 asymptotically stabilizes (S) to Ω , the largest invariant subset of { x ∈ R n | L f V ( x ) = L g i V ( x ) = 0 , for every i = 1 , . . . , m } . Remark The strategy is more robust, since it works for any ε . One needs a continuous evaluation of the control. 20 / 40

  32. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis ε = 0 . 5 21 / 40

  33. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis ε = 0 . 1 21 / 40

  34. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis ε = 0 . 01 21 / 40

  35. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3: Sparse feedback with hysteresis Similarly to Strategy 2, Strategy 3 is efficient along axes. It is not efficient around the switching set. 21 / 40

  36. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Comparison between strategies 8 6 4 2 0 0 2 4 6 8 Orange: “Time to target=1” for Strategy 1 for τ → 0 . Red: Strategy 2 for τ → 0 and Strategy 3 for ε → 0 . Green: Non-sparse Jurdjevic-Quinn. 22 / 40

  37. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Comparison between strategies 8 6 4 2 0 0 2 4 6 8 Orange: “Time to target=1” for Strategy 1 for τ → 0 . Red: Strategy 2 for τ → 0 and Strategy 3 for ε → 0 . Green: Non-sparse Jurdjevic-Quinn. 22 / 40

  38. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Sparse control of the Hegselmann-Krause model Each agent has an opinion x i ( t ) , that evolves by interaction with his neighbours only: � x i = ˙ ( x j − x i ) | x j − x i | < 1 23 / 40

  39. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Sparse control of the Hegselmann-Krause model Each agent has an opinion x i ( t ) , that evolves by interaction with his neighbours only: � x i = ˙ ( x j − x i ) + u i . | x j − x i | < 1 23 / 40

  40. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Sparse control of the Hegselmann-Krause model Each agent has an opinion x i ( t ) , that evolves by interaction with his neighbours only: � x i = ˙ ( x j − x i ) + u i . | x j − x i | < 1 We aim to enforce consensus by minimizing the variance � ( x i − x j ) 2 . V = i,j 23 / 40

  41. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Sparse control of the Hegselmann-Krause model Each agent has an opinion x i ( t ) , that evolves by interaction with his neighbours only: � x i = ˙ ( x j − x i ) + u i . | x j − x i | < 1 We aim to enforce consensus by minimizing the variance � ( x i − x j ) 2 . V = i,j L f V ≤ 0 , thus we can apply Jurdjevic-Quinn. 23 / 40

  42. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Simulation with 50 randomized agents in [0 , 10] 24 / 40

  43. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1 for Hegselmann-Krause model τ = 1 25 / 40

  44. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1 for Hegselmann-Krause model τ = 0 . 5 25 / 40

  45. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 1 for Hegselmann-Krause model τ = 0 . 1 25 / 40

  46. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2 for Hegselmann-Krause model τ = 1 26 / 40

  47. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2 for Hegselmann-Krause model τ = 0 . 5 26 / 40

  48. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 2 for Hegselmann-Krause model τ = 0 . 1 26 / 40

  49. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3 for Hegselmann-Krause model ε = 0 . 2 27 / 40

  50. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3 for Hegselmann-Krause model ε = 0 . 05 27 / 40

  51. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Strategy 3 for Hegselmann-Krause model ε = 0 . 01 27 / 40

  52. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Section 2 Sparse Jurdjevic-Quinn stabilization for mean-field equations 28 / 40

  53. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Mean-Field equations Let us consider a large number N of interacting agents: Analysis of a very large-dimensional ODE is hard; Numerical simulations are hard; Control is hard. If agents are all identical, we can replace the positions of agents with the density of the crowd µ . What is the dynamics of µ ? 29 / 40

  54. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Mean-Field equations Let us consider a large number N of interacting agents: Analysis of a very large-dimensional ODE is hard; Numerical simulations are hard; Control is hard. If agents are all identical, we can replace the positions of agents with the density of the crowd µ . What is the dynamics of µ ? 29 / 40

  55. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Mean-Field equations Let us consider a large number N of interacting agents: Analysis of a very large-dimensional ODE is hard; Numerical simulations are hard; Control is hard. If agents are all identical, we can replace the positions of agents with the density of the crowd µ . What is the dynamics of µ ? 29 / 40

  56. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Proposition If f is regular, the mean-field limit of x i = 1 � ˙ f ( x j − x i ) (ODE) N j � = i is ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 . (PDE) where F [ µ ] = f ⋆ µ. Conditions of existence, uniqueness, continuous dependence...? 30 / 40

  57. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Proposition If f is regular, the mean-field limit of x i = 1 � ˙ f ( x j − x i ) (ODE) N j � = i is ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 . (PDE) where F [ µ ] = f ⋆ µ. Conditions of existence, uniqueness, continuous dependence...? 30 / 40

  58. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Proposition Assume that F is defined by F [ µ ] ( x ) := v ( f ⋆ µ ( x )) with v, f Lipschitz. Then there exists a unique solution of the Cauchy problem � ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 µ | t =0 = µ 0 , Remark This is a particular case of a theory in Wasserstein spaces: Ambrosio-Gangbo, Colombo-Garavello-Mercier, Piccoli-Rossi... Proposition (Piccoli-Tosin 2011, Piccoli-Rossi 2013) Numerical schemes are available. 31 / 40

  59. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles x i = 1 ˙ � j f ( x i − x j ) N 32 / 40

  60. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles x i = 1 ˙ � j f ( x i − x j ) N Control of the density 32 / 40

  61. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles Evolution of the density mean-field − − − − − − → x i = 1 ˙ � j f ( x i − x j ) ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 N Control of the density 32 / 40

  62. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles Evolution of the density mean-field − − − − − − → x i = 1 ˙ � j f ( x i − x j ) ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 N ↓ control? Control of the density 32 / 40

  63. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles Evolution of the density mean-field − − − − − − → x i = 1 ˙ � j f ( x i − x j ) ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 N control ↓ ↓ control? Controlled ODE Control of the density x i = 1 � ˙ j f ( x i − x j )+ u i ( t ) N 32 / 40

  64. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation N interacting particles Evolution of the density mean-field − − − − − − → x i = 1 ˙ � j f ( x i − x j ) ∂ t µ + ∇ · ( F [ µ ] µ ) = 0 N control ↓ ↓ control? Controlled ODE mean-field? − − − − − − → Control of the density x i = 1 � ˙ j f ( x i − x j )+ u i ( t ) N 32 / 40

  65. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation How to control the transport equation? How to define controls? Remark Controls on each agent � x i = � ˙ j f ( x j − x i )+ u i ( t ) make no sense in the mean-field limit. Definition The control is a localized vector field χ ω u . It acts as follows ∂ t µ + ∇ · [( F [ µ ] + χ ω u ) µ ] = 0 . We need χ ω u Lipschitz. 33 / 40

  66. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Control of the transport equation How to control the transport equation? How to define controls? Remark Controls on each agent � x i = � ˙ j f ( x j − x i )+ u i ( t ) make no sense in the mean-field limit. Definition The control is a localized vector field χ ω u . It acts as follows ∂ t µ + ∇ · [( F [ µ ] + χ ω u ) µ ] = 0 . We need χ ω u Lipschitz. 33 / 40

  67. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Definition (Sparsity constraint) Fix a small area c > 0 of the configuration space and impose the control set ω ( t ) satisfies | ω ( t ) | ≤ c ; the control function u ( t ) is defined on ω ( t ) with � u ( t ) � ∞ ≤ 1 We aim to generalize Strategy 3 to this setting. Remark One control only, for more controls adapt finite-dimensional. The hardest part is to understand how to generalize objets for the Jurdjevic-Quinn method: proper Lyapunov function, Lie derivatives, steepest descent... 34 / 40

  68. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Definition (Sparsity constraint) Fix a small area c > 0 of the configuration space and impose the control set ω ( t ) satisfies | ω ( t ) | ≤ c ; the control function u ( t ) is defined on ω ( t ) with � u ( t ) � ∞ ≤ 1 We aim to generalize Strategy 3 to this setting. Remark One control only, for more controls adapt finite-dimensional. The hardest part is to understand how to generalize objets for the Jurdjevic-Quinn method: proper Lyapunov function, Lie derivatives, steepest descent... 34 / 40

  69. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations Definition (Sparsity constraint) Fix a small area c > 0 of the configuration space and impose the control set ω ( t ) satisfies | ω ( t ) | ≤ c ; the control function u ( t ) is defined on ω ( t ) with � u ( t ) � ∞ ≤ 1 We aim to generalize Strategy 3 to this setting. Remark One control only, for more controls adapt finite-dimensional. The hardest part is to understand how to generalize objets for the Jurdjevic-Quinn method: proper Lyapunov function, Lie derivatives, steepest descent... 34 / 40

  70. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations The idea is to have Lipschitz-mollified characteristic functions χ η [ a,b ] a − η a b b + η ( a, b, η ) | | ω ( a, b, η ) | ≤ c and η ≥ t − 1 � � Ω t = . At time t n , choose one of the maximizers ( a ∗ , b ∗ , η ∗ ) in Ω t n of V [ e τχ η � [ a,b ] g µ ( t n )] − V [ µ ( t n )] � � � s t n ( a, b, η ) := | L χ η [ a,b ] g V [ µ ( t n )] | = � lim � � � τ � τ → 0 � 35 / 40

  71. Sparse Jurdjevic-Quinn stabilization in finite dimension Sparse Jurdjevic-Quinn stabilization for mean-field equations If either s t n ( a ∗ , b ∗ , η ∗ ) < t − 1 or Ω t n is empty, then choose the n zero control. a, ¯ η ) ∈ Ω ′ Let the measure evolve up to the existence of (¯ b, ¯ t a, ¯ η ) ≥ 2 t − 1 . such that s t (¯ b, ¯ If s t n ( a ∗ , b ∗ , η ∗ ) ≥ t − 1 n , then choose the control = − χ η ∗ u ( t, · ) [ a ∗ ,b ∗ ] sign( L χ η ∗ [ a ∗ ,b ∗ ] g [ µ ( t )] V [ µ ( t )]) . Let the measure evolve up to one of the following conditions: either s t ( a ∗ , b ∗ , η ∗ ) ≤ t − 1 2 ; a, ¯ or there exists (¯ b, ¯ η ) ∈ Ω ′ t such that a, ¯ s t ( a ∗ , b ∗ , η ∗ ) ≤ (1 − ε ) s t (¯ b, ¯ η ) . Theorem (CPRT, submitted) For each ε ∈ (0 , 1) , this strategy drives the solution to (PDE) to Z = { µ ∈ P c ( R n ) | L f V [ µ ] = L ug V [ µ ] = 0 ∀ u ∈ Lip( R n , R ) } 36 / 40

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Technology and Technology and Stabilization Stabilization Workshop on GHG Stabilization

Technology and Technology and Stabilization Stabilization Workshop on GHG Stabilization

Technology and Technology and Stabilization Stabilization Workshop on GHG Stabilization Scenarios Jae Edmonds 22 January 2004 National Institutes for Environmental Studies Tsukuba, Japan Thanks To Thanks To Thanks To National Institutes

459 views • 41 slides
Session A4 Climate &amp; Urbanism Francisca Quinn, Quinn &amp; Partners Inc. May 10, 2016

Session A4 Climate &amp; Urbanism Francisca Quinn, Quinn &amp; Partners Inc. May 10, 2016

Session A4 Climate &amp; Urbanism Francisca Quinn, Quinn &amp; Partners Inc. May 10, 2016 TRUSTED ADVISOR IN ALL MATTERS RELATED TO CORPORATE SUSTAINABILITY Quinn &amp; Partners is a sustainability strategy and integration advisory

480 views • 9 slides
PostgreSQL Adores a Vacuum Quinn Weaver &lt;quinn@pgexperts.com&gt;

PostgreSQL Adores a Vacuum Quinn Weaver &lt;quinn@pgexperts.com&gt;

PostgreSQL Adores a Vacuum Quinn Weaver &lt;quinn@pgexperts.com&gt; http://fairpath.com/vacuum-slides.pdf This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit

901 views • 54 slides
Sparse Matrices Example Of Sparse Matrices diagonal tridiagonal sparse many elements are

Sparse Matrices Example Of Sparse Matrices diagonal tridiagonal sparse many elements are

Sparse Matrices Example Of Sparse Matrices diagonal tridiagonal sparse many elements are zero lower triangular (?) dense few elements are zero These are structured sparse matrices. May be mapped into a 1D array so that a mapping

391 views • 6 slides
Sparse Matrices sparse many elements are zero dense few elements are zero Example Of

Sparse Matrices sparse many elements are zero dense few elements are zero Example Of

Sparse Matrices sparse many elements are zero dense few elements are zero Example Of Sparse Matrices diagonal tridiagonal lower triangular (?) These are structured sparse matrices. May be mapped into a 1D array so that a mapping

698 views • 11 slides
+ + + Stabilization Growth Moving = Forward 4 www.budget.illinois.gov I LLINOIS M OVING F

+ + + Stabilization Growth Moving = Forward 4 www.budget.illinois.gov I LLINOIS M OVING F

I LLINOIS M OVING F ORWARD I LLINOIS M OVING F ORWARD Fiscal Year 2013 Budget Fiscal Year 2013 Budget Governor Pat Quinn Governor Pat Quinn M OVING F ORWARD Stabilizing Our Budget Building &amp; Growing Our Economy 1 I LLINOIS M OVING F

641 views • 26 slides
Rent Stabilization in Mountain View Community Stabilization and Fair Rent Act (CSFRA) Measure

Rent Stabilization in Mountain View Community Stabilization and Fair Rent Act (CSFRA) Measure

Rent Stabilization in Mountain View Community Stabilization and Fair Rent Act (CSFRA) Measure V mountainview.gov/rentstabilization What is Rent Stabilization? A local program designed to limit the amount of rent increases a landlord may

486 views • 37 slides
Page 1 of 19 Rent Stabilization Board RE RENT STABILIZATION BOARD DATE: April 21, 2014 TO:

Page 1 of 19 Rent Stabilization Board RE RENT STABILIZATION BOARD DATE: April 21, 2014 TO:

Page 1 of 19 Rent Stabilization Board RE RENT STABILIZATION BOARD DATE: April 21, 2014 TO: Honorable Membe bers of the Rent Stabilization Board FROM: Jay Kelekian, Exe xecutive Director on and Good Cause for Eviction in the 21 st Centur

1.24k views • 64 slides
Instability Cases Arthroscopic Rx for 21yr Stabilization now Old Linebacker Open

Instability Cases Arthroscopic Rx for 21yr Stabilization now Old Linebacker Open

Instability Cases Arthroscopic Rx for 21yr Stabilization now Old Linebacker Open Stabilization with now Nonoperative Subluxation Treatment and Bony Stabilization at end Bankart?? of Season Case Report 21 Yr Old

684 views • 42 slides
AIM application on GHG stabilization scenarios Workshop on GHG Stabilization Scenarios Tsukuba,

AIM application on GHG stabilization scenarios Workshop on GHG Stabilization Scenarios Tsukuba,

AIM application on GHG stabilization scenarios Workshop on GHG Stabilization Scenarios Tsukuba, Japan on January 22-23, 2004 Yuzuru MATSUOKA Kyoto University, Japan and Mikiko KAINUMA National Institute for Environmental Studies, Japan

467 views • 13 slides
FASOM GHG2, Stabilization Scenarios and GHG2, Stabilization Scenarios and FASOM other

FASOM GHG2, Stabilization Scenarios and GHG2, Stabilization Scenarios and FASOM other

FASOM GHG2, Stabilization Scenarios and GHG2, Stabilization Scenarios and FASOM other items other items Bruce A. McCarl Regents Professor of Agricultural Economics, Texas A&amp;M University Presented at Land Modeling Subgroup Session

394 views • 35 slides
Legislative Participation Concerning Section 301 Tariffs Ye Joo (Quinn) Lee Mentor: Dr. Jack

Legislative Participation Concerning Section 301 Tariffs Ye Joo (Quinn) Lee Mentor: Dr. Jack

Legislative Participation Concerning Section 301 Tariffs Ye Joo (Quinn) Lee Mentor: Dr. Jack Zhang My name is Quinn Lee and I am a junior in the College of Liberal Arts and Sciences at KU studying economics and political science. This project

643 views • 13 slides
Outline New Developments in Point-Stabilization and Path-Following First Part: Point

Outline New Developments in Point-Stabilization and Path-Following First Part: Point

Outline New Developments in Point-Stabilization and Path-Following First Part: Point Stabilization Switched seesaw system A. Pedro Aguiar Seesaw control systems design pedro@isr.ist.utl.pt Stabilization of underactuated vehicles

218 views • 18 slides
Illinois Budget Briefing | Fiscal Year 2015 | Pat Quinn, Governor F IVE Y EAR B LUEPRINT T O S

Illinois Budget Briefing | Fiscal Year 2015 | Pat Quinn, Governor F IVE Y EAR B LUEPRINT T O S

State of Illinois Budget Briefing | Fiscal Year 2015 | Pat Quinn, Governor F IVE Y EAR B LUEPRINT T O S ECURE I LLINOIS E CONOMIC F UTURE FY 2015-2019 1 S ECURING I LLINOIS F INANCIAL F UTURE Governor Pat Quinn Governor Pat Quinn FY

530 views • 20 slides
Chorus America Conference Earning the Right to Ask for a Gift June 28, 2019 Richard Quinn

Chorus America Conference Earning the Right to Ask for a Gift June 28, 2019 Richard Quinn

Chorus America Conference Earning the Right to Ask for a Gift June 28, 2019 Richard Quinn Quinn Consulting Associates Whats the #1 reason individuals make gifts to non-profit organizations? The answer Because they were asked.

1.36k views • 106 slides
= + N 1 N M measurements M 1 sparse signal Problem : Solve for x Basis pursuit,

= + N 1 N M measurements M 1 sparse signal Problem : Solve for x Basis pursuit,

Sparse processing / Compressed sensing Model : y = Ax + n , x is sparse = + N 1 N M measurements M 1 sparse signal Problem : Solve for x Basis pursuit, LASSO (convex objective function) Matching pursuit (greedy method)

210 views • 17 slides
Practical denoising of clipped or overexposed noisy images Alessandro Foi www.cs.tut.fi/~foi

Practical denoising of clipped or overexposed noisy images Alessandro Foi www.cs.tut.fi/~foi

1 TAMPERE UNIVERSITY OF TECHNOLOGY Department of Signal Processing Transforms and Spectral Methods Group Practical denoising of clipped or overexposed noisy images Alessandro Foi www.cs.tut.fi/~foi EUSIPCO-2008 - Lausanne, Switzerland,

2.88k views • 40 slides
Stochastic Models of an Uncertain World x = F ( x , u ) x = F ( x , u , 1 ) Lecture

Stochastic Models of an Uncertain World x = F ( x , u ) x = F ( x , u , 1 ) Lecture

Stochastic Models of an Uncertain World x = F ( x , u ) x = F ( x , u , 1 ) Lecture 9: y G ( x ) y G ( x , 2 ) = = Observers and Kalman Filters Actions are uncertain. CS 344R/393R: Robotics Observations are

228 views • 3 slides
Scalable Stream Processing - Spark Streaming and Beam Amir H. Payberah payberah@kth.se

Scalable Stream Processing - Spark Streaming and Beam Amir H. Payberah payberah@kth.se

Scalable Stream Processing - Spark Streaming and Beam Amir H. Payberah payberah@kth.se 26/09/2019 The Course Web Page https://id2221kth.github.io 1 / 65 Where Are We? 2 / 65 Stream Processing Systems Design Issues Continuous vs.

1.49k views • 104 slides
Single versus coincidence detection of cell-derived vesicles by flow cytometry Edwin van der Pol

Single versus coincidence detection of cell-derived vesicles by flow cytometry Edwin van der Pol

Single versus coincidence detection of cell-derived vesicles by flow cytometry Edwin van der Pol 1,2 Martin van Gemert 1 , Auguste Sturk 2 , Rienk Nieuwland 2 , and Ton van Leeuwen 1 February 3rd, 2013 1 Biomedical Engineering and Physics; 2

412 views • 27 slides
Outline Assessing the precision of estimates of variance Estimates and standard errors

Outline Assessing the precision of estimates of variance Estimates and standard errors

Outline Assessing the precision of estimates of variance Estimates and standard errors components Summarizing mixed-effects model fits Douglas Bates A simple (but real) example Department of Statistics University of Wisconsin Madison A

371 views • 7 slides
Understanding and Mitigating the Tradeoff Between Robustness and Accuracy Aditi Raghunathan*

Understanding and Mitigating the Tradeoff Between Robustness and Accuracy Aditi Raghunathan*

Understanding and Mitigating the Tradeoff Between Robustness and Accuracy Aditi Raghunathan* Sang Michael Xie* Fanny Yang John C. Duchi Percy Liang Stanford University Adversarial examples Standard training leads to models that are not

619 views • 50 slides
Blockchains, the web and standardization: the big opportunity conversation starter Keynote

Blockchains, the web and standardization: the big opportunity conversation starter Keynote

Blockchains, the web and standardization: the big opportunity conversation starter Keynote Arvind Narayanan Princeton University @random_walker Standardization: is it too soon? Are Bitcoin and other blockchains sound? Academic research on

445 views • 27 slides
A Statement on Standardization Orr Dunkelman 1 , Atul Luykx 2 , Lo Perrin 3 1 orrd@cs.haifa.ac.il

A Statement on Standardization Orr Dunkelman 1 , Atul Luykx 2 , Lo Perrin 3 1 orrd@cs.haifa.ac.il

A Statement on Standardization Orr Dunkelman 1 , Atul Luykx 2 , Lo Perrin 3 1 orrd@cs.haifa.ac.il 2 Atul.Luykx@esat.kuleuven.be 2 leo.perrin@uni.lu March 7, 2017 Fast Sofware Encryption 2017 ISO/IEC ISO, IEC ISO promotes worldwide

136 views • 9 slides

More recommend