1250918
1250918
https://ebookbell.com/product/regularity-and-complexity-in-
dynamical-systems-1st-edition-albert-c-j-luo-auth-2501836
https://ebookbell.com/product/regularity-and-complexity-in-dynamical-
systems-2nd-edition-albert-c-j-luo-5766106
https://ebookbell.com/product/regularity-and-irregularity-of-
superprocesses-with-1-stable-branching-mechanism-1st-edition-leonid-
mytnik-5737214
https://ebookbell.com/product/regularity-and-stochasticity-of-
nonlinear-dynamical-systems-leoncini-6751844
https://ebookbell.com/product/regularity-and-approximability-of-
electronic-wave-functions-1st-edition-harry-yserentant-auth-1441398
Global Regularity And Uniqueness Of Solutions In A Surface Growth
Model Using Rigorous Aposteriori Methods 1st Edition Christian Nolde
https://ebookbell.com/product/global-regularity-and-uniqueness-of-
solutions-in-a-surface-growth-model-using-rigorous-aposteriori-
methods-1st-edition-christian-nolde-51628324
https://ebookbell.com/product/smoothness-regularity-and-complete-
intersection-1st-edition-javier-majadas-2504454
https://ebookbell.com/product/progress-in-partial-differential-
equations-asymptotic-profiles-regularity-and-wellposedness-michael-
reissig-m-ruzhansky-eds-4158386
https://ebookbell.com/product/structure-and-regularity-of-group-
actions-on-onemanifolds-1st-ed-2021-sanghyun-kim-51679986
https://ebookbell.com/product/existence-and-regularity-of-minimal-
surfaces-on-riemannian-manifolds-mn27-course-book-jon-t-pitts-51954186
Regularity and Complexity in Dynamical Systems
Albert C. J. Luo
123
Albert C. J. Luo
School of Engineering
Mechanical and Industrial Engineering
Southern Illinois University
Room: EB2064
Edwardsville, IL 62026-1805
USA
e-mail: [email protected]
vii
viii Preface
ix
x Contents
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
Chapter 1
Nonlinear Continuous Dynamical Systems
where ẋ = dx/dt is differentiation with respect to time t, which is simply called the
velocity vector of the state variables x. With an initial condition of x(t0 ) = x0 , the
solution of Eq. (1.1) is given by
(i) The ordinary differential equation with the initial condition is called a dynamical
system.
(ii) The vector function f(x, t, p) is called a vector field on domain .
(iii) The solution (x0 , t − t0 , p) is called the flow of dynamical systems.
(iv) The projection of the solution (x0 , t − t0 , p) on domain is called the tra-
jectory, phase curve or orbit of dynamical system, which is defined as
Definition 1.2 If the vector field of the dynamical system in Eq. (1.1) is independent
of time, such a system is called an autonomous dynamical system. Thus, Eq. (1.1)
becomes
n
||A|| = |ai j |. (1.6)
i, j=1
∂f/∂x is called the spatial derivative of f(x, t, p) at x0 , and the derivative is given
by the Jacobian matrix
∂f(x, t, p) ∂ fi
= . (1.10)
∂x ∂ x j n×n
1.1 Continuous Dynamical Systems 3
Proof The proof of this theorem can be referred to the book by Coddington and
Levinson (1955).
for all t ∈ [t0 , t1 ] and δ1 and δ2 are positive constants. For t ∈ [t0 , t1 ], one obtains
Since δ1 > 0 and δ2 > 0 are constants, one obtains G(t) > 0 and G(t) ≥ g(t) for
t ∈ [t0 , t1 ]. The derivative of the foregoing equation gives
Ġ(t) = δ1 g(t).
Further,
so
d
(log G(t)) ≤ δ1 ⇒ log G(t) ≤ δ1 (t − t0 ) − log G(t0 ).
dt
Proof From the method of successive approximations, with the local Lipschitz con-
dition, the two initial value problems become
t t
x(t) = x0 + f(x, τ, p)dτ and y(t) = y0 + f(y, τ, p)dτ.
t0 t0
Thus,
t
x(t) − y(t) = x0 − y0 + (f(x, τ, p) − f(y, τ, p))dτ ,
t0
t
||x(t) − y(t)|| ≤ ||x0 − y0 || + ||f(x, τ, p) − f(y, τ, p)||dτ .
t0
Using the local Lipschitz condition of ||f(x, τ, p) − f(y, τ, p)|| ≤ L||x − y|| gives
t
||x(t) − y(t)|| ≤ ε + L||x(τ ) − y(τ )||dτ .
t0
f(x∗ , p) = 0. (1.18)
The linearized system of the nonlinear system ẋ = f(x, p) in Eq. (1.4) at the equi-
librium point x∗ is given by
corresponding vectors for the negative, positive and zero eigenvalues of Df(x∗ , p) are
{uki } (ki ∈ Ni , i = 1, 2, 3), respectively. The stable, unstable and invariant subspaces
of the linearized nonlinear system in Eq. (1.19) are defined as
6 1 Nonlinear Continuous Dynamical Systems
E s = span uk |(Df(x∗ , p) − λk I)uk = 0, λk < 0, k ∈ N1 ⊆ N ∪ ∅ ;
E u = span uk |(Df(x∗ , p) − λk I)uk = 0, λk > 0, k ∈ N2 ⊆ N ∪ ∅ ; (1.20)
E i = span uk |(Df(x∗ , p) − λk I)uk = 0, λk = 0, k ∈ N3 ⊆ N ∪ ∅ .
The stable, unstable, center subspaces of Eq. (1.19) are linear subspaces spanned by
{uki , vki } (ki ∈ Ni , i = 1, 2, 3), respectively. Set Ni = {l1 , l2 , · · · , ln i } ∪ ∅ ⊆ N ∪ ∅
and N = {1, 2, · · · , n−m} with l j ∈ N ( j = 1, 2, · · · , n i ). ∪i=1
3 N = N , ∩3 N = ∅
i i=1 i
and i=1 n i = n − m. The stable, unstable, center subspaces of the linearized non-
3
where E s , E u and E c are the stable, unstable and center spaces E s , E u and E c ,
respectively.
Proof This proof is the same as the linear system in Appendix A.
Definition 1.12 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ and f(x, p) is C r (r ≥ 1)-
continuous in a neighborhood of the equilibrium x∗ . The corresponding solution is
x(t) = (x0 , t − t0 , p) = t (x0 ). The linearized system of the nonlinear system at
1.2 Equilibriums and Stability 7
Sloc (x, x∗ ) = {x ∈ U (x∗ )| lim x(t) = x∗ , x(t) ∈ U (x∗ ) for all t ≥ 0} (1.25)
t→∞
is called the local stable manifold of x∗ , and the corresponding global stable
manifold is defined as
(a) The local stable invariant manifold Sloc (x, x∗ ) possesses the following
properties:
8 1 Nonlinear Continuous Dynamical Systems
(i) for x∗ ∈ Sloc (x, x∗ ), Sloc (x, x∗ ) possesses the same dimension of E s
and the tangential space of Sloc (x, x∗ ) is identical to E s ;
(ii) for x0 ∈ Sloc (x, x∗ ), x(t) ∈ Sloc (x, x∗ ) for all time t ≥ t0 and lim x(t) =
t→∞
x∗ ;
/ Sloc (x, x∗ ), ||x − x∗ || ≥ δ for δ > 0 with t ≥ t1 ≥ t0 .
(iii) for x0 ∈
(b) The local unstable invariant manifold Uloc (x, x∗ ) possesses the following prop-
erties:
(i) for x∗ ∈ Uloc (x, x∗ ), Uloc (x, x∗ ) possesses the same dimension of E u
and the tangential space of Uloc (x, x∗ ) is identical to E u ;
(ii) for x0 ∈ Uloc (x, x∗ ), x(t) ∈ Uloc (x, x∗ ) for all time t ≤ t0 and lim x(t) =
t→−∞
x∗ ;
/ Uloc (x, x∗ ), ||x − x∗ || ≥ δ for δ > 0 with t ≤ t1 ≤ t0 .
(iii) for x0 ∈
Proof The proof for stable and unstable manifold can be referred to Hartman (1964).
The proof for center manifold can be referenced to Marsden and McCracken (1976)
or Car (1981).
Theorem 1.6 Consider an n-dimensional, autonomous, nonlinear dynamical system
ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ . Suppose there is a neighborhood
of the equilibrium x∗ as U (x∗ ) ⊂ , then f(x, p) is C r (r ≥ 1)-continuous in a
neighborhood of the equilibrium x∗ . The corresponding solution is x(t) = (x0 , t −
t0 , p). The linearized system of the nonlinear system at the equilibrium point x∗ is
ẏ = Df(x∗ , p)y (y = x − x∗ ) in Eq. (1.19). If the homeomorphism between the local
invariant subspace E(x, x∗ ) ⊂ U (x∗ ) under the flow (x0 , t − t0 , p) of ẋ = f(x, p)
in Eq. (1.4) and the eigenspace E of the linearized system exists with the condition in
Eq. (1.24), in addition to the local stable and unstable invariant manifolds, there is a
C r −1 center manifold Cloc (x, x∗ ). The center manifold possesses the same dimension
of E c for x∗ ∈ Cloc (x, x∗ ), and the tangential space of Cloc (x, x∗ ) is identical to E c .
Thus, the local invariant subspace is decomposed by
Proof The proof for stable and unstable manifold can be referred to Hartman (1964).
The proof for center manifold can be referenced to Marsden and McCracken (1976)
or Car (1981).
Definition 1.13 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ and f(x, p) is C r (r ≥ 1)-
continuous in a neighborhood of the equilibrium x∗ .
(i) The equilibrium x∗ is stable if all ε > 0, there is a δ > 0 such that for all
x0 ⊂ Uδ (x∗ ) where Uδ (x∗ ) = { x| ||x − x∗ || < δ} and t ≥ 0,
/ Uδ (x∗ ).
(x0 , t − t0 , p) ∈ (1.32)
(iii) The equilibrium x∗ is asymptotically stable if all ε > 0, there is a δ > 0 such
that for all x0 ⊂ Uδ (x∗ ) where Uδ (x∗ ) = { x| ||x − x∗ || < δ} and t ≥ 0,
where
αk βk cos βk t sin βk t
Ei = and Bk = . (1.40)
−βk αk − sin βk t cos βk t
lim ||c(k) || = lim eαk t ||Bk || × ||c0(k) || = 0 for Reλk = αk < 0. (1.41)
t→∞ t→∞
(iii) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is on the invariant circle if
(k) (k)
lim ||c(k) || = lim eαk t ||Bk || × ||c0 || = ||c0 || for Reλk = αk = 0.
t→∞ t→∞
(1.43)
(iv) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is degenerate in the direction
of uk if Imλk = 0.
Definition 1.16 Consider an n-dimensional, autonomous, nonlinear dynamical
system ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ . Suppose there is a
neighborhood of the equilibrium x∗ as U (x∗ ) ⊂ , then f(x, p) is C r (r ≥ 1)-
continuous in the neighborhood, and Eq. (1.24) holds. The corresponding solution
is x(t) = (x0 , t − t0 , p). The linearized system of the nonlinear system at the
equilibrium point x∗ is ẏ = Df(x∗ , p)y (y = x − x∗ ) in Eq. (1.19).
(i) The equilibrium x∗ is said a hyperbolic equilibrium if none of eigenvalues of
Df(x∗ , p) is nonzero real part (i.e., Reλk = 0 (k = 1, 2, · · · , n)).
(ii) The equilibrium x∗ is said a sink if all of eigenvalues of Df(x∗ , p) have negative
real parts (i.e., Reλk < 0 (k = 1, 2, · · · , n)).
(iii) The equilibrium x∗ is said a source if all of eigenvalues of Df(x∗ , p) have
positive real parts (i.e., Reλk > 0 (k = 1, 2, · · · , n)).
(iv) The equilibrium x∗ is said a saddle if it is a hyperbolic equilibrium and
Df(x∗ , p) have at least one eigenvalue with a positive real part (i.e., Reλ j >
0 ( j ∈ {1, 2, · · · , n}) and one with a negative real part (i.e., Reλk < 0 (k ∈
{1, 2, · · · , n}).
(v) The equilibrium x∗ is called a center if all of eigenvalues of Df(x∗ , p) have
zero real parts (i.e., Reλ j = 0 ( j = 1, 2, · · · , n)) with distinct eigenvalues.
1.2 Equilibriums and Stability 11
(vi) The equilibrium x∗ is called a stable node if all of eigenvalues of Df(x∗ , p) are
real λk < 0 (k = 1, 2, · · · n).
(vii) The equilibrium x∗ is called an unstable node if all of eigenvalues of Df(x∗ , p)
are real λk > 0 (k = 1, 2, · · · n).
(viii) The equilibrium x∗ is called a degenerate case if all of eigenvalues of Df(x∗ , p)
are zero λk = 0 (k = 1, 2, · · · n).
As in Appendix A, the refined classification of the linearized nonlinear system
at equilibrium points should be discussed. The generalized stability and bifurcation
of flows in linearized, nonlinear dynamical systems in Eq. (1.4) will be discussed as
follows.
Definition 1.17 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ . Suppose there is a
neighborhood of the equilibrium x∗ as U (x∗ ) ⊂ , then f(x, p) is C r (r ≥ 1)-
continuous in the neighborhood, and Eq. (1.24) holds. The corresponding solu-
tion is x(t) = (x0 , t − t0 , p). The matrix Df(x∗ , p) in Eq. (1.19) possesses n
eigenvalues λk (k = 1, 2, · · · , n). Set N = {1, 2, · · · , n}, Ni = {l1 , l2 , · · · , ln i } ∪ ∅
with l ji ∈ N ( ji = 1, 2, · · · , n i , i = 1, 2, · · · , 6) and i3= 1 n i + 2 i=4
6 n = n. ∪6
i i=1
∗
Ni = N and ∩i=1 Ni = ∅. Ni = ∅ if n i = 0. The matrix Df(x , p) possesses n 1 -
6
(iv) The value p0 in Eq. (1.4) is called a bifurcation value of p if the dynamical char-
acteristics at point (x0∗ , p0 ) change from one stable state into another unstable
state.
Definition 1.22 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ . Suppose there is a
neighborhood of the equilibrium x∗ as U (x∗ ) ⊂ , then f(x, p) is C r (r ≥ 1)-
continuous in the neighborhood, and Eq. (1.24) holds. The corresponding solu-
tion is x(t) = (x0 , t − t0 , p). The matrix Df(x∗ , p) in Eq. (1.19) possesses n
eigenvalues λk (k = 1, 2, · · · , n). Set N = {1, 2, · · · , n}, Ni = {l1 , l2 , · · · , ln i } ∪ ∅
with l ji ∈ N ( ji = 1, 2, · · · , n i , i = 1, 2, · · · , 6) and i=1
3 n + 2 6 n = n.
i i=4 i
∪i=1 Ni = N and ∩i=1 Ni = ∅.Ni = ∅ if n i = 0. The matrix Df(x∗ , p) possesses
6 6
To extend the idea of Definitions 1.14 and 1.15, a new function will be defined to
determine the stability and the stability state switching.
Definition 1.25 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ and f(x, p) is C r (r ≥ 1)-
continuous in a neighborhood of the equilibrium x∗ . The corresponding solution is
x(t) = (x0 , t − t0 , p). Suppose U (x∗ ) ⊂ is a neighborhood of equilibrium x∗ ,
and there are n linearly independent vectors vk (k = 1, 2, · · · , n). For a perturbation
of equilibrium y = x − x∗ , let y(k) = ck vk and ẏ(k) = ċk vk ,
and
1.3 Bifurcation and Stability Switching 21
G (1)
sk (x, p) = vk · Dsk f(x(sk ), p)
T
G (m) (m)
sk (x, p) = vk · Dsk f(x(sk ), p)
T
(1.47)
= vkT · Dsk (Ds(m−1)
k
f(x(sk ), p))
for all x ∈ U (x∗ ) ⊂ and all t ∈ [t0 , ∞). The equilibrium x∗ is called to be
degenerate on the direction vk .
Theorem 1.7 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ and f(x, p) is C r (r ≥ 1)-
continuous in a neighborhood of the equilibrium x∗ (i.e., U (x∗ ) ⊂ ). The corre-
sponding solution is x(t) = (x0 , t −t0 , p). Suppose Eq. (1.24) holds in U (x∗ ) ⊂ .
For a linearized dynamical system in Eq. (1.19), consider a real eigenvalue λk of
matrix Df(x∗ , p) (k ∈ N = {1, 2, · · · , n}) with an eigenvector vk . Let y(k) = ck vk
and ẏ(k) = ċk vk , sk = vkT · y = vkT · (x − x∗ ) in Eq. (1.44) with sk = ck ||vk ||2 .
Define
vice versa.
(ii) For sk = vkT · (x(t) − x∗ ) > 0
vice versa.
(iii) For sk = vkT · (x(t) − x∗ ) > 0
vice versa.
24 1 Nonlinear Continuous Dynamical Systems
G (m) ∗
sk (x , p) = 0 (m = 0, 1, 2, · · · ) (1.63)
and
G (1) ∗
sk (x , p) = vk · Dx f(x(sk ), p)∂ck x∂sk ck
T
ṡk = G (1) ∗
sk (x , p)sk + o(sk ) = λk sk + o(sk ).
(1)
Thus, G sk (x∗ , p) = λk < 0.
(ii) For sk > 0
(1)
Thus, G sk (x∗ , p) = λk > 0.
(iii) For sk > 0
(1)
Thus, G sk (x∗ , p) = λk = 0 and the higher order should be considered. The
higher-order Taylor series expansion gives
For sk > 0
So we have
G (2) ∗
sk (x , p) > 0.
So
G (2) ∗
sk (x , p) < 0.
m=1
(N = 1, 2, · · · )
for all x ∈ U (x∗ ) ⊂ and all t ∈ [t0 , ∞). The equilibrium x∗ is called the
sink (or stable node) of the (2m k + 1)th order on the direction vk .
(ii) x(k) at the equilibrium x∗ on the direction vk is unstable of the (2m k + 1)th
order if
G (r k) ∗
sk (x , p) = 0, rk = 0, 1, 2, · · · , 2m k ;
vkT · (x(t + ε) − x(t)) > 0 for vkT · (x(t) − x∗ ) > 0; (1.65)
∗
vkT · (x(t + ε) − x(t)) < 0 for vkT · (x(t) − x ) < 0
for all x ∈ U (x∗ ) ⊂ and all t ∈ [t0 , ∞). The equilibrium x∗ is called the
source (or unstable node) of the (2m k + 1)th order on the direction vk .
(iii) x(k) at the equilibrium x∗ on the direction vk is increasingly unstable of the
(2m k )th order if
G (r k) ∗
sk (x , p) = 0, rk = 0, 1, 2, · · · , 2m k − 1;
vkT · (x(t + ε) − x(t)) > 0 for vkT · (x(t) − x∗ ) > 0; (1.66)
∗
vkT · (x(t + ε) − x(t)) > 0 for vkT · (x(t) − x ) < 0
for all x ∈ U (x∗ ) ⊂ and all t ∈ [t0 , ∞). The equilibrium x∗ is called the
increasing saddle of the (2m k )th order on the direction vk .
(iv) x(k) at the equilibrium x∗ on the direction vk is decreasingly unstable of the
(2m k )th order if
G (r k) ∗
sk (x , p) = 0, rk = 0, 1, 2, · · · , 2m k − 1;
vkT · (x(t + ε) − x(t)) < 0 for vkT · (x(t) − x∗ ) > 0; (1.67)
∗
vkT · (x(t + ε) − x(t)) < 0 for vkT · (x(t) − x ) < 0
for all x ∈ U (x∗ ) ⊂ and all t ∈ [t0 , ∞). The equilibrium x∗ is called the
decreasing saddle of the (2m k )th order on the direction vk .
28 1 Nonlinear Continuous Dynamical Systems
2m k
2m k +1 2m k +1
G (r k) ∗ (2m k +1) ∗
r
ṡk = vkT · f(x, p) = sk (x , p)sk + G sk
k
(x , p)sk + o(sk )
rk =1
and
G (r k) ∗
sk (x , p) = 0 for r k = 0, 1, 2, · · · , m k
1.3 Bifurcation and Stability Switching 29
(2m +1)
Thus, G sk k (x∗ , p) < 0.
(ii) For sk > 0
2m k +1
G k (x, p) = ṡk = G s(2m
k
k +1) (x ∗ , p)s
k >0
Thus, G (2m
sk
k +1)
(x∗ , p) > 0.
(iii) For x = x∗ , sk = 0. Using Taylor series expansion gives
2m k −1
G (r k) ∗ (2m k ) ∗
r 2m k 2m k
ṡk = vkT · f(x, p) = sk (x , p)sk + G sk
k
(x , p)sk + o(sk )
rk =1
and
G (r k) ∗
sk (x , p) = 0 for r k = 0, 1, · · · , 2m k − 1
For sk > 0
So we have
G s(2m
k
k ) (x∗ , p) > 0
30 1 Nonlinear Continuous Dynamical Systems
So
G (2m
sk
k ) (x∗ , p) < 0.
rk = ck uk + dk vk =rk erk ,
ṙk = ċk uk + ḋk vk = ṙk erk + rk ėrk (1.72)
and
1 1
ck = [2 (ukT · y) − 12 (vkT · y)] and dk = [1 (vkT · y) − 12 (ukT · y)]
1 = ||uk ||2 , 2 = ||vk ||2 , 12 = ukT · vk and = 1 2 − 212 . (1.73)
where uk⊥ and vk⊥ are the normal vectors of uk and vk , respectively.
1
ċk = [2 G ck (x, p) − 12 G dk (x, p)]
(1.76)
1
ḋk = [1 G dk (x, p) − 12 G dk (x, p)]
1.3 Bifurcation and Stability Switching 31
where
∞
G ck (x, p) = ukT · f(x, p) = G (m) ∗
ck (x , p)r k ,
m
m=1
∞
(1.77)
G dk (x, p) = vkT · f(x, p) = G (m) ∗
dk (x , p)r k ;
m
m=1
m
G (m)
ck (x ∗
, p) = u T
k · ∂ (m)
x f(x, p)[u k cos θk + vk sin θk ] ∗ ,
(x ,p)
(1.78)
(m) ∗ (m) m
G dk (x , p) = vk · ∂x f(x, p)[uk cos θk + vk sin θk ] ∗ .
T
(x ,p)
Thus
∞
ṙk = ċk cos θk + ḋk sin θk = G r(m)
k
(θk )rkm
m=1
∞
(1.79)
(m)
θ̇k = rk−1 (ḋk cos θk − ċk sin θk ) = rk−1 G θk (θk )rkm−1
m=1
where
1
G r(m) (θk ) = [(2 cos θk − 12 sin θk )ukT + (2 sin θk − 12 cos θk )vkT ]
k
·∂x(m) f(x, p)(uk cos θk + vk sin θk )m ∗ ,
(x ,p)
1
G (m)
θk (θk ) = − [(2 sin θk + 12 cos θk )ukT − (1 cos θk − 12 sin θk )vkT ]
·∂x(m) f(x, p)(uk cos θk + vk sin θk )m ∗ .
(x ,p)
(1.80)
From the foregoing definition, consider the first order terms of G-function
(1) (1)
G (1)
ck (x, p) = G ck 1 (x, p) + G ck 2 (x, p)
(1) (1) (1)
(1.81)
G dk (x, p) = G dk 1 (x, p) + G dk 2 (x, p)
where
G (1)
ck 1 (x, p) = uk · Dx f(x, p)∂ck x = uk · Dx f(x, p)uk
T T
and
(1)
G dk 1 (x, p) = vkT · Dx f(x, p)∂ck x = vkT · Dx f(x, p)uk
= vkT · (−βk vk + αk uk ) = −βk 2 + αk 12 ,
(1.83)
G (1)
dk 2 (x, p) = vk · Dx f(x, p)∂dk x = vk · Dx f(x, p)vk
T T
1 (1) (1)
G r(1) (θk ) = [(G (1) (1)
ck 2 − G dk 12 ) cos θk + (G dk 1 − G ck 12 ) sin θk ] = αk ;
k
1
G (1)
θk (θk ) = [(G (1) (1) (1) (1)
dk 1 − G ck 12 ) cos θk − (G ck 2 − G dk 12 ) sin θk ] = −βk .
(1.85)
Furthermore, Eq. (1.79) gives
and
where
αk βk cos βk t sin βk t
Ei = and Bk = . (1.91)
−βk αk − sin βk t cos βk t
(m) (m)
If G rk (θk ) and G θk (θk ) are dependent on θk , Eq. (1.79) gives the dynamical
systems based on the polar coordinates on the invariant plane of (uk , vk ) of matrix
(m) (m)
Df(x∗ , p) with a pair of eigenvectors uk ± ivk . If G rk (θk ) and G θk (θk ) are inde-
pendent of θk , the deformed dynamical system on the plane of (uk , vk ) is dependent
on rk , then the G-functions can be used to determine the stability of x(k) at the
equilibrium x∗ on the plane of (uk , vk ).
Definition 1.29 Consider an n-dimensional, autonomous, nonlinear dynamical sys-
tem ẋ = f(x, p) in Eq. (1.4) with an equilibrium point x∗ and f(x, p) is C r (r ≥ 1)-
continuous in a neighborhood of the equilibrium x∗ . The corresponding solution
is x(t) = (x0 , t − t0 , p). Suppose U (x∗ ) ⊂ is a neighborhood of equilib-
rium x∗ , and there are n linearly independent vectors vk (k = 1, 2, · · · , n). For a
linearized dynamical system in Eq. (1.19), √ consider a pair of complex eigenvalue
αk ± iβk (k ∈ N = {1, 2, · · · , n}, i = −1) of matrix Df(x∗ , p) with a pair of
(k) (k)
eigenvectors uk ± ivk . On the invariant plane of (uk , vk ), consider y(k) = y+ + y−
with Eqs. (1.72) and (1.74). For any arbitrarily small ε > 0, the stability of the
equilibrium x∗ on the invariant plane of(uk , vk ) can be determined.
(i) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is spirally stable if
(iii) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is stable with the m k th-order
singularity if for θk ∈ [0, 2π ]
(iv) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is spirally unstable with the
m k th-order singularity if for θk ∈ [0, 2π ]
(v) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is circular if for θk ∈ [0, 2π ]
rk (t + ε) − rk (t) = 0. (1.96)
34 1 Nonlinear Continuous Dynamical Systems
(vi) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is degenerate in the direction
of uk if
G r(1)
k
(θk ) = αk < 0. (1.98)
(ii) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is spirally unstable if and
only if
G r(1)
k
(θk ) = αk > 0. (1.99)
(iii) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is stable with the m k th-order
singularity if and only if for θk ∈ [0, 2π ]
(iv) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is spirally unstable with the
m k th-order singularity if and only if for θk ∈ [0, 2π ]
(v) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is circular if and only if for
θk ∈ [0, 2π ]
(vi) x(k) at the equilibrium x∗ on the plane of (uk , vk ) is degenerate in the direction
of uk if and only if
I mλk = βk = 0 and G (s k)
θk (θk ) = 0 for sk = 2, 3, · · · . (1.103)
Proof For x = x∗ , sk = 0. The first order approximation of ċk and ḋk in the Taylor
series expansion gives
1 (1)
ċk = [2 G (1)
ck (x, p) − 12 G dk (x, p)]
1
ḋk = [1 G (1) (1)
dk (x, p) − 12 G ck (x, p)]
where rk = ck2 + dk2 and
(1)
G ck 1 (x, p) = ukT · Dx f(x, p)∂ck x = ukT · Dx f(x, p)uk
= ukT · (−βk vk + αk uk ) = αk 1 − βk 12 ,
G (1) T T
ck 2 (x, p) = uk · Dx f(x, p)∂dk x = uk · Dx f(x, p)vk
= ukT · (βk uk + αk vk ) = αk 12 + βk 1 ;
and
(1)
G dk 1 (x, p) = vkT · Dx f(x, p)∂ck x = vkT · Dx f(x, p)uk
= vkT · (−βk vk + αk uk ) = −βk 2 + αk 12 ,
G (1)
dk 2 (x, p) = vk · Dx f(x, p)∂dk x = vk · Dx f(x, p)vk
T T
Therefore, using
(1) (1)
G (1)
ck (x, p) = G ck 1 (x, p)ck + G ck 2 (x, p)dk
or
ċk αk βk ck
= .
ḋk −βk βk dk
rk = ck uk + dk vk = rk erk ,
36 1 Nonlinear Continuous Dynamical Systems
where
ck = rk cos θk , dk = rk sin θk ;
erk = cos θk uk + sin θk vk ,
eθk = − cos θk uk⊥ 3 + sin θk vk⊥ 4
and
ṙk = ċk uk + ḋk vk = ṙk erk + rk ėrk ,
ėrk = −θ̇k uk sin θk + θ̇k vk cos θk .
Thus
ṙk = ċk cos θk + ḋk sin θk ,
θ̇k = rk−1 (ḋk cos θk − ċk sin θk ).
For the first approximation of the relative change rate in the erk direction, we obtain
ṙk = (αk ck + βk dk ) cos θk + (−βk ck + αk dk ) sin θk
= αk rk .
Further
ṙ k = αk rk .
Therefore,
(1)
G r(1)
k
(θk ) = αk and G θk (θk ) = −βk .
In fact, the relative change rate in the erk direction is of interest. The corresponding
higher-order expression is given by
m k −1
G r(skk ) (θk )rk k + G r(m k ) (θ )r m k + o(r m k ).
s
ṙk = k k k
sk =1 k
(1)
(i) For equilibrium stability, rk > 0 and rk → 0. If G rk (θk ) = αk = 0, we have
ṙk = G r(1)
k
(θk )rk = αk rk .
ṙk = G r(m k ) (θ )r k .
m
k k k
If G r(m
k
k)
(θk ) is independent of θk (i.e., G r(m
k
k)
(θk ) = const), it can be used to
(m )
determine the equilibrium stability. Due to rk > 0, if G rk k (θk ) < 0, then
ṙk > 0. Therefore,
then
rk (t + ε) − rk (t) = ṙk ε = 0
vice versa. Therefore rk (t) is constant. x(k) at the equilibrium x∗ on the plane
of (uk , vk ) is circular.
Other documents randomly have
different content
XIX.
SYDÄNMAALLE.
— Klaus-herrako, mikä sen nyt olisi näin muuttanut? Toista hän oli
poikaa, kun kolme vuotta sitten kävi meitä kurittamassa.
— Eikä ole suurta toivoa siitäkään, miten käräjillä käy, jos niitä
jäämme odottamaan, varoittelivat vanhat miehet vanhoilta
kokemuksiltaan. — Lainlukijat ovat herrain kätyreitä, he tuomitsevat,
niinkuin toiset herrat tahtovat, taikka kiskovat meiltä määrättömiä
lahjuksia päittemme hintana.
Silloin oli ainakin Apajan kylän miehille asema ilman muuta selvä:
heidän on sittenkin kiireellä painuttava salolle!
KAKSI HAKUMIESTÄ.
— Ja poikako myöskin…?
Tämän hiihdon hän oli nyt suorittanut aina Sysikorpeen asti. Oli
vältellyt teitä ja pääkyliä, ainoastaan syrjäkylissä hän oli
Rautalammen halki hiihtäessäänkin pistäytynyt. Sikäläiset uutiset oli
hän kuullut: ryöstetyt olivat taas talot, sotaväkeä oli pitkin talvea
samonnut edes ja takaisin, Apajan kylässäkin se oli moneen kertaan
vieraillut; ankaria käräjiä oli pidetty ja etsiskelty oli nuijamiehiä,
mutta tuskin yhtään oli tavattu. Henkiin jääneet olivat kaikki
pakosalla… Nuoren papin luultiin lähteneen uusille armoretkille,
mihin lie sitten häipynytkin, ei ollut kuulunut enää palkkapitäjäänsä…
Erkki oli hiihtänyt siitä edelleen ja nyt hän Sysmän Karmalan törmällä
kautta rantain tiedusteli Tuomas-isännältä sitä asiataan, jota varten
hän oli matkalle lähtenyt.
*****
— Mikä tieto? — Erkki oivalsi, että tällä välin oli asiassa joku
odottamaton, ratkaiseva käänne tapahtunut, mutta ei käsittänyt,
mikä. — Mikä tieto tuli Turkuun?
— Missä ja milloin?
— Ja siellä salolla siis kovia kärsinyt poikani nyt on, totesi Martti,
samalla iloisena ja kiitollisena, kun nyt vihdoin tarkoin tunsi hänen
vaiheensa ja hänen nykyiset olonsa.
— Ja poikaniko myöskin…?
— Heitä aikoi hän pappina palvella siellä salolla. Mutta itse hän
aikoinaan teille aikomuksistaan ilmoittanee, — minä vien hänelle
terveisenne…
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
ebookbell.com