Control and Synchronization in complex networks

Hoang M. Tran April 13, 2012

Contents
1 Synchronization network of 1.1 The model . . . . . . . . . 1.2 Lyapunov direct method . 1.3 Master stability function . N identical linear diusively coupled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 1 3 4

2 Controllability of complex networks

3 Interconnected system approach to study the stability of dynamical complex network 5

1
1.1

Synchronization network of N identical linear diusively coupled nodes

The general model L et al. [2004], Li and Chen [2003], Wang and Chen [2002]:
N

xi

f (xi ) +
j=1

aij (t)G(t)xj , i = 1, 2, ..., N

xi = (xi1 , xi2 , ..., xin )T Rn - state vector of node i G(t) = (gij (t))nn Rnn - inner coupled link matrix b/t 2 components of xi .

1.2

Lyapunov direct method

Existence of synchronous solution: aij (t)

Under which conditions of G(t), aij (t), the chaotic synchronous state x1 (t) = x2 (t) = ... = xN (t) = s(t) is stable. 1

 Results: stability of chaotic synchronous state is determined by means of G(t) and eigenvalues and eigenvectors of A(t) = (aij (t))N N , L et al. [2004], Li and Chen [2003], Wang and Chen [2002] Method: Lyapunovs direct method, use the mode decomposition method L et al. [2002] to translate (equivalently) the synchronous stability problem into the stability problem of N 1 independent ndimensional linear time-varying systems: w = [Df (s(t)) + k (t)G(t) k (t)In ]w, k = 2, ..., N (1)

Df (s(t)) : Jacobian of f (x) at s(t), k (t), k (t) : 1 (t)A(t)T (t) = diag{1 (t), ..., N (t)}, 1 (t)(t) = diag{1 (t), ..., N (t)} Some conditions for the system (1) stable:L et al. [2004], Li and Chen [2003], Wang and Chen [2002] Other problems: robustness, fragility of synchronization: study the system (1). A special case:Wu and Chua [1995]
N

xi1 xi2 .. xin

= = .. =

f1 (xi ) + c
j=1

aij xj1

f2 (xi ) .. fn (xi )

fi (0) = 0, c > 0. 1 is the largest nonzero eigenvalue of A, the synchronization state is stable if 1 T where c T >0 such that 0 is an exponentially stable point of the system = f1 (z) T z1 = f2 (z) .. .. = fn (z)

z1 z2 .. zn

Small world nw: Wang and Chen [2002] used the condition 1 T to explore numerically the c stability of synchronous state of small-world network. + Given N > T /c, there exist p : p 1 the small-world nw will synchronize. p : probability p of adding connection b/t each pair of vertices. + Given p, N : N N ,the small world nw will synchronize. Scale free network:

1.3

Master stability function

The method Pecora and Carroll [1998], Pecora [1998]: xi Rm - state vector of node i, dynamics of each node: xi = F (xi ) +
j j Gij = 0. G = (Gij )N N is adjacency matrix, H(xj ) - coupling among state variables. Denote x = (x1 , x2 , ..., xN ), F (x) = [F (x1 ), ..., F (xN )], H(x) = [H(x1 ), H(x2 ), ..., H(xN ) the whole system dynamics can be rewritten:

Gij H(xj )

(2)

x = F (x) + G H(x) x is a synchronous state (x1 = ... = xN ), perturb it by (3) becomes: = [1N DF + G DH] DF , DH are Jacobian matrices at x. Diagonalize (4) we get: k = [DF + k DH]k , k is eigenvalue of G, k = 0, 1, ..., N 1.

(3)

(4)

Master stability function: max - the maximum Floquet or Lyapunov exponent for the variational equation: = [DF + ( + i)DH] max is a function of , ( = + i). sign(max ) reveal the stability of eigenmode. ( = limt limZ0 0 1 ln |Z(t)| ) t |Z0 | Applications : Barahona and Pecora [2002]: for many oscillators (periodic, quasi-periodic, chaotic,...) sync. state stable if k (1 , 2 ) or max /1 < 2 /1 = . SW with each node is connected (coupled) with 2k nearest nodes, stable synchronous state if 0 k > kmin N (3 + 2)/(2 3 ) 0.429 1/2 N Motter et al. [2005]: Asymmetric coupling, use R = N /2 numerically and approximately: synchronization is determined by average degree of nodes not degree distribution or system size. SF: synchronizability is highly correlated with heterogeneity. Cost involved in nw coupling is signicantly reduced compared to unweighted nw. Some special cases Rangarajan and Ding [2002] studied the system: xi (t) = f (xi (t))+ j=1 bij xj (t), j bij = hmax 0. The synchronized state is stable if bij > N , i, j, i = j, hmax is the largest Lyapunov exponent of xi (t) = f (xi (t)). Extensions: use Jordan canonical form and the Gershgnrin disk theory to generalize the o concept of master stability function for nw with 3
N

 diagonalization is not necessary possible Nishikawa and Motter [2006b,a], Chen et al. [2002], asymmetry coupling Wu and Jiao [2008].

Controllability of complex networks

Controllability in general x(t) = Ax(t) + Bu(t) Liu et al. [2011]: ND = max{N |M |, 1}, M : Maximal matching. Wang et al. [2012]: perturb the structure of the network: algorithm on adding minimum number of edges to make nw controllable by one node Ideas to develop: Ferrarini [2011]. Li et al. [2004] xik xik = f (xik (t), t) + = f (xik (t), t) +
N j=1,=ik cik aik j G(xj (t) xik (t)) + uik , k = 1, 2, ..., l N j=1,=ik cik aik j G(xj (t) xik (t)), k = l + 1, ..., N

Conditions for globally and locally stable about the synchronous state (x1 = ... = xN = x, f (x) = 0) Pinning scheme for radon nw and scale - free nw Method: Lyapunov direct method. Sorrentino et al. [2007] dxi Lij h(xj ) = f (xi ) dt j=1
N n

(i ck )i (xi s)
k=1

Use pinned nodes C = {c1 , ..., cn } to lead other nodes toward some desired reference evolution s(t) Change variable yi = xi , yN +1 = s(t), transform to (2) then apply the method of MSF. Chen et al. [2007] x1 xi = f (x1 (t), t) + c j=1 a1j xj (t) c(x1 (t) s(t)) . N = f (xi (t), t) + c j=1 a1j xj (t), i = 2, ..., N
N

Sucient conditions for this nw to be driven to s(t). Adaptive adjustment of coupling strength.

Interconnected system approach to study the stability of dynamical complex network

We will use the method developed in Khalil [2002] to study the stability, region of attraction for complex network which can be representend in the following form (Li et al. [2004]): xik xik = = f (xik ) + cik aik j (xj xik ) cik ik dik (xik x), k = 1, 2, ..., l N f (xik ) + j=1,=ik cik aik j (xj xik ), k = l + 1, ..., N
N j=1,=ik

(5)

with x satises: f () = 0 or x is the equilibrium of xi = f (xi ) and the coupling strength x N cik ik satises: cik ik aik ik + j=1 cik j aik j = 0, dik > 0.
j=i

For the whole network to be stable, each node must be stable or equilibrium point of the following equation is stable. xi = f (xi ) (6) Applying the converse Lyapunov Theorem, there must be exist a positive denite decrescent Lyapunov function V (t, x) such that its derivative along trajectories of each node dynamics (6) is nagative denite. Theorem 1 (Khalil [2002]). Let x = 0 be an equilibrium poit for the nonlinear system x = f (t, x) where f : [0, ) D Rn is cont. di, D = {x Rn |||x|| < r} and the Jacobian matrix [f /x] is bounded on D, uni. in t. Let k, , r0 be positve constants with r0 < r/k. Let D0 = {x Rn |||x|| < r0 }. Assume that the trajectories of the system satisy: ||x(t)|| k||x(t0 )||e(tt0 ) , x(t0 ) D0 , t t0 > 0 Then V : [0, ) D0 R that satises the inequalities c1 ||x||2 V V + f (t, x) t x || V || c4 ||x|| x for some positive constants c1 , c2 , c3 , c4. Moreover, if r = and the origin is glob. expo. stable, then V (t, x)is dened and satises the aforementioned inequalites on Rn . Furthermore, if the system is autonomous, V can be chosen independent of t. The function V (t, x) = i=1 di V (t, xi ) is a composite Lyapunov function for the whole network. We try V (t, x) as a Lyapunov function candidate for (5). The derivative of V (t, x) along the trajectories of (5) is given by: l N N V V V + f (xi ) + dik cik aik j (xj xik ) cik ik dik (xik x) V (t, x) = di t xi xik i=1 k=1 j=1,=ik N N V + d ik cik aik j (xj xik ) xik
k=l+1 j=1,=ik n

V (t, x) c2 ||x||2 c3 ||x||2

Now, let us estimate the other terms in the above expression:

gik (xik ) =
j=1,=ik

cik aik j (xj xik ) cik ik dik (xik x)

cik aik j (xj xik )

gik (xik ) =
j=1,=ik

Contents References
Mauricio Barahona and Louis M. Pecora. Synchronization in small-world systems. Phys. Rev. Lett., 89:054101, Jul 2002. T. Chen, X. Liu, and W. Lu. Pinning complex networks by a single controller. Circuits and Systems I: Regular Papers, IEEE Transactions on, 54(6):13171326, 2007. Y. Chen, G. Rangarajan, and M. Ding. General stability analysis of synchronized dynamics in coupled systems. Arxiv preprint nlin/0212051, 2002. A. Ferrarini. Some thoughts on the control of network systems. Network Biology, 1(3-4):186188, 2011. H.K. Khalil. Nonlinear Systems. Prentice Hall, 2002. http://books.google.com/books?id=t_d1QgAACAAJ. ISBN 9780130673893. URL

X. Li and G. Chen. Synchronization and desynchronization of complex dynamical networks: An engineering viewpoint. Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions on, 50(11):13811390, 2003. X. Li, X. Wang, and G. Chen. Pinning a complex dynamical network to its equilibrium. Circuits and Systems I: Regular Papers, IEEE Transactions on, 51(10):20742087, 2004. Y.Y. Liu, J.J. Slotine, and A.L. Barabsi. Controllability of complex networks. Nature, 473(7346): 167173, 2011. J. L, T. Zhou, and S. Zhang. Chaos synchronization between linearly coupled chaotic systems. Chaos, Solitons & Fractals, 14(4):529541, 2002. J. L, X. Yu, and G. Chen. Chaos synchronization of general complex dynamical networks. Physica A: Statistical Mechanics and its Applications, 334(1):281302, 2004. A.E. Motter, CS Zhou, and J. Kurths. Enhancing complex-network synchronization. EPL (Europhysics Letters), 69:334, 2005. T. Nishikawa and A.E. Motter. Maximum performance at minimum cost in network synchronization. Physica D: Nonlinear Phenomena, 224(1):7789, 2006a. 6

T. Nishikawa and A.E. Motter. Synchronization is optimal in nondiagonalizable networks. Physical Review E, 73(6):065106, 2006b. L.M. Pecora and T.L. Carroll. Master stability functions for synchronized coupled systems. Physical Review Letters, 80(10):21092112, 1998. Louis M. Pecora. Synchronization conditions and desynchronizing patterns in coupled limit-cycle and chaotic systems. Phys. Rev. E, 58:347360, Jul 1998. G. Rangarajan and M. Ding. Stability of synchronized chaos in coupled dynamical systems. Physics Letters A, 296(4-5):204209, 2002. Francesco Sorrentino, Mario di Bernardo, Franco Garofalo, and Guanrong Chen. Controllability of complex networks via pinning. Phys. Rev. E, 75:046103, Apr 2007. W.X. Wang, X. Ni, Y.C. Lai, and C. Grebogi. Optimizing controllability of complex networks by minimum structural perturbations. Physical Review E, 85(2):026115, 2012. X.F. Wang and G. Chen. Synchronization in scale-free dynamical networks: robustness and fragility. Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions on, 49(1):54 62, 2002. C.W. Wu and L.O. Chua. Synchronization in an array of linearly coupled dynamical systems. Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions on, 42(8): 430447, 1995. J. Wu and L. Jiao. Synchronization in complex dynamical networks with nonsymmetric coupling. Physica D: Nonlinear Phenomena, 237(19):24872498, 2008.