Applied Mathematics and Computation 181 (2006) 220–246 www.elsevier.com/locate/amc
Center manifolds and normal forms for a class of retarded functional differential equations with parameter associated with Fold-Hopf singularity R. Qesmi a b
a,*
, M. Ait Babram b, M.L. Hbid
a
De´partement de Mathe´matiques, Faculte´ des Sciences Semlalia, Universite´ Cadi Ayyad, B.P. S15, Marrakech, Morocco De´partement de Mathe´matiques, Faculte´ des Sciences et te´chniques Gue´liz, Universite´ Cadi Ayyad, Marrakech, Morocco
Abstract In this paper, we present explicit formulas for computing the coefficients of a center manifolds for the Fold-Hopf singularity in autonomous retarded functional differential equations. As consequence, normal forms associated with the flow on a center manifold up to an arbitrary order are derived. The explicit formulas have been implemented using the computer algebra system Maple. We apply our results to a delayed system in order to show the applicability of the methodology. 2006 Elsevier Inc. All rights reserved.
1. Introduction The dynamics at the onset of several instabilities in a physical system undergoing a bifurcation near an equilibrium point can often be reduced to a simple of ordinary differential equations by the application of both center manifolds and normal forms theories. References on the center manifolds and normal forms may be found, for example in [3,8]. Here the case of retarded functional differential equations (RFDEs) with parameters is considered following the computation method of center manifolds we developed in [12] for RFDEs associated with Hopf singularity without parameters. In the past few years, symbolic computations using computer languages such as Maple, Mathematica, and Macsysma have been introduced in computing center manifolds. However, it seems that even with a symbolic manipulator, the computation of center manifolds is still limited to lower-order approximation, since executing such a symbolic program usually quickly runs out of computer memory as the order of center manifolds increases. thus, computationally efficients methodologies and symbolic computer programs need to be developed, in particular, for computing higher-order center manifolds.
*
Corresponding author. E-mail address:
[email protected] (R. Qesmi).
0096-3003/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.amc.2006.01.030
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
221
The center manifolds and normal forms of Fold-Hopf bifurcation has been discussed in [10,13] for ordinary differential equations. In [5,6], Faria and Magalhae`s have considered the computation of coefficients of normal forms for RFDEs of both Hopf and Bogdanov singularities. However, it is difficult to apply the method to compute the explicit expressions for these coefficients since it demands much more computation efforts for high-order normal forms. In [1,2], M. Ait Babram et al have considered RFDEs without parameters, and they derive an initial value problem (IVP) in finite dimension to compute the terms of center manifolds. However, the ingredients of the obtained (IVP) have not an explicit form, and the technicalities used to compute the solution of the (IVP) are theoretical and can not be programmed on machine. The attention of this paper will be focused on the development of methodology and software for computing the center manifolds of FoldHopf bifurcation for RFDEs with parameters. According to the structure of linearized equation of a retarded system evaluated at an equilibrium, the case considered in this paper corresponds to a pair of purely imaginary and a simple zero eigenvalues (Fold-Hopf), while the case studied in [12] for FDEs without parameters is only for one pair of purely imaginary eigenvalues (Hopf). As an important consequence, we obtain, for the general situation of bifurcation of Fold-Hopf for RFDEs, explicit formulas giving the coefficients of normal forms in terms of the coefficients of the original equation. For the notation background about the theory of RFDEs and all needed results in the remainder of this paper, we follow [9], as recalled in Section 2 of [12], but we use C n ¼ Cð½r; 0; Rn Þ; r P 0 since we need to work in realization spaces with different dimensions, depending on whether the parameters are incorporated or not incorporated in the realization space variables. The paper is organized as follows: A theorem of characterization of a center manifolds for RFDEs with parameters and its proof are given in Section 2, as consequence, a computational schemes of a center manifold and normal forms associated with Fold-Hopf are presented in the same section. Section 3 outlines the symbolic computation procedure using Maple, and an illustrative example is given in Section 4 to demonstrate the applicability of the obtained results. Conclusions are drawn in Section 5. The results of the examples including the center manifolds and normal forms obtained by executing the Maple programs are listed in the appendices. 2. Center manifolds for FDEs with parameters and main results In this section we present our main results concerning the computation of the terms of center manifolds and normal forms for the next class of RFDEs m X d xðtÞ ¼ Lj ðaÞxðt rj Þ þ f ðxðtÞ; xðt r1 Þ; . . . ; xðt rm Þ; aÞ; dt j¼0
ð2:1Þ
where a 2 Rp , a # Lj(a) are a C1 functions with values in the space of square matrices of order n. f : Rnm Rp ! Rn is assumed to be sufficiently smooth (f 2 C1) such that: f(0, . . . , 0, a) = 0 and Df(0, . . . , 0, a) = 0 for all a 2 Rp , r0 = 0 and rj > 0 for all 0 < j 6 m. If we denote r :¼ max16i6m{ri}, C n :¼ Cð½r; 0; Rn Þ the space of continuous functions from [r, 0] to Rn , and for / 2 Cn LðaÞ/ :¼ L0 ðaÞ/ð0Þ þ
m X
Lj ðaÞ/ðrj Þ; gð/; aÞ :¼ f ð/ð0Þ; /ðr1 Þ; . . . ; /ðrm Þ; aÞ
j¼1
then Eq. (2.1) reads as equation x_ ðtÞ ¼ LðaÞxt þ F ðxt ; aÞ.
ð2:2Þ
We denote L0 = L(0). In the sequel of this paper we assume that the following hypothesis is satisfied: (H) The linear equation x_ ðtÞ ¼ L0 xt has a pure imaginary pair (±ix) and a simple zero (k = 0) as characteristic values and no other characteristic values with zero real part.
222
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
One way of considering center manifolds for a differential equations with parameters is to reduce the situation to the case of differential equations without parameters by considering the system x_ ðtÞ ¼ L0 xt þ ½LðaÞ L0 xt þ F ðxt ; aÞ; _ ¼ 0. aðtÞ
ð2:3Þ ð2:4Þ
e :¼ C nþp , and the The solutions of this system are of the form ~xðtÞ :¼ ðxðtÞ; aðtÞÞ 2 Rnþp , the phase space is C system can be written ~x_ ðtÞ ¼ e L~xt þ Fe ð~xt Þ;
ð2:5Þ
where e Lðu; vÞ :¼ ðL0 u; 0Þ and Fe ðu; vÞ :¼ ð½Lðvð0ÞÞ L0 u þ F ðu; vð0ÞÞ; 0Þ with u 2 Cn, v 2 Cp. It is now possible to apply to (2.5) the center manifold theory as introduced in [12]. e denote the infinitesimal generators associated with the equations x_ ðtÞ ¼ L0 xt , and x_ ðtÞ ¼ e Let A and A Lxt , _ ¼ 0, has a unique characteristic value k = 0, with multiplicity p, and respectively. The equation in Cp, aðtÞ the associated generalized eigenspace consists of the elements of Cp which are constant functions, and it is denoted here also by Rp . Let K :¼ rðAÞ \ iR and consider the decomposition Cn = Xc Xs obtained as in Section 2 of [12]. In particular, we consider bases for Xc and X c denoted by U = (/1, . . . , /2), W = col(w1, . . . , w2), W e ¼ K f0g, X e s ¼ X s R where R = e c ¼ X c Rp ; X respectively, and satisfying (W, U) = I. We define K e and the rows e e {v 2 Cp:v(0) = 0}, and consider for bases of X c and X c , respectively, the columns of the matrix U e of the matrix W, e ¼ U
U
0
0
Ip
;
e ¼ W
W
0
0
Ip
;
e C e defined as in Section 2 of [12], and e Ui e ¼ I nþp , where hÆ , Æi is the bilinear form in C which satisfies h W; _e eB e with U¼U 2
0
0
0
3
6 0 e¼6 B 6 4 0
ix 0
0 0
0 0
7 7 7. 5
0
0
0
0p
ix
e ¼X e associated with K. e ec X e s , where X e c is the invariant space of A Then we have the decomposition C In the sequel, we recall the definition of a local center manifold associated to Eq. (2.5). e s , the graph of ~h is said to be a local center manifold Definition 2.1 [9]. Given a C1 map ~ h from R3þp into X ~ ~ associated to Eq. (2.5) if and only if hð0Þ ¼ 0, Dhð0Þ ¼ 0, and there exists a neighborhood V of zero in R3þp such that, for each n 2 V, there exist d = d(n) > 0 and a function x defined on ]dr,d[ such that e þ~ x0 ¼ Un hðnÞ and x verifies Eq. (2.5) on ]d,d[ and satisfies the identity e xt ¼ UzðtÞ þ~ hðzðtÞÞ for t 2 ½0; d½; where z(t) is the unique solution of the ordinary differential equation ( d e e e zðtÞ ¼ BzðtÞ þ Wð0Þ Fe ð UzðtÞ þ~ hðzðtÞÞÞ; dt zð0Þ ¼ n;
n 2 R3þp .
ð2:6Þ
Remark 2.2. (i) If we write ~ hðn; aÞ ¼ ðhðn; aÞ; h0 ðn; aÞÞ for n 2 R3 and a 2 Rp , then Eq. (2.6) is equivalent to BzðtÞ d zðtÞ ¼ þ dt a 0
Wð0Þ½ðLðað0Þ þ h0 ðzðtÞ; aÞð0ÞÞ L0 ÞðUzðtÞ þ hðzðtÞ;aÞÞ þ F ðUzðtÞ þ hðzðtÞ;aÞ; að0Þ þ h0 ðzðtÞ;aÞð0ÞÞ 0
!
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
223
2
3 ix 0 0 with B ¼ 4 0 ix 0 5 and zð0Þ ¼ n 2 R3 . Noting that h0(z(t), a)(0) = 0, because h0(z(t), a) 2 R, and drop0 0 0 ping the auxiliary equations introduced for handling the parameter, we get the reduced equation on the center manifold—represented as a graph over the n and a variables of h(n, a) for n and a sufficiently smalls—associated with the parametrized Eq. (2.1) as ( d zðtÞ ¼ BzðtÞ þ Wð0Þ½ðLðaÞ L0 ÞðUzðtÞ þ hðzðtÞ; aÞÞ þ F ðUzðtÞ þ hðzðtÞ; aÞ; aÞ; dt zð0Þ ¼ n;
n 2 R3 .
(ii) It is noted that the invariance properties of center manifolds guarantee that any small solutions bifurcating from (0, 0, 0) must lie in any center manifold and thus we may follow the local evolution of bifurcating families of solutions in this suspended family of center manifolds, (see [13] for more details). 2.1. Characterization of a local center manifolds for the RFDEs In the following, we give an analytic characterization of a center manifold associated to Eq. (2.1). Theorem 2.3. Given a C1 map h from R3þp into Xs with h(0) = 0 and Dh(0) = 0, a necessary condition for the graph of h to be a local center manifold of Eq. (2.1) is that there exists a neighborhood V of zero in R3þp such that, for each (n, a) 2 V o ohðn; aÞ ohðn; aÞ ohðn; aÞ ðhðn; aÞÞðhÞ ¼ ix ðhÞWð0Þ½ðLðaÞ Lð0ÞÞðUn þ hðn; aÞ ðhÞn1 þ ix ðhÞn2 þ oh on1 on2 on þ
ohðn; aÞ ðhÞWð0ÞF ðUn þ hðn; aÞ; aÞ þ UðhÞWð0Þ½ðLðaÞ Lð0ÞÞðUn þ hðn; aÞ on
þ UðhÞWð0ÞF ðUn þ hðn; aÞ; aÞ o ðhðn; aÞÞð0Þ ¼ L0 hðn; aÞ þ ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞ. oh
ð2:7Þ ð2:8Þ
Proof. Let h be a graph of a local center manifold of Eq. (2.1), then from Definition 2.1 there exist a neighborhood V of zero in R3þp such that, for each (n, a) 2 V, there exist d > 0 such that the solution of (2.1) with initial data Un + h(n, a) exists on the interval ]dr,d[ and it is given by xt ¼ UzðtÞ þ hðzðtÞ; aÞ for t 2 d; d½; such that z(t) is solution of the equation ( d zðtÞ ¼ BzðtÞ þ Wð0Þ½ðLðaÞ L0 ÞðUzðtÞ þ hðzðtÞ; aÞÞ þ F ðUzðtÞ þ hðzðtÞ; aÞ; aÞ; dt zð0Þ ¼ n; n 2 R3 ; 3 2 ix 0 0 with B ¼ 4 0 ix 0 5. 0 0 0 The variation of constants formula of Eq. (2.1) can be written as Z t xt ¼ T ðtÞ/ þ T ðt sÞX 0 ½ðLðaÞ L0 Þxs þ F ðxs ; aÞ ds; t P 0; 0
where (T(t))tP0 is the semi group solution of the linear equation x_ ðtÞ ¼ L0 xt . It follow from the decomposition of the phase space by K : Cn = Xc Xs that the function h satisfies:
224
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
hðzðtÞ; aÞ ¼ T ðtÞhðzðtÞ; aÞ þ
Z
t
T ðt rÞX s0 F ðUzðrÞ þ hðzðrÞ; aÞ; aÞ dr. 0
Then 1 1 1 ½T ðtÞhðn; aÞ hðn; aÞ ¼ ½hðzðtÞ; aÞ hðzð0Þ; aÞ t t t
Z
t
T ðt rÞX s0 ½ðLðaÞ L0 ÞðUzðrÞ þ hðzðrÞ; aÞÞ 0
þ F ðUzðrÞ þ hðzðrÞ; aÞ; aÞ dr; which implies from the fact that h and F are smooth and T(.) is a strongly continuous semi group on the Banach space Cn, that h(n) is in the domain of A, and ohðn; aÞ Ahðn; aÞ ¼ fBn þ Wð0Þ½ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞg on ¼ X s0 ½ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞ Consequently, we have by evaluating the above equation at h 5 0 that o ohðn; aÞ ðhðn; aÞÞðhÞ ¼ ðhÞ fBn þ Wð0Þ½ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞg oh on þ UðhÞWð0Þ½ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞ;
ð2:9Þ
which is the formula (2.7) of theorem. On the other hand, it result from the fact that the semi-flow t#xt = Uz(t) + h(z(t), a) exists on the open ]d, d[ that for h 2 ]d, 0] d d d ðUðhÞn þ hðn; aÞðhÞÞ ¼ x0 ðhÞ ¼ xðhÞ ¼ LðaÞðUzðhÞ þ hðzðhÞ; aÞÞ þ F ðUzðhÞ þ hðzðhÞ; aÞ; aÞ. dh dh dh Consequently, by the fact that
d Uð0Þn dh
¼ Uð0ÞBn ¼ L0 ðUnÞ, we obtain
o ðhðn; aÞÞð0Þ ¼ L0 hðnÞ þ ðLðaÞ L0 ÞðUn þ hðn; aÞÞ þ F ðUn þ hðn; aÞ; aÞ; oh
ð2:10Þ
which achieve the proof of theorem. h 2.2. The computational scheme Let us recall that the function h which represent the center manifold for Eq. (2.1) has the same regularity as the nonlinearity F. From this fact and in view of the assumed smoothness on F, for all m 2 N, we can write m X hðn; aÞ ¼ hk ðn; aÞ þ vðn; aÞ for n 2 V ; k¼2
where hk is the homogeneous part of degree k and v (n, a) = o(j(n, a)jm). Let k 2 N, k P 2. The homogeneous parts of degree k of Eqs. (2.7) and (2.8) are respectively given by ohk ðn; aÞ ohk ðn; aÞ ohk ðn; aÞ ¼ ix n1 þ ix n2 þ Nk1 ðn; aÞ; oh on1 on2 o ðhk ðn; aÞÞð0Þ ¼ L0 hk ðn; aÞ þ Rk1 ðn; aÞ; oh where Nk1 ðn; aÞ ¼ UWð0ÞRk1 ðn; aÞ þ
k1 X ohkjþ1 ðn; aÞWð0ÞRj1 on j¼2
ð2:11Þ ð2:12Þ
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
225
and Ri1 is the homogeneous part of degree i of R(n , a) = (L(a) L0)(Un + h(n, a)) + F(Un + h(n, a), a). In particular, R1 is the homogeneous part of degree 2 of (L(a) L0)Un + F(Un, a) which is independent from terms of a center manifolds. l If P n ¼ ðn1 ; n2 ; n3 Þ 2 R3 , q = (q1, q2, q3), a = (a1, . . . , ap), al ¼ al11 al22 app for l ¼ ðl1 ; . . . ; lp Þ 2 Np , and p jlj ¼ i¼1 li , then we can write X q q q hk ðn; aÞ ¼ hkðq;lÞ n11 n22 n33 al for some hkðq;lÞ 2 X s ; ðq;lÞ2Dk k1
N
X
ðn; aÞðhÞ ¼
q
q
q
3 l 1 2 N k1 ðq;lÞ n1 n2 n3 a
for some N kðq;lÞ 2 X s
ð2:13Þ
ðq;lÞ2Dk
and Rk1 ðn; aÞ ¼
X
q
q
q
n for some Rk1 ðq;lÞ 2 R ;
3 l 1 2 Rk1 ðq;lÞ n1 n2 n3 a
ð2:14Þ
ðq;lÞ2Dk
where Dk ¼ fðq; lÞ 2 N3 Np : jðq; lÞj ¼ kg. Theorem 2.4. Assume that (H) holds. Then the vector of the coefficients of the homogeneous part of degree k of a local center manifold associated with Eq. (2.1) is given in a unique way by the following recursive formulas: For k = 2, (q1, q2, q3, l) 2 D2: If (q1 q2 1)(q1 q2 + 1)(q1 q2) 5 0 then h2ðq1 ;q2 ;q3 ;lÞ ðhÞ ¼ eixðq1 q2 Þh h2ðq1 ;q2 ;q3 ;lÞ ð0Þ þ
1 ðeixh eixðq1 q2 Þh Þ/1 ð0Þw1 ð0Þ ixðq1 q2 1Þ
1 ðeixh eixðq1 q2 Þh Þ/2 ð0Þw2 ð0Þ ixðq1 q2 þ 1Þ 1 ð1 eixðq1 q2 Þh Þ/3 ð0Þw3 ð0Þ R1ðq1 ;q2 ;q3 ;lÞ . þ ixðq1 q2 Þ þ
ð2:15Þ
If q1 q2 1 = 0 then h2ðq1 ;q2 ;q3 ;lÞ ðhÞ
eixh h2ðq1 ;q2 ;q3 ;lÞ ð0Þ
þ heixh /1 ð0Þw1 ð0Þ þ 1 ixh þ ð1 e Þ/3 ð0Þw3 ð0Þ R1ðq1 ;q2 ;q3 ;lÞ . ix
¼
sinðxhÞ ixh ðe eixh Þ/2 ð0Þw2 ð0Þ x ð2:16Þ
If q1 q2 + 1 = 0 then
sinðxhÞ /1 ð0Þw1 ð0Þ þ ðheixh Þ/2 ð0Þw2 ð0Þ x 1 þ ð1 eixh Þ/3 ð0Þw3 ð0Þ R1ðq1 ;q2 ;q3 ;lÞ . ix
h2ðq1 ;q2 ;q3 ;lÞ ðhÞ ¼ eixh h2ðq1 ;q2 ;q3 ;lÞ ð0Þ þ
ð2:17Þ
If q1 q2 = 0 then h2ðq1 ;q2 ;q3 ;lÞ ðhÞ
1 ðeixh 1Þ/1 ð0Þw1 ð0Þ þ h/3 ð0Þw3 ð0Þ ix 1 ixh þ ðe 1Þ/2 ð0Þw2 ð0Þ R1ðq1 ;q2 ;q3 ;lÞ . ix
¼
h2ðq1 ;q2 ;q3 ;lÞ ð0Þ
þ
ð2:18Þ
For k > 2 and (q1, q2, q3, l) 2 Dk: hkðq1 ;q2 ;q3 ;lÞ ðhÞ ¼ eixðq1 q2 Þh hkðq1 ;q2 ;q3 ;lÞ ð0Þ þ
Z
h 0
eixðq1 q2 ÞðhsÞ N k1 ðq1 ;q2 ;q3 ;lÞ ðsÞ ds;
ð2:19Þ
226
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
where the vectors hkðq1 ;q2 ;q3 ;lÞ ð0Þ; ðq1 ; q2 ; q3 ; lÞ 2 Dk are given by solving the following systems: For (q1 q2 1) · (q1 q2 + 1)(q1 q2) 5 0 hkðq1 ;q2 ;lÞ ð0Þ ¼ ½Dðixðq1 q2 ÞÞ1 Ek1 ðq1 ;q2 ;lÞ
ð2:20Þ
for q1 = q2 + 1: M1
hkðq1 ;q2 ;q3 ;lÞ ð0Þ
! ¼
0
Ek1 ðq1 ;q2 ;q3 ;lÞ
! ð2:21Þ
vk1 ðq1 ;q2 ;q3 ;lÞ
for q1 = q2 1 M1
hkðq1 ;q2 ;q3 ;lÞ ð0Þ
! ¼
0
Ek1 ðq1 ;q2 ;q3 ;lÞ
! ð2:22Þ
vk1 ðq1 ;q2 ;q3 ;lÞ
and for q1 = q2 M2
hkðq1 ;q2 ;q3 ;lÞ ð0Þ 0
! ¼
Ek1 ðq1 ;q2 ;q3 ;lÞ
! ð2:23Þ
.
vk1 ðq1 ;q2 ;q3 ;lÞ
With M1 and M2 are the (n + 1) · (n + 1) matrices defined by ! DðxiÞ w> 1 ð0Þ M1 ¼ hw ; exi. i 0 1
and M2 ¼
! Dð0Þ w> 3 ð0Þ ; hw3 ; I Rn i 0
k1 and the second members Ek1 ðq1 ;q2 ;q3 ;lÞ and vðq1 ;q2 ;q3 ;lÞ are a vector given by means of the coefficients of the center manifolds already computed (see (2.27) and (2.29) in the proof below).
Proof. Let (q1, q2, q3, l) 2 Dk, from relation (2.11), we have ohkðq1 ;q2 ;q3 ;lÞ ðhÞ oh
¼ ixðq1 q2 Þhkðq1 ;q2 ;q3 ;lÞ ðhÞ þ N k1 ðq1 ;q2 ;q3 ;lÞ ðhÞ;
ð2:24Þ
or equivalently hkðq1 ;q2 ;q3 ;lÞ ðhÞ ¼ eixðq1 q2 Þh hkðq1 ;q2 ;q3 ;lÞ ð0Þ þ
Z 0
h
eixðq1 q2 ÞðhsÞ N k1 ðq1 ;q2 ;q3 ;lÞ ðsÞ ds
ð2:25Þ
However, for k = 2, we have that N1 ðn; aÞðhÞ ¼ UðhÞWð0ÞR1 ðn; aÞ ¼ ½eixh /1 ð0Þw1 ð0Þ þ eixh /2 ð0Þw2 ð0Þ þ /3 ð0Þw3 ð0ÞR1 ðn; aÞ; it follow that: Z Z h eixðq1 q2 ÞðhsÞ N 1ðq1 ;q2 ;q3 ;lÞ ðsÞ ds ¼ eixðq1 q2 Þh 0
Z
0
h
h
eixðq1 q2 1Þs ds/1 ð0Þw1 ð0Þþ
eixðq1 q2 Þs ds/3 ð0Þw3 ð0ÞÞ R1 ðn; aÞ;
Z
h
eixðq1 q2 þ1Þs ds/2 ð0Þw2 ð0Þ
0
0
which implies, by discussing the cases whether q1 q2 2 {1, 0, 1} or not, the value of h2ðq1 ;q2 ;q3 ;lÞ ðhÞ given in theorem.
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
227
The relation (2.12) is equivalent to ohkðq1 ;q2 ;q3 ;lÞ oh
ð0Þ ¼ L0 hkðq1 ;q2 ;q3 ;lÞ þ Rk1 ðq1 ;q2 ;q3 ;lÞ ;
which give by evaluating the relation (2.24) at h = 0 that k k1 L0 hkðq1 ;q2 ;q3 ;lÞ þ Rk1 ðq1 ;q2 ;q3 ;lÞ ¼ ixðq1 q2 Þhðq1 ;q2 ;q3 ;lÞ ð0Þ þ N ðq1 ;q2 ;q3 ;lÞ ð0Þ:
On the other hand, by taking the values of the linear operator L0 at the both hand sides of relation (2.25), we have Z . 0 k ixðq1 q2 Þ. k 0 ixðq1 q2 Þð.sÞ k1 L hðq1 ;q2 ;q3 ;lÞ ¼ L0 ðe Þhðq1 ;q2 ;q3 ;lÞ ð0Þ þ L e N ðq1 ;q2 ;q3 ;lÞ ðsÞ ds . 0
Consequently, we obtain from the expression of the characteristic equation that Dðixðq1 q2 ÞÞhkðq1 ;q2 ;q3 ;lÞ ð0Þ ¼ Ek1 ðq1 ;q2 ;q3 ;lÞ where Ek1 ðq1 ;q2 ;q3 ;lÞ
¼
Rk1 ðq1 ;q2 ;q3 ;lÞ
þL
0
Z . e
for ðq1 ; q2 ; lÞ 2 Dk ;
ixðq1 q2 Þð.sÞ
0
N k1 ðq1 ;q2 ;q3 ;lÞ ðsÞ ds
N k1 ðq1 ;q2 ;q3 ;lÞ ð0Þ.
ð2:26Þ
ð2:27Þ
We will use also the fact that all center manifolds has rank in the subspace Xs, which implies in particular that hw1 ; hkðq1 ;q2 ;q3 ;lÞ ð.Þi ¼ 0 for all (q1, q2, q3, l) 2 Dk and k > 1. Consequently, we obtain from relation (2.25) that hw ; eixðq1 q2 Þ. ihk ð0Þ ¼ vk1 for ðq ; q ; q ; lÞ 2 D ; ð2:28Þ 1
ðq1 ;q2 ;q3 ;lÞ
1
2
3
k
is given by Z . ¼ w1 ; eixðq1 q2 Þð.sÞ N k1 ðsÞ ds . ðq1 ;q2 ;q3 ;lÞ
where the vector vk1 ðq1 ;q2 ;q3 ;lÞ
ðq1 ;q2 ;q3 ;lÞ
vk1 ðq1 ;q2 ;q3 ;lÞ
0
We introduce the (n + 1) · (n + 1) matrices M1 and M2 defined by ! DðixÞ w> ð0Þ 1 M1 ¼ hw1 ; eix. i 0 and M2 ¼
Dð0Þ hw3 ; I Rn i
w> 3 ð0Þ 0
!
then it follows from the systems (2.26) and (2.28) that the vectors defined by: ! k h ð0Þ ðq1 ;q2 ;q3 ;lÞ ^ hkðq1 ;q2 ;q3 ;lÞ ð0Þ ¼ 0 satisfy M 1^ hkðq1 ;q2 ;q3 ;lÞ ð0Þ ¼
Ek1 ðq1 ;q2 ;q3 ;lÞ vk1 ðq1 ;q2 ;q3 ;lÞ
for q1 = q2 + 1, M 1^ hkðq1 ;q2 ;q3 ;lÞ ð0Þ for q1 = q2 1, and
¼
!
Ek1 ðq1 ;q2 ;q3 ;lÞ vk1 ðq1 ;q2 ;q3 ;lÞ
!
ð2:29Þ
228
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
M 2^ hkðq1 ;q2 ;q3 ;lÞ ð0Þ for q1 = q2.
¼
Ek1 ðq1 ;q2 ;q3 ;lÞ
!
vk1 ðq1 ;q2 ;q3 ;lÞ
h
The uniqueness of hkðq1 ;q2 ;q3 ;lÞ ð0Þ is a consequence of the next lemma: Lemma. The (n + 1) · (n + 1) matrices M1 and M2 are invertibles. Proof. Let x 2 Rn , g 2 R such that x M1 ¼ 0; g then (
DðixÞx þ gw> 1 ð0Þ ¼ 0; ix. hw ; e ix ¼ 0; 1
then from the fact that w1(0)D(ix) = 0, we have g = 0, and the above system becomes DðixÞx ¼ 0; hw ; eix. ix ¼ 0. 1
From the fact that dim ker D(ix) = 1 and since we have D(ix)/1(0) = 0, then the first equation implies that x 2 span{/1(0)}, it follows that there exist c 2 R such that x = c/1(0), and from the second equation we have chw1,/1i = 0 which yields c = 0 and finally we obtain x = 0. Consequently M1 is invertible. By the same manner as before, we prove that M2 is invertible. Which completes the proof of theorem. h 2.3. Normal forms Having computed the center manifolds of Eq. (2.1), we obtain the associated reduced equation as follow: d zðtÞ ¼ BzðtÞ þ H ðzðtÞ; aÞ; dt 0 1 ix 0 0 where B ¼ @ 0 ix 0 A; z 2 R3 ; a 2 Rp and 0 0 0
ð2:30Þ
H ðz; aÞ :¼ Wð0Þ½ðLðaÞ L0 ÞðUz þ hðz; aÞÞ þ F ðUz þ hðz; aÞ; aÞ. It is now easy to compute its normal forms up to a desired order. The basic idea of normal form theory consists of employing successive, near identity nonlinear transformations to eliminate the so-called non-resonant nonlinear terms, and the terms called resonant which cannot be eliminated are remained in normal forms. Assume that by a nonlinear transformation z ¼ u þ T ðu; aÞ;
u 2 R3 ;
a 2 Rp ;
ð2:31Þ
the above system is transformed to its normal form d uðtÞ ¼ BuðtÞ þ N ðuðtÞ; aÞ. dt
ð2:32Þ
Then if we replace (2.31) and (2.32) in Eq. (2.30) we obtain the following formula: DT ðu; aÞBu BT ðu; aÞ ¼ H ðu þ T ðu; aÞ; aÞ DT ðu; aÞN ðu; aÞ N ðu; aÞ.
ð2:33Þ
In view of the regularity of h, we restrict our attention to the terms of degree lower than m for each m P 2.
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
229
Let k X X
T ðu; aÞ ¼
q
q
q
T jðq;lÞ u11 u22 u33 al ;
N ðu; aÞ ¼
j¼2 ðq;lÞ2Dj
T j;1 ðq;lÞ
1
B j;2 C C T jðq;lÞ ¼ B @ T ðq;lÞ A; T j;3 ðq;lÞ
q
q
q
N jðq;lÞ u11 u22 u33 al ;
j¼2 ðq;lÞ2Dj
H ðu þ T ðu; aÞ; aÞ DT ðu; aÞN ðu; aÞ ¼ 0
k X X
0
N j;1 ðq;lÞ
X
k X
1
q
j¼2 ðq;lÞ2Dj
B j;2 C C N jðq;lÞ ¼ B @ N ðq;lÞ A and N j;3 ðq;lÞ
q
q
C jðq;lÞ u11 u22 u33 al ; 0
C j;1 ðq;lÞ
1
B j;2 C C C jðq;lÞ ¼ B @ C ðq;lÞ A. C j;3 ðq;lÞ
Then from the fact that DT ðu; aÞBu ¼ ix
k X X oT ðu; aÞ oT ðu; aÞ q q q u1 þ ix u2 ¼ ixðq1 q2 ÞT jðq;lÞ u11 u22 u33 al ou1 ou2 j¼2 ðq;lÞ2D j
and
0
k P
P
q1 q2 q3 l T j;1 ðq;lÞ u1 u2 u3 a
B ix B j¼2 ðq;lÞ2Dj B B k BT ðu; aÞ ¼ B P P j;2 q1 q2 q3 l B ix T ðq;lÞ u1 u2 u3 a B j¼2 ðq;lÞ2Dj @ 0
1 C C C C C; C C A
the coefficients of the nonlinear transformation and the normal form, satisfy the following system for (q1, q2, q3, l) 2 Dj: j;1 j;1 ixðq1 q2 þ 1ÞT j;1 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ N ðq1 ;q2 ;q3 ;lÞ ; j;2 j;2 ixðq1 q2 ÞT j;2 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ N ðq1 ;q2 ;q3 ;lÞ ;
and j;3 j;3 ixðq1 q2 1ÞT j;3 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ N ðq1 ;q2 ;q3 ;lÞ .
Remark 2.5. Note that for a fixed k P 2, the coefficients C kðq1 ;q2 ;q3 ;lÞ ; ðq1 ; q2 ; q3 ; lÞ 2 Dk are given in terms of T jðq ;q ;q ;lÞ ; N jðq ;q ;q ;lÞ ; j 2 f2 . . . k 1g, because of T(0, 0) = DT(0, 0) = N(0, 0) = DN(0, 0) = 0. 1
2
3
1
2
3
Finally, the algorithm of computation of normal forms is the following: Being computed N jðq1 ;q2 ;q3 ;lÞ ; T jðq1 ;q2 ;q3 ;lÞ for all j 2 {2 . . . , k 1} and (q1, q2, q3, l) 2 Dj then N kðq1 ;q2 ;q3 ;lÞ ; T kðq1 ;q2 ;q3 ;lÞ are given by For (q1 q2 + 1)(q1 q2)(q1 q2 1) 5 0 then we can choose k;2 k;3 N k;1 ðq1 ;q2 ;q3 ;lÞ ¼ N ðq1 ;q2 ;q3 ;lÞ ¼ N ðq1 ;q2 ;q3 ;lÞ ¼ 0;
T k;1 ðq1 ;q2 ;q3 ;lÞ ¼
C k;1 ðq1 ;q2 ;q3 ;lÞ ixðq1 q2 þ 1Þ
;
T k;2 ðq1 ;q2 ;q3 ;lÞ ¼
C k;2 ðq1 ;q2 ;q3 ;lÞ ixðq1 q2 Þ
;
T k;3 ðq1 ;q2 ;q3 ;lÞ ¼
For q1 q2 + 1 = 0 then k;1 N k;1 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ ;
T k;1 ðq1 ;q2 ;q3 ;lÞ ¼ 0;
N k;2 ðq1 ;q2 ;q3 ;lÞ ¼ 0;
T k;2 ðq1 ;q2 ;q3 ;lÞ ¼
C k;2 ðq1 ;q2 ;q3 ;lÞ ix
;
N k;3 ðq1 ;q2 ;q3 ;lÞ ¼ 0; T k;3 ðq1 ;q2 ;q3 ;lÞ ¼
C k;3 ðq1 ;q2 ;q3 ;lÞ 2ix
.
C k;3 ðq1 ;q2 ;q3 ;lÞ ixðq1 q2 1Þ
230
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
For q1 q2 = 0 then N k;1 ðq1 ;q2 ;q3 ;lÞ ¼ 0; T k;1 ðq1 ;q2 ;q3 ;lÞ ¼
k;2 N k;2 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ ;
C k;1 ðq1 ;q2 ;q3 ;lÞ ix
;
T k;2 ðq1 ;q2 ;q3 ;lÞ ¼ 0;
N k;3 ðq1 ;q2 ;q3 ;lÞ ¼ 0; T k;3 ðq1 ;q2 ;q3 ;lÞ ¼
C k;3 ðq1 ;q2 ;q3 ;lÞ ix
.
For q1 q2 1 = 0 then N k;1 ðq1 ;q2 ;q3 ;lÞ ¼ 0; T k;1 ðq1 ;q2 ;q3 ;lÞ ¼
N k;2 ðq1 ;q2 ;q3 ;lÞ ¼ 0;
C k;1 ðq1 ;q2 ;q3 ;lÞ 2ix
;
k;3 N k;3 ðq1 ;q2 ;q3 ;lÞ ¼ C ðq1 ;q2 ;q3 ;lÞ ;
T k;2 ðq1 ;q2 ;q3 ;lÞ ¼
C k;2 ðq1 ;q2 ;q3 ;lÞ ix
;
T k;3 ðq1 ;q2 ;q3 ;lÞ ¼ 0.
3. Outline of symbolic computer programs All the formulas presented in the previous section are given explicitly in terms of the coefficients of the original differential equations, and thus can be easily implemented on a symbolic computation system. The symbolic manipulation language Maple has been used to code these explicit formulas. In this section, we shall outline the computer programs. 3.1. Create the input file (a) Set and define the variables: n the dimension of the system. m the number of delays. r[l], l = 1, . . . , m the delays. omega the positive imaginary part of the purely imaginary eigenvalues. phi[i, j] (i = 1, 2, 3; j = 1, . . . , n) the jth component of the ith element of the basis U, of the generalized subspace associated with the critical eigenvalues. m0 the order of the center manifold to be computed. m1 the order of the normal forms to be computed. (b) Create the vector field of the original system: In order to create the vector field of the system, we must consider the parameters al, l = 1, . . . , p as a new dependent variables as follows: We put x[1, n + l] alpha[l],l = 1, . . . , p, where x[i, j] note the jth component of the dependent variables associated with the delay r[i]. (see the examples for more clarification). LP[l, i, j] the (i, j) th coefficient of the linear part of the vector field associated with the delay r[l] of the system. NL[i] the ith component of the nonlinearity of the vector field of the system. 3.2. Compute the center manifold and normal form (a) Compute the basis W. (b) Compute recursively the coefficients hiðq1 ;q2 ;q3 ;lÞ ; q1 þ q2 þ q3 þ l ¼ i; i ¼ 2; . . . ; j of a center manifolds: Pj1 • Compute the homogeneous part Hj of F ðUn þ i¼2 hi ðn; aÞ; aÞ. • Compute the coefficients of the nonlinear terms Nj1(n,a)(h) and Rj1(n, a). • Compute the initial data hjðq1 ;q2 ;q3 ;lÞ ð0Þ by formulas (2.20)–(2.23). • Compute the vector hjðq1 ;q2 ;q3 ;lÞ ðhÞ by formulas (2.15)–(2.19). • Construct the jth homogeneous part hj(n, a)(h) of a center manifolds.
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
231
(c) Compute the normal forms N associated up to the desired order. (d) Write the reduced equation on center manifold and the normal form N in the output file.
4. Examples In this section, we shall apply the results presented in Section 2 and the Maple programs we developed to compute the center manifolds and normal forms for three examples. All the results including the center manifolds, normal forms obtained by executing the Maple programs on a PC (Pentium 2—400 MHz) are given in the appendices. We describe also in these appendices how to create the input file of each example. 4.1. Example 1 Consider the scalar RFDE x_ ðtÞ ¼ axðtÞ bxðt 1Þ cxðt rÞ þ f ðxðtÞ; xðt 1Þ; xðt rÞÞ; where f 2 C 1 ðR3 ; RÞ and f(0,0,0) = Df(0,0,0) = 0. and assume the characteristic equation k ¼ a bek cekr has only the eigenvalues ±ix, x > 0, and zero on the imaginary axis. This latter hypothesis is equivalent to a þ b þ c ¼ 0; b sinðxÞ þ c sinðxrÞ ¼ x; a þ b cosðxÞ þ c cosðxrÞ ¼ 0. If we suppose r, x such that sin((r 1)x) 5 0, then given r, x, these later equations uniquely determine the coefficients a0, b0, c0. If we introduce the small parameters a1, a2 and a3 by a1 ¼ a þ a0 ; a2 ¼ b þ b0 and a3 ¼ c þ c0 ; then the RFDE can be reads as x_ ðtÞ ¼ a0 xðtÞ b0 xðt 1Þ c0 xðt rÞ þ a1 xðtÞ þ a2 xðt 1Þ þ a3 xðt rÞ þ f ðxðtÞ; xðt 1Þ; xðt rÞÞ. Here, we consider the case x :¼ p2 and r:¼2. which give a0 ¼ p4 ; b0 ¼ p2 and c0 ¼ p4. The quadratic term is considered to be x2 ðtÞ xðtÞxðt 1Þ þ 2xðt 1Þxðt 2Þ. Then executing the Maple program developed in this paper yields the reduced equation on center manifold and the associated normal form up to third order in Appendix A. 4.2. Examples 2 Consider the following model for a network of two neurons with self-connection u_ 1 ðtÞ ¼ u1 ðtÞ þ a11 f ðu1 ðt sÞÞ þ a12 f ðu2 ðt sÞÞ; u_ 2 ðtÞ ¼ u2 ðtÞ þ a21 f ðu1 ðt sÞÞ þ a22 f ðu2 ðt sÞÞ;
ð4:1Þ
where aij, i, j = 1, 2 are real constants, the delay s is positive and f 2 C 3 ðR; RÞ such that f(0) = 0. The particular case f(u) = tan gh, has been studied by many researchers (see [11,7]). The above mentioned authors investigated the linearized stability and delay induced oscillations. In [4], Faria studied the local bifurcation of system (4.1) under the assumptions f00 (0) = 0 and f000 (0) 5 0. Thus by deriving the normal form up to second
232
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
order of the system. Here, we will consider the system (4.1) with the unique assumption f(0) = 0. We will derive the reduced equation on the center manifolds and the normal form associated to the equation. Scaling the time variable, t ! t/s and separating the linear from the nonlinear terms, (4.1) becomes _ ¼ sLut þ sF ðut Þ; uðtÞ
ð4:2Þ
where ut 2 C :¼ Cð½1; 0; R2 Þ, and L : C ! R2 ; F : C ! R2 are given by L/ ¼
/1 ð0Þ þ a11 f 0 ð0Þ/1 ð1Þ þ a12 f 0 ð0Þ/2 ð1Þ
!
/2 ð0Þ þ a21 f 0 ð0Þ/1 ð1Þ þ a22 f 0 ð0Þ/2 ð1Þ ! a11 gð/1 ð1ÞÞ þ a12 gð/2 ð1ÞÞ F ð/Þ ¼ ; a21 gð/1 ð1ÞÞ þ a22 gð/2 ð1ÞÞ
for / ¼ ð/1 ; /2 Þ and g(x) = f(x) f 0 (0)x. The characteristic equation for the linearization of Eq. (4.2) at (0, 0) is 2
Dðk; sÞ ¼ ðk þ sÞ sT ðk þ sÞek þ s2 De2k ¼ 0; where 1 T ¼ ða11 þ a22 Þf 0 ð0Þ; 2
2
D ¼ ða11 a22 a12 a21 Þf 0 ð0Þ .
The following result, which was proved in [4], gives the necessary and sufficient conditions for existence of Fold-Hopf singularity. pffiffiffiffiffiffiffiffiffiffiffiffi Theorem 4.1. Assume D > T2, D > 1 and let q ¼ D 1. Then for r > 0,s > 0, D(ir, s) = 0 if and only if there is an n 2 N , such that s = sn and r = rn, where sn,rn are defined by rn sn ¼ ; q
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi T þ q D T2 cos rn ¼ ; D
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi D T 2 qT sin rn ¼ D
and rn 2 ð2np; 2ðn þ 1ÞpÞ. At s ¼ sn ; n 2 N , the eigenvalues ±irn are simple. Moreover, k = 0 is a simple eigenvalue of the linearization of Eq. (4.2) at ð 0; 0 Þ if and only if D = 2T 1. The basis U can be chosen as U ¼ ð/1 ; /2 ; /3 Þ; /1 ðhÞ ¼ eirn h u; /2 ðhÞ ¼ /1 ðhÞ; /3 ðhÞ ¼ v 0 f ð0Þa12 c1 with u ¼ 1; f 0 ð0Þa , v ¼ 1; and cj :¼ f 0 (0)ajj 1, j = 1, 2. c2 12 For a fixed n 2 N , introduce the new parameter a1 by a1 = s sn, then the RFDE (4.2) is written as _ ¼ sn Lut þ a1 Lut þ ða1 þ sn ÞF ðut Þ. uðtÞ pffiffiffi 3. The nonlinear terms are considered to be as ! 8/21 ð1Þ 6/22 ð1Þ þ 2/31 ð1Þ 32 /32 ð1Þ . 4/21 ð1Þ 2/22 ð1Þ þ /31 ð1Þ 12 /32 ð1Þ
pffiffi, q ¼ We choose D = 2, which give rn ¼ 5p , sn ¼ 35p 3 3
F ð/Þ ¼
Again we execute the Maple program developed in this paper to find the reduced equation on the center manifolds and the associated normal form, up to fourth-order, given in Appendix B.
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
233
4.3. Example 3 Consider the 3-dimensional RFDE given by 8 2 2 2 > < x_ ðtÞ ¼ axðt 1Þ þ 2x ðt 1Þ þ 3zðtÞzðt 1Þ þ y ðt 1Þ þ 5z ðtÞ; _ ¼ yðt 1Þ þ x2 ðt 1Þ þ y 2 ðt 1Þ yðtÞxðt 1Þ; yðtÞ > : z_ ðtÞ ¼ bzðtÞ þ x2 ðtÞ y 2 ðtÞ þ 2z2 ðtÞ; where a, b are a real parameters. The characteristic equation of the linear part is given by ðk bÞðk aek Þðk ek Þ ¼ 0. It has ±ix, x > 0, as purely imaginary eigenvalues and simple zero eigenvalue if and only if ak ¼ p2 þ 2kp and xk ¼ p2 þ 2kp. We choose as critical parameters a0 ¼ p2 and x0 ¼ p2. If we introduce the small parameters a1 and a2 by a1 ¼ a þ
p 2
and
a2 ¼ b;
then the RFDE can be reads as 8 p 2 2 2 > < x_ ðtÞ ¼ 2 xðt 1Þ þ a1 xðt 1Þ þ 2x ðt 1Þ þ 3zðtÞzðt 1Þ þ y ðt 1Þ þ 5z ðtÞ; _ ¼ yðt 1Þ þ x2 ðt 1Þ þ y 2 ðt 1Þ yðtÞxðt 1Þ; yðtÞ > : z_ ðtÞ ¼ a2 zðtÞ þ x2 ðtÞ y 2 ðtÞ þ 2z2 ðtÞ. Executing the Maple program developed in this paper yields he results in Appendix C. 5. Conclusions Methodology and computer programs have been developed for computing explicit canter manifolds and normal forms for Fold-Hopf singularity. The calculations and formulas are given in a explicit iterative procedure, and thus are very easy to be implemented on a symbolic computation system. Symbolic computer programs written in Maple have been developed for automating the computations. It has been shown that by several examples that the method is computationally efficient and fast, particularly suitable for the computations of high-dimensional systems and higher-order center manifolds and normal forms. Appendix A The input data file of Example 1: n :¼ 1: m :¼ 3: r[1] :¼ 0: r[2] :¼ 1: r[3] :¼ 2: omega :¼ Pi/2: phi[1, 1] :¼ exp(I*omega*theta): phi[2, 1] :¼ exp(I*omega*theta): phi[3, 1] :¼ 1: LP[1, 1, 1] :¼ Pi/4: LP[2, 1, 1] :¼ Pi/2: LP[3, 1, 1] :¼ Pi/4: NL[1] :¼ x[1, 2]*x[1, 1]+x[1, 3]*x[2, 1]+x[1,4]*x[3, 1]+x[1, 1]**2x[1, 1]*x[2, 1]+2*x[2, 1]*x[3, 1]: m0 :¼ 2: The reduced equation on a center manifold up to the third-order is
234
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
1 dn1=dt ¼ pn2 2ðð4l2p 8l2p2 4l2p4 þ 8l2p3 þ 8l1l3 12l12 2 þ 8l2l3 4l22 8l2l1 þ 4l32 þ 12l1pl2 þ 10l1p3 l2 2l1p4 l2 14l1p2 l2 4l1l3p3 þ 14l1l3p2 12l1l3p 2l2l3p3 þ 2l2l3p2 4l2l3p þ p5 l2 þ 8l12 p3 l12 p4 21l12 p2 þ 22l12 p þ 6pl22 p2 l22 4l32 p3 þ l32 p4 þ 7l32 p2 10l32 pÞn1Þ=ð2 2p þ p2 Þ3 ðð16l1p3 þ 8l1p4 þ 16l1p2 þ 8l3p þ 16l3p3 16l3p2 8l3p4 8l1p 16l2l3 16l22 16l2l1 þ 2l1p4 l3 2l2p4 l3 2p5 l1 þ 2p5 l3 þ 16l22 p3 3l22 p4 þ 32l1pl2 þ 16l1p3 l2 2l1p4 l2 40l1p2 l2 8l1l3p3 þ 24l1l3p2 24l1l3p þ 8l2l3p3 16l2l3p2 þ 32l2l3p 4l12 p3 þ l12 p4 þ 4l12 p þ 28pl22 1 3 28p2 l22 þ 12l32 p3 3l32 p4 24l32 p2 þ 20l32 pÞn2Þ=ð2 2p þ p2 Þ þ ððl1p4 þ l2p4 2 þ l3p4 þ 4p3 10l3p3 6l2p3 2l1p3 þ 10l2p2 8p2 þ 2l1p2 þ 18l3p2 þ 8l2p þ 8p 1 16l1p 32l2Þðl1 þ l2 þ l3Þn3Þ=ð2 2p þ p2 Þ2 ðð3l1p6 þ 3p6 l2 þ 3l3p6 12p5 l1 12 36p5 l3 24p5 l2 þ 72l2p4 þ 36l1p4 þ 84l3p4 þ 56l1p3 200l3p3 þ 40l2p3 þ 532l3p2 244l2p2 508l1p2 496l3p þ 480l2p þ 592l1p 496l2 528l1 þ 432l3Þn12 Þ=ð2 2p þ p2 Þ 2 þ ðð9p5 36p4 6l3p4 6l1p4 þ 6l2p4 þ 72p3 þ 62l1p3 14l3p3 40l2p3 þ 20l3p2 3 þ 116l2p2 72p2 116l1p2 24l3p þ 36p 68l2p þ 120l1p þ 56l3 1 3 þ 24l2 104l1Þn1n2Þ=ð2 2p þ p2 Þ ðð48l3 þ 496l2 þ 144l1 96p4 þ 96p þ 192p3 12 192p2 1056l2p þ 1324l2p2 þ 24p5 488l1p3 þ 108l1p4 þ 1060l1p2 þ 496l3p þ 152l3p3 652l3p2 þ 60l3p4 784l1p þ 192l2p4 808l2p3 12p5 l1 36p5 l3 24p5 l2 þ 3p6 l2 1 3 þ 3l3p6 þ 3l1p6 Þn22 Þ=ð2 2p þ p2 Þ ððl1p6 þ p6 l2 þ l3p6 20p5 l3 20p5 l1 þ 4p5 2 5 4 4 4 20p l2 þ 48l2p þ 40l1p 16p þ 104l3p4 þ 32p3 272l3p3 þ 80l1p3 þ 436l3p2 108l2p2 300l1p2 32p2 þ 16p þ 160l2p 448l3p þ 416l1p 192l1 1 þ 256l3Þn1n3Þ=ð2 2p þ p2 Þ3 ððl1p6 þ p6 l2 þ l3p6 þ 4p5 þ 16l1p4 2 48l2p4 16p4 48l3p4 þ 264l2p3 40l1p3 þ 200l3p3 þ 32p3 636l2p2 428l3p2 44l1p2 32p2 þ 160l1p þ 480l3p þ 800l2p þ 16p 224l1 3
480l2 224l3Þn2n3Þ=ð2 2p þ p2 Þ þ 2ðð3l1p4 þ 3l2p4 þ 3l3p4 23l3p3 þ 4p3 19l2p3 15l1p3 8p2 þ 40l3p2 þ 32l2p2 þ 24l1p2 þ 8p 26l1p 2l2p 10l3p 8l3 2
8l1 40l2Þn32 Þ=ð2 2p þ p2 Þ þ
1 ðp4 18p3 þ 30p2 56p þ 184Þn13 2 4 ð2 2p þ p2 Þ
þ
1 ð3p4 þ 6p3 126p2 þ 1264p 1368Þn12 n2 1 ð3p4 54p3 þ 18p2 þ 232p 1064Þn1n22 þ 2 2 12 12 ð2 2p þ p2 Þ ð2 2p þ p2 Þ
þ
1 ð3p4 þ 6p3 þ 138p2 736p þ 504Þn23 12 ð2 2p þ p2 Þ2
1 ð6p6 51p5 þ 129p4 148p3 þ 90p2 þ 68p 88Þn12 n3 3 ð2 2p þ p2 Þ3
3
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
4 ð18p4 45p3 þ 76p2 98p þ 32Þn1n2n3 3 3 ð2 2p þ p2 Þ
1 ð6p6 51p5 þ 189p4 416p3 þ 474p2 356p þ 184Þn22 n3 3 3 ð2 2p þ p2 Þ
2 2
235
ðp6 20p5 þ 73p4 106p3 þ 92p2 48p þ 56Þn1n32 ð2 2p þ p2 Þ
3
ðp6 24p4 þ 118p3 310p2 þ 400p 256Þn2n32 ð2 2p þ p2 Þ
3
þ8
ð2p4 13p3 þ 22p2 10p 8Þn33 ð2 2p þ p2 Þ
2
;
1 dn2=dt ¼ pn1 þ 2ðð4l2p 8l2p2 4l2p4 þ 8l2p3 þ 8l1l3 12l12 þ 8l2l3 2 4l22 8l2l1 þ 4l32 þ 12l1pl2 þ 10l1p3 l2 2l1p4 l2 14l1p2 l2 4l1l3p3 þ 14l1l3p2 12l1l3p 2l2l3p3 þ 2l2l3p2 4l2l3p þ p5 l2 þ 8l12 p3 l12 p4 21l12 p2 3
þ 22l12 p þ 6pl22 p2 l22 4l32 p3 þ l32 p4 þ 7l32 p2 10l32 pÞð2 þ pÞn1Þ=ðð2 2p þ p2 Þ pÞ þ ðð16l1p3 þ 8l1p4 þ 16l1p2 þ 8l3p þ 16l3p3 16l3p2 8l3p4 8l1p 16l2l3 16l22 16l2l1 þ 2l1p4 l3 2l2p4 l3 2p5 l1 þ 2p5 l3 þ 16l22 p3 3l22 p4 þ 32l1pl2 þ 16l1p3 l2 2l1p4 l2 40l1p2 l2 8l1l3p3 þ 24l1l3p2 24l1l3p þ 8l2l3p3 16l2l3p2 þ 32l2l3p 4l12 p3 þ l12 p4 þ 4l12 p þ 28pl22 28p2 l22 þ 12l32 p3 3l32 p4 24l32 p2 þ 20l32 pÞ 1 3 ð2 þ pÞn2Þ=ðð2 2p þ p2 Þ pÞ ððl1p4 þ l2p4 þ l3p4 þ 4p3 10l3p3 6l2p3 2l1p3 2 þ 10l2p2 8p2 þ 2l1p2 þ 18l3p2 þ 8l2p þ 8p 16l1p 32l2Þð2 þ pÞðl1 þ l2 1 2 þ l3Þn3Þ=ðpð2 2p þ p2 Þ Þ þ ðð3l1p6 þ 3p6 l2 þ 3l3p6 12p5 l1 36p5 l3 24p5 l2 12 þ 72l2p4 þ 36l1p4 þ 84l3p4 þ 56l1p3 200l3p3 þ 40l2p3 þ 532l3p2 244l2p2 508l1p2 496l3p þ 480l2p þ 592l1p 496l2 528l1 þ 432l3Þð2 þ pÞn12 Þ=ðð2 2p þ p2 Þ3 pÞ 2 ðð9p5 36p4 6l3p4 6l1p4 þ 6l2p4 þ 72p3 þ 62l1p3 14l3p3 40l2p3 þ 20l3p2 3 þ 116l2p2 72p2 116l1p2 24l3p þ 36p 68l2p þ 120l1p þ 56l3 þ 24l2 104l1Þ 1 ð2 þ pÞn1n2Þ=ðð2 2p þ p2 Þ3 pÞ þ ðð48l3 þ 496l2 þ 144l1 96p4 þ 96p þ 192p3 192p2 12 1056l2p þ 1324l2p2 þ 24p5 488l1p3 þ 108l1p4 þ 1060l1p2 þ 496l3p þ 152l3p3 652l3p2 þ 60l3p4 784l1p þ 192l2p4 808l2p3 12p5 l1 36p5 l3 24p5 l2 þ 3p6 l2 þ 3l3p6 1 þ 3l1p6 Þð2 þ pÞn22 Þ=ðpð2 2p þ p2 Þ3 Þ þ ððl1p6 þ p6 l2 þ l3p6 20p5 l3 20p5 l1 2 þ 4p5 20p5 l2 þ 48l2p4 þ 40l1p4 16p4 þ 104l3p4 þ 32p3 272l3p3 þ 80l1p3 þ 436l3p2 108l2p2 300l1p2 32p2 þ 16p þ 160l2p 448l3p þ 416l1p 192l1 þ 256l3Þð2 1 3 þ pÞn1n3Þ=ðð2 2p þ p2 Þ pÞ þ ððl1p6 þ p6 l2 þ l3p6 þ 4p5 þ 16l1p4 48l2p4 16p4 2 48l3p4 þ 264l2p3 40l1p3 þ 200l3p3 þ 32p3 636l2p2 428l3p2 44l1p2 32p2 3
þ 160l1p þ 480l3p þ 800l2p þ 16p 224l1 480l2 224l3Þð2 þ pÞn2n3Þ=ðð2 2p þ p2 Þ pÞ 2ðð3l1p4 þ 3l2p4 þ 3l3p4 23l3p3 þ 4p3 19l2p3 15l1p3 8p2 þ 40l3p2 þ 32l2p2 þ 24l1p2 þ 8p 26l1p 2l2p 10l3p 8l3 8l1 40l2Þ
236
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246 2
ð2 þ pÞn32 Þ=ðpð2 2p þ p2 Þ Þ
1 ðp4 18p3 þ 30p2 56p þ 184Þð2 þ pÞn13 2 4 pð2 2p þ p2 Þ
1 ð3p4 þ 6p3 126p2 þ 1264p 1368Þð2 þ pÞn12 n2 2 12 pð2 2p þ p2 Þ
1 ð3p4 54p3 þ 18p2 þ 232p 1064Þð2 þ pÞn1n22 2 12 pð2 2p þ p2 Þ
1 ð3p4 þ 6p3 þ 138p2 736p þ 504Þð2 þ pÞn23 2 12 pð2 2p þ p2 Þ
þ
1 ð6p6 51p5 þ 129p4 148p3 þ 90p2 þ 68p 88Þð2 þ pÞn12 n3 3 3 pð2 2p þ p2 Þ
þ
4 ð18p4 45p3 þ 76p2 98p þ 32Þð2 þ pÞn1n2n3 3 pð2 2p þ p2 Þ3
þ
1 ð6p6 51p5 þ 189p4 416p3 þ 474p2 356p þ 184Þð2 þ pÞn22 n3 3 pð2 2p þ p2 Þ3
þ2 þ2 8
ðp6 20p5 þ 73p4 106p3 þ 92p2 48p þ 56Þð2 þ pÞn1n32 ð2 2p þ p2 Þ3 p ðp6 24p4 þ 118p3 310p2 þ 400p 256Þð2 þ pÞn2n32 ð2 2p þ p2 Þ3 p ð2p4 13p3 þ 22p2 10p 8Þð2 þ pÞn33 pð2 2p þ p2 Þ2
dn3=dt ¼ ðð4l2p 8l2p2 4l2p4 þ 8l2p3 þ 8l1l3 12l12 þ 8l2l3 4l22 8l2l1 þ 4l32 þ 12l1pl2 þ 10l1p3 l2 2l1p4 l2 14l1p2 l2 4l1l3p3 þ 14l1l3p2 12l1l3p 2l2l3p3 þ 2l2l3p2 4l2l3p þ p5 l2 þ 8l12 p3 l12 p4 21l12 p2 þ 22l12 p 1 2 þ 6pl22 p2 l22 4l32 p3 þ l32 p4 þ 7l32 p2 10l32 pÞn1Þ=ðð2 2p þ p2 Þ pÞ ðð16l1p3 2 4 2 3 2 4 þ 8l1p þ 16l1p þ 8l3p þ 16l3p 16l3p 8l3p 8l1p 16l2l3 16l22 16l2l1 þ 2l1p4 l3 2l2p4 l3 2p5 l1 þ 2p5 l3 þ 16l22 p3 3l22 p4 þ 32l1pl2 þ 16l1p3 l2 2l1p4 l2 40l1p2 l2 8l1l3p3 þ 24l1l3p2 24l1l3p þ 8l2l3p3 16l2l3p2 þ 32l2l3p 4l12 p3 þ l12 p4 þ 4l12 p þ 28pl22 28p2 l22 þ 12l32 p3 3l32 p4 24l32 p2 þ 20l32 pÞn2Þ=ðpð2 2p 1 þ p2 Þ2 Þ þ ðð8l2p 8l2p2 þ 4l1p3 8l1p2 þ 8l3p þ 4l3p3 8l3p2 þ 8l1p þ 4l2p3 32l2l3 4 32l22 32l2l1 þ 2l1p4 l3 þ 2l2p4 l3 6l22 p3 þ l22 p4 8l1pl2 8l1p3 l2 þ 2l1p4 l2 þ 12l1p2 l2 12l1l3p3 þ 20l1l3p2 16l1l3p 16l2l3p3 þ 28l2l3p2 þ 8l2l3p 2l12 p3 þ l12 p4 þ 2l12 p2 16l12 p þ 8pl22 þ 10p2 l22 10l32 p3 þ l32 p4 þ 18l32 p2 Þn3Þ=ðpð2 2p 1 ðð3l1p6 þ 3p6 l2 þ 3l3p6 12p5 l1 36p5 l3 24p5 l2 þ 72l2p4 þ 36l1p4 þ 84l3p4 24 þ 56l1p3 200l3p3 þ 40l2p3 þ 532l3p2 244l2p2 508l1p2 496l3p þ 480l2p þ 592l1p þ p2 ÞÞ
1 2 496l2 528l1 þ 432l3Þn12 Þ=ðð2 2p þ p2 Þ pÞ þ ðð9p5 36p4 6l3p4 6l1p4 þ 6l2p4 3 þ 72p3 þ 62l1p3 14l3p3 40l2p3 þ 20l3p2 þ 116l2p2 72p2 116l1p2 24l3p þ 36p
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
1 ðð48l3 þ 496l2 þ 144l1 24 96p4 þ 96p þ 192p3 192p2 1056l2p þ 1324l2p2 þ 24p5 488l1p3 þ 108l1p4 þ 1060l1p2 68l2p þ 120l1p þ 56l3 þ 24l2 104l1Þn1n2Þ=ðpð2 2p þ p2 Þ2 Þ
þ 496l3p þ 152l3p3 652l3p2 þ 60l3p4 784l1p þ 192l2p4 808l2p3 12p5 l1 36p5 l3 1 24p5 l2 þ 3p6 l2 þ 3l3p6 þ 3l1p6 Þn22 Þ=ðpð2 2p þ p2 Þ2 Þ ððl1p6 þ p6 l2 þ l3p6 20p5 l3 4 20p5 l1 þ 4p5 20p5 l2 þ 48l2p4 þ 40l1p4 16p4 þ 104l3p4 þ 32p3 272l3p3 þ 80l1p3 þ 436l3p2 108l2p2 300l1p2 32p2 þ 16p þ 160l2p 448l3p þ 416l1p 192l1 1 þ 256l3Þn1n3Þ=ðpð2 2p þ p2 Þ2 Þ ððl1p6 þ p6 l2 þ l3p6 þ 4p5 þ 16l1p4 48l2p4 16p4 4 4 3 3 48l3p þ 264l2p 40l1p þ 200l3p3 þ 32p3 636l2p2 428l3p2 44l1p2 32p2 þ 160l1p 2
þ 480l3p þ 800l2p þ 16p 224l1 480l2 224l3Þn2n3Þ=ðpð2 2p þ p2 Þ Þ þ ðð3l1p4 þ 3l2p4 þ 3l3p4 23l3p3 þ 4p3 19l2p3 15l1p3 8p2 þ 40l3p2 þ 32l2p2 þ 24l1p2 þ 8p 26l1p 2l2p 10l3p 8l3 8l1 40l2Þn32 Þ=ðpð2 2p þ p2 ÞÞ þ
1 ðp4 18p3 þ 30p2 56p þ 184Þn13 1 ð3p4 þ 6p3 126p2 þ 1264p 1368Þn12 n2 þ 8 24 pð2 2p þ p2 Þ pð2 2p þ p2 Þ
þ
1 ð3p4 54p3 þ 18p2 þ 232p 1064Þn1n22 24 pð2 2p þ p2 Þ
þ
1 ð3p4 þ 6p3 þ 138p2 736p þ 504Þn23 24 pð2 2p þ p2 Þ
1 ð6p6 51p5 þ 129p4 148p3 þ 90p2 þ 68p 88Þn12 n3 2 6 pð2 2p þ p2 Þ
2 ð18p4 45p3 þ 76p2 98p þ 32Þn1n2n3 2 3 pð2 2p þ p2 Þ
1 ð6p6 51p5 þ 189p4 416p3 þ 474p2 356p þ 184Þn22 n3 2 6 ð2 2p þ p2 Þ p
ðp6 20p5 þ 73p4 106p3 þ 92p2 48p þ 56Þn1n32 pð2 2p þ p2 Þ
2
ðp6 24p4 þ 118p3 310p2 þ 400p 256Þn2n32
þ4
pð2 2p þ p2 Þ
2
ð2p4 13p3 þ 22p2 10p 8Þn33 ð2 2p þ p2 Þp.
The normal form in polar coordinate: order = 3 3 3 5 dr=dt ¼ ððl1l3p5 l2l3p5 þ l12 p5 l22 p5 l32 p5 þ l2l1p5 þ 16l2l3 2 2 2 þ 16l22 þ 16l2l1 2l1p4 l3 þ 8l2p4 l3 29l22 p3 þ 11l22 p4 32l1pl2 22l1p3 l2 þ 44l1p2 l2 þ 6l1l3p3 24l1l3p2 þ 16l1l3p 18l2l3p3 þ 36l2l3p2 48l2l3p þ 25l12 p3 11l12 p4 20l12 p2 þ 8l12 p 32pl22 þ 36p2 l22 31l32 p3 þ 13l32 p4 þ 44l32 p2 24l32 pÞrrÞ=ðp%1Þ þ ðð112l3 þ 240l2 þ 112l1 520l2p þ 478l2p2 þ 84l1p3 38l1p4 42l1p2 360l3p 316l3p3 þ 446l3p2
237
238
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
þ 142l3p4 88l1p þ 90l2p4 264l2p3 6p5 l1 38p5 l3 24p5 l2 9 9 9 þ p6 l2 þ l3p6 þ l1p6 ÞrrzÞ=ðp%1Þ 2 2 2 ð18p6 97p5 þ 272p4 638p3 þ 1068p2 1112p þ 512Þrrz2 þ p%1 ð 94 p4 þ 12 p3 þ 13p2 15p þ 3Þrr3 þ pð4 8p þ 8p2 4p3 þ p4 Þ %1 :¼ 8 24p þ 36p2 32p3 þ 18p4 6p5 þ p6 ; 1 1 3 1 dh=dt ¼ p þ ð16l1l3 þ l1l3p5 l2l3p5 l12 p5 l22 p5 l32 p5 2 2 2 2 2 2 2 5 3l2l1p þ 24l1 16l2l3 þ 8l2 þ 16l2l1 8l3 8l1p4 l3 þ 2l2p4 l3 15l22 p3 þ 8l22 p4 40l1pl2 54l1p3 l2 þ 22l1p4 l2 þ 56l1p2 l2 þ 34l1l3p3 52l1l3p2 þ 32l1l3p 2l2l3p3 þ 8l2l3p2 þ 8l2l3p 37l12 p3 þ 8l12 p4 þ 66l12 p2 56l12 p 24pl22 þ 22p2 l22 þ 3l32 p3 14l32 p2 þ 24l32 pÞ=ðp%1Þ þ ðð10l1p4 11 1 þ 24p5 l3 l3p6 þ l3p7 128l3 70l3p4 þ 232l3p þ 138l3p3 2 2 11 1 186l2p3 þ 10p5 l2 p6 l2 þ p7 l2 210l3p2 þ 294l1p2 þ 24p5 l1 2 2 11 1 6 2 4 l1p þ 294l2p þ 42l2p þ l1p7 þ 96l1 126l1p3 312l1p 2 2 200l2pÞzÞ=ðp%1Þ ð2p7 22p6 þ 89p5 134p4 6p3 þ 168p2 104p 112Þz2 p%1 1 5 9 4 21 3 56 2 ð 4 p þ 4 p 2 p þ 3 p 403 p þ 373Þrr2 %1 :¼ 8 24p þ 36p2 32p3 þ 18p4 6p5 þ p6 þ pð4 8p þ 8p2 4p3 þ p4 Þ 1 1 3 1 dz=dt ¼ 8l2l3 8l22 8l2l1 þ l1p4 l3 þ l2p4 l3 l22 p3 þ l22 p4 2 2 2 4 1 2l1pl2 2l1p3 l2 þ l1p4 l2 þ 3l1p2 l2 3l1l3p3 þ 5l1l3p2 2 1 1 4l1l3p 4l2l3p3 þ 7l2l3p2 þ 2l2l3p l12 p3 þ l12 p4 2 4 1 2 2 5 5 1 9 2 2 2 2 2 2 3 2 4 2 þ l1 p 4l1 p þ 2pl2 þ p l2 l3 p þ l3 p þ l3 p z =ðpð2 2p þ p2 ÞÞ 2 2 2 4 2 þ
þ ðð3l1p4 þ 3l2p4 þ 3l3p4 23l3p3 19l2p3 15l1p3 þ 24l1p2 þ 40l3p2 þ 32l2p2 10l3p 2l2p 26l1p 8l1 8l3 1 1 1 40l2Þz2 Þ=ðpð2 2p þ p2 ÞÞ þ l1p6 p6 l2 l3p6 þ p5 l2 8 8 8 3 5 1 5 11 þ p l3 þ p l1 3l1p4 l2p4 3l3p4 þ l3p3 þ 9l1p3 þ 16l2p3 2 2 2 5 45 23 2 2 2 þ l3p l2p l1p þ 4l1p þ 12l2p 8l3 þ 8l1 rr2 ðpð4 8p þ 8p2 4p3 þ p4 ÞÞ 2 2 2 6 17 5 53 4 p þ 2 p 2 p þ 47p3 47p2 þ 24p 8 rr2 z ð8p4 52p3 þ 88p2 40p 32Þz3 þ . þ pð4 8p þ 8p2 4p3 þ p4 Þ ð2 2p þ p2 Þp
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
Appendix B The input data file of Example 2: n :¼ 2: m :¼ 2: omega :¼ 5*Pi/3: b11 :¼ 4:b12 :¼ 3:b21 :¼ 2:b22 :¼ 1: phi[1, 1] :¼ exp(I*omega*theta): phi[1, 2] :¼ (b22 1)*exp(I*omega*theta)/b21: phi[2, 1] :¼ exp(I*omega*theta): phi[2, 2] :¼ (b22 1)*exp(I*omega*theta)/b21: phi[3, 1] :¼ 1:phi[3, 2] :¼ (1 b11)/b12: r[1] :¼ 0: r[2] :¼ 1: LP[1, 1, 1]:¼ 1:LP[1, 1, 2] :¼ 0:LP[1, 2, 1] :¼ 0:LP[1, 2, 2]:¼ 1:LP[2, 1, 1] :¼ 4:LP[2, 1, 2] :¼ 3:LP[2, 2, 1] :¼ 2:LP[2, 2, 2] :¼ 1: NL[1] :¼ x[1, 3]*x[1, 1] + 4*x[1, 3]*x[2, 1] 3*x[1, 3]*x[2, 2] + 8*(x[1, 3] + 5*Pi/ (3*sqrt(3)))*(x[2, 1]**2) 6*(x[1, 3] + (5*Pi/(3*sqrt(3))))*(x[2, 2]**2) + 2*(x[1, 3] + (5*Pi/ (3*sqrt(3))))*(x[2, 1]**3) 3*(x[1, 3] + (5*Pi/(3*sqrt(3))))*(x[2, 2]**3)/2: NL[2] :¼ x[1, 3]*x[1, 2] + 2*x[1, 3]*x[2, 1] x[1, 3]*x[2, 2] + 4*(x[1, 3] + (5*Pi/ (3*sqrt(3))))*(x[2, 1]**2) 2*(x[1, 3] + (5*Pi/(3*sqrt(3))))*(x[2, 2]**2) + (x[1, 3] + (5*Pi/ (3*sqrt(3))))*(x[2, 1]**3) (x[1, 3] + (5*Pi/(3*sqrt(3))))*(x[2, 2]**3)/2: m0 :¼ 2: The reduced equation on a center manifold up to the third order is pffiffiffi pffiffiffi 5 1 %2l1 15309l1 3 þ 1984500p3 1531250p6 3 dn1=dt ¼ pn2 3 8575 pffiffiffi pffiffiffi pffiffiffi pffiffiffi pffiffiffi 4 6 3 þ 953125l1p 3 198450p2 3 2756250p4 3 þ 591300l1p2 3 þ 2542500l1p
2723625l1p3 4021875l1p5 229635l1p þ 5512500p5 n1 =ð%13 p2 Þ pffiffiffi pffiffiffi pffiffiffi 2 3 %2l1 1701 3 24975p2 3 þ 17415p þ 52875p3 16875p4 3 þ 3125p5 n2 þ 3 1715 pffiffiffi pffiffiffi p%1 4 2 2 3 150 210p þ 189 þ 1325p 3 459p 3 2875p %2l1n1 þ 343 %33 p ffiffi ffi pffiffiffi pffiffiffi 4 237500p5 3 691875p4 þ 249375p3 3 90450p2 6615p 3 þ 5103 %2l1n1n2 þ 343 pffiffiffi pffiffiffi pffiffiffi 2 =ðp%13 Þ þ 134375p5 3 428125p4 þ 168000p3 3 71775p2 2835p 3 þ 3402 343 . pffiffiffi pffiffiffi 8 3%2l1n22 ðp%13 Þ %2 918750p5 3 15309l1 þ 99225p2 þ 1378125p4 5145 pffiffiffi pffiffiffi pffiffiffi 330750p3 3 þ 765625p6 þ 136080l1 3p þ 1694250l1p3 3 859375l1p6 1174500l1p2 . pffiffiffi pffiffiffi 8 %2 15309l1 3 þ 992250p3 4027500l1p4 þ 1687500l1p5 3 n1n3 ðp%33 Þ 15435 pffiffiffi pffiffiffi pffiffiffi pffiffiffi pffiffiffi 765625p6 3 99225p2 3 1378125p4 3 þ 684450l1p2 3 þ 1946250l1p4 3 . pffiffiffi þ 359375l1p6 3 2662875l1p3 2278125l1p5 280665l1p þ 2756250p5 n2n3 ðp%13 Þ pffiffiffi pffiffiffi pffiffiffi 5 %2p 621000p3 3 664200p2 þ 107730p 3 35721 990000p4 þ 200000p5 3 n13 252 25 . pffiffiffi pffiffiffi pffiffiffi %2p 93240p2 3 %3 10000p4 þ 9000p3 3 14400p2 þ 3510p 3 1701 .252 pffiffiffi pffiffiffi pffiffiffi %1 10000p4 9000p3 3 þ 53622p þ 228600p3 þ 40000p5 94000p4 3 1701 3 n12 n2 pffiffiffi pffiffiffi 5 %2p 174600p2 þ 55890p 3 þ 18711 350000p4 þ14400p2 3510p 3 þ 1701 252
239
240
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
. pffiffiffi pffiffiffi pffiffiffi pffiffiffi þ237000p3 3 þ 200000p5 3 n1n22 %3 10000p4 þ 9000p3 3 14400p2 þ 3510p 3 pffiffiffi pffiffiffi pffiffiffi 5 %2p 3 63000p3 3 þ 70200p2 þ 2430p 3 þ 15309 210000p4 1701 756 pffiffiffi 3 . pffiffiffi pffiffiffi 5 þ 200000p 3 n2 ð%1ð10000p4 9000p3 3 þ 14400p2 3510p 3 þ 1701ÞÞ
ÞÞ
pffiffiffi pffiffiffi pffiffiffi 40 ðð34375p5 3 þ 10206 384750p2 þ 25515p 3 618750p4 þ 464625p3 3Þ%2n12 n3Þ=%33 3087 pffiffiffi pffiffiffi pffiffiffi 80 þ ðð42525p þ 322875p3 þ 3402 3 þ 6750p2 3 405000p4 3 3087 pffiffiffi pffiffiffi 40 þ 465625p5 Þ%2n1n2n3Þ=%13 þ 19845p þ 370125p3 þ 3402 3 38250p2 3 3087 . pffiffiffi p ffiffi ffi 1 %2 3529575p 53679375p3 338750p4 3 þ 321875p5 %2 3n22 n3 %13 9261 pffiffiffi pffiffiffi pffiffiffi pffiffiffi þ163296 3 þ 17593750p6 3 þ 10870200p2 3 þ 50490000p4 3 pffiffiffi pffiffiffi 1 77971875p5 n1n32 =%13 %2 5196875p5 3 þ 2543400p2 513135p 3 3087 . pffiffiffi 4 3 þ8550000p 3036375p 3 þ 4531250p6 þ 163296 n2n32 %13 pffiffiffi %1 :¼ 25p2 15p 3 þ 9 pffiffiffi pffiffiffi %2 :¼ 45p þ 9 3 þ 25p2 3 pffiffiffi %3 :¼ 9 25p2 þ 15p 3; pffiffiffi pffiffiffi pffiffiffi 5 2 15309l1 3 þ 1984500p3 1531250p6 3 198450p2 3 dn2=dt ¼ pn1 3 8575 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 2756250p4 3 þ 591300l1p2 3 þ 2542500l1p4 3 þ 953125l1p6 3 2723625l1p3 4021875l1p5 . 229635l1p þ 5512500p5 l1n1 ðp2 %12 Þ pffiffiffi pffiffiffi pffiffiffi 52875p3 17415p þ 16875p4 3 þ 24975p2 3 3125p5 þ 1701 3 l12 n2 6 1715 p%22 pffiffiffi pffiffiffi 2 3 300 l1 210p þ 189 þ 1325p 3 459p 3 2875p4 n12 343 %22 pffiffiffi p ffiffi ffi pffiffiffi 8 l1 237500p5 3 691875p4 þ 249375p3 3 90450p2 6615p 3 þ 5103 n1n2 þ 343 %12 p pffiffiffi pffiffiffi p ffiffi ffi pffiffiffi 4 l1 3 134375p5 3 428125p4 þ 168000p3 3 71775p2 2835p 3 þ 3402 n22 þ 343 %12 p p ffiffi ffi pffiffiffi pffiffiffi 16 918750p5 3 þ 15309l1 99225p2 1378125p4 þ 330750p3 3 765625p6 136080l1 3p 5145 . pffiffiffi pffiffiffi 1694250l1p3 3 þ 859375l1p6 þ 1174500l1p2 þ 4027500l1p4 1687500l1p5 3 n1n3 ð%12 pÞ pffiffiffi pffiffiffi pffiffiffi pffiffiffi pffiffiffi 16 15309l1 3 þ 992250p3 765625p6 3 99225p2 3 1378125p4 3 þ 684450l1p2 3 15435 pffiffiffi pffiffiffi þ 1946250l1p4 3 þ 359375l1p6 3 2662875l1p3 2278125l1p5 280665l1p . þ 2756250p5 n2n3 ð%12 pÞ pffiffiffi pffiffiffi pffiffiffi 5 621000p3 3 664200p2 þ 107730p 3 35721 990000p4 þ 200000p5 3 pn13 pffiffiffi pffiffiffi þ 126 10000p4 þ 9000p3 3 14400p2 þ 3510p 3 1701 þ
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
pffiffiffi pffiffiffi pffiffiffi 25 93240p2 3 þ 1701 3 53622p 228600p3 þ 94000p4 3 40000p5 pn12 n2 pffiffiffi pffiffiffi 126 10000p4 þ 9000p3 3 14400p2 þ 3510p 3 1701 pffiffiffi pffiffiffi 5 174600p2 þ 55890p 3 þ 18711 350000p4 þ 237000p3 3 126 . pffiffiffi pffiffiffi pffiffiffi þ200000p5 3 pn1n22 10000p4 9000p3 3 þ 14400p2 3510p 3 þ 1701 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 5 63000p3 3 þ 70200p2 þ 2430p 3 þ 15309 210000p4 þ 200000p5 3 p 3n23 pffiffiffi pffiffiffi 378 10000p4 9000p3 3 þ 14400p2 3510p 3 þ 1701 pffiffiffi pffiffiffi pffiffiffi 80 34375p5 3 þ 10206 384750p2 þ 25515p 3 618750p4 þ 464625p3 3 n12 n3 2 3087 %1p p ffiffi ffi ffiffiffi pffiffiffi 160 42525p 322875p3 3402 3 6750p2 3 þ 405000p4 3 465625p5 n1n2n3 3087 %22 pffiffiffi p ffiffi ffi pffiffiffi pffiffiffi 80 19845p 370125p3 3402 3 þ 38250p2 3 þ 338750p4 3 321875p5 3n22 n3 3087 %22 pffiffiffi p ffiffiffi 2 3529575p 53679375p3 þ 163296 3 þ 17593750p6 3 9261 . pffiffiffi pffiffiffi þ10870200p2 3 þ 50490000p4 3 77971875p5 n1n32 %12 pffiffiffi pffiffiffi pffiffiffi 2 ðð 5196875p5 3 þ 2543400p2 513135p 3 þ 8550000p4 3036375p3 3 3087 þ4531250p6 þ 163296 n2n32 =%12 p ffiffi ffi %1 :¼ 25p2 15p 3 þ 9 pffiffiffi %2 :¼ 9 25p2 þ 15p 3; pffiffiffi pffiffiffi pffiffiffi 1 dn3=dt ¼ l1 30618l1 3 þ 1984500p3 1531250p6 3 198450p2 3 1960 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 2756250p4 3 þ 828225l1p2 3 þ 2171250l1p4 3 þ 1156250l1p6 3 2750625l1p3 pffiffiffi 1 3731250l1p5 433755l1p þ 5512500p5 n1=ð%12 p2 Þ þ l1 367500p4 3 þ 39690p 392 pffiffiffi pffiffiffi pffiffiffi þ551250p3 132300p2 3 þ 306250p5 þ 15309l1 3 þ 223425l1p2 3 þ 18750l1p5 pffiffiffi pffiffiffi 1 169695l1p 380250l1p3 þ 60000l1p4 3 n2=ðp%12 Þ 78121827l1 3 8820 pffiffiffi pffiffiffi þ8197969500p3 þ 144703125000p9 122232796875p6 3 148837500000p8 3 pffiffiffi pffiffiffi pffiffiffi 506345175p2 3 322776562500l1p8 3 22968750000p10 3 þ 228234375000l1p9 pffiffiffi pffiffiffi pffiffiffi þ 30156250000l1p10 3 þ 451967343750l1p7 21767484375p4 3 þ 34886149200l1p2 3 pffiffiffi pffiffiffi þ 254828328750l1p4 3 þ 60863906250l1p6 3 214691502375l1p3 . 485536781250l1p5 7861783860l1p þ 284651718750p7 þ 108502537500p5 n12 ðp%2%12 Þ pffiffiffi pffiffiffi 1 12055837500p5 3 þ 31627968750p7 3 þ 677055834l1 168781725p2 1470 pffiffiffi pffiffiffi 7656250000p10 þ 16078125000p9 3 7255828125p4 þ 910885500p3 3 953564062500l1p8 pffiffiffi pffiffiffi þ 2812500000l1p10 þ 168281250000l1 3p9 þ 867211875000l1 3p7 40744265625p6 pffiffiffi pffiffiffi 49612500000p8 2414612025l1 3p þ 38386679625l1p3 3 1426704890625l1p6 . pffiffiffi þ1951897500l1p2 333901591875l1p4 þ 502901662500l1p5 3 n1n2 ðp%2%12 Þ pffiffiffi pffiffiffi 1 þ 1328071059l1 3 þ 2732656500p3 þ 48234375000p9 40744265625p6 3 8820
241
242
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
pffiffiffi pffiffiffi pffiffiffi pffiffiffi 49612500000p8 3 168781725p2 3 1755435937500l1p8 3 7656250000p10 3 pffiffiffi pffiffiffi þ 878484375000l1p9 þ 5781250000l1p10 3 þ 4998041718750l1p7 7255828125p4 3 pffiffiffi pffiffiffi pffiffiffi 7949963700l1p2 3 757538527500l1p4 3 2865511687500l1p6 3 þ 304777312875l1p3 þ 3187292793750l1p5 11867077530l1p þ 94883906250p7 pffiffiffi pffiffiffi 2 . 1 þ þ 36167512500p5 n22 1837500p5 3 þ 45927l1 p%3 9 25p2 þ 15p 3 294 pffiffiffi pffiffiffi pffiffiffi 2 4 3 6 198450p 2756250p þ 661500p 3 1531250p 371385l1 3p 3550500l1p3 3 . pffiffiffi þ 1875000l1p6 þ 2818800l1p2 þ 7571250l1p4 3078125l1p5 3 n1n3 ðp%12 Þ pffiffiffi pffiffiffi 5 þ 15309l1 3 þ 396900p3 306250p6 3 882 pffiffiffi pffiffiffi pffiffiffi pffiffiffi pffiffiffi 39690p2 3 551250p4 3 þ 114210l1p2 3 þ 42750l1p4 3 þ 212500l1p6 3 48600l1p3 5 pffiffiffi 5 pffiffiffi 466875l1p5 137781l1p þ 1102500p5 n2n3 =ðp%12 Þ þ p 3 þ l1 l1 3p n32 9 18 pffiffiffi pffiffiffi pffiffiffi 3 2 4 25 p 621000p 3 664200p þ 107730p 3 35721 990000p þ 200000p5 3 n13 þ %2 288 pffiffiffi pffiffiffi pffiffiffi 2 3 93240p þ 94000p4 3 40000p5 pn12 n2 3 þ 1701 3 53622p 228600p 125 þ %3 288 . pffiffiffi pffiffiffi pffiffiffi 25 2 þ 174600p þ 55890p 3 þ 18711 350000p4 þ 237000p3 3 þ 200000p5 3 pn1n22 ð%2Þ 288 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 25 p 3 63000p3 3 þ 70200p2 þ 2430p 3 þ 15309 210000p4 þ 200000p5 3 n23 %3 864 p ffiffi ffi pffiffiffi 1 5677690190625p5 3 þ 5386021875000p7 3 þ 6666395904 345545562600p2 7056 pffiffiffi pffiffiffi þ 5740645365p 3 160625000000p10 þ 541968750000p9 3 5320406025000p4 pffiffiffi þ 1058529961125p3 3 11965985718750p6 4311759375000p8 n12 n3 . pffiffiffi 2 1 %3 9 25p2 þ 15p 3 139753335015p 497418114375p3 10584 pffiffiffi pffiffiffi pffiffiffi þ 7499695392 3 þ 4813781250000p9 11250786093750p6 3 8970215625000p8 3 pffiffiffi pffiffiffi pffiffiffi þ 285018369300p2 3 þ 110000000000p10 3 1204444687500p4 3 þ 22947840000000p7 þ 9627328153125p5 n1n2n3 =ð%2%12 Þ pffiffiffi 1 pffiffiffi 3 87428244645p þ 731775940875p3 þ 7221928896 3 þ 3542906250000p9 21168 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 11450488218750p6 3 7255696875000p8 3 þ 74530335600p2 3 þ 60625000000p10 3 pffiffiffi 2584885770000p4 3 þ 20482824375000p7 þ 12103053159375p5 n22 n3=ð%2%12 Þ pffiffiffi pffiffiffi pffiffiffi 25 þ p 950859 15196275p2 þ 3518750p5 3 þ 3405240p 3 þ 12078000p3 3 10584 pffiffiffi pffiffiffi 5 46875p5 3 þ 707400p2 830655p 3 15924375p4 n1n32 =%12 þ 3528 . p ffiffi ffi 25 2 5 pffiffiffi 2 2 4 3 6 %1 þ p þ p 3 n33 7710000p þ 2885625p 3 þ 326592 þ 4531250p n2n3 18 36 pffiffiffi 2 %1 :¼ 25p 15p 3 þ 9 pffiffiffi pffiffiffi %2 :¼ 10000p4 9000p3 3 þ 14400p2 3510p 3 þ 1701 pffiffiffi pffiffiffi %3 :¼ 10000p4 þ 9000p3 3 14400p2 þ 3510p 3 1701.
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
The normal form in polar coordinate: order = 3 pffiffiffi pffiffiffi 3 l12 1171875p5 3 þ 417150p2 59535p 3 þ 2013750p4 dr=dt ¼ 17150 pffiffiffi 4 l1ð357210p 658125p3 3 þ 15309 þ 921875p6 rr=ðp2 %1Þ 3087 p ffiffi ffi p ffiffi ffi pffiffiffi p ffiffi ffi 4114800p3 þ 15309 3 þ 659375p6 3 þ 978480p2 3 þ 3195000p4 3 pffiffiffi 1 489888 36384375p5 3 þ 15957000p2 3948750p5 rrz=ðp%1Þ 6174 . pffiffiffi pffiffiffi 2202795p 3 þ 67590000p4 23965875p3 3 þ 26656250p6 rrz2 ð%1Þ pffiffiffi pffiffiffi 5 p 4640085p þ 64010250p3 168399 3 44250000p6 3 504 pffiffiffi pffiffiffi 13456125p2 3 62640000p4 3 þ 25000000p7 þ 117675000p5 rr3 . pffiffiffi pffiffiffi ð250000p6 375000p5 3 þ 855000p4 384750p3 3 þ 330075p2 pffiffiffi 57105p 3 þ 15309 pffiffiffi pffiffiffi %1 :¼ 625p4 750p3 3 þ 1125p2 270p 3 þ 81 pffiffiffi pffiffiffi 5 1 l12 6571125p3 þ 30618 3 535815p þ 1443825p2 3 dh=dt ¼ p 3 17150 . pffiffiffi pffiffiffi 4 l1ð5103 þ 5878125p4 3 þ 1953125p6 3 8803125p5 ðp2 %1Þ 1715 p ffiffi ffi pffiffiffi p ffiffi ffi þ 554850p2 59535p 3 833625p3 3 þ 2036250p4 871875p5 3 pffiffiffi 1 80031375p3 163296 3 2440935p þ 453125p6 z=ðp%1Þ 18522 . pffiffiffi pffiffiffi pffiffiffi 2 þ 14110200p 3 þ 75330000p4 3 þ 21593750p6 3 109171875p5 z2 ð%1Þ pffiffiffi pffiffiffi 5 p 443961 13743675p2 þ 1826145p 3 þ 21269250p3 3 504 pffiffiffi pffiffiffi 62280000p4 þ 36975000p5 3 37750000p6 þ 5000000p7 3 rr2 pffiffiffi pffiffiffi =ð250000p6 375000p5 3 þ 855000p4 384750p3 3 þ 330075p2 pffiffiffi 57105p 3 þ 15309 pffiffiffi pffiffiffi %1 :¼ 625p4 750p3 3 þ 1125p2 270p 3 þ 81 pffiffiffi pffiffiffi 5 pffiffiffi 1 52329375p4 3 2268000p2 3 dz=dt ¼ l1 3p þ l1 z2 þ 18 8820 pffiffiffi pffiffiffi þ 60187500p5 3240405p þ 413343 3 þ 1796875p6 3 þ 46433250p3 l1rr2 . pffiffiffi pffiffiffi p 625p4 þ 750p3 3 1125p2 þ 270p 3 81 pffiffiffi pffiffiffi 25 5 pffiffiffi 1 6984375p5 3 þ 10117800p2 1652805p 3 þ p 2 þ p 3 z3 þ 18 36 1176 pffiffiffi þ27216 8330000p4 4894125p3 3 þ 1843750p6 rr2 z . pffiffiffi pffiffiffi 625p4 þ 750p3 3 1125p2 þ 270p 3 81 .
243
244
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
Appendix C The input data file of Example 3: n :¼ 3: m :¼ 2: omega :¼ Pi/2: phi[1, 1] :¼ exp(I*omega*theta): phi[1, 2] :¼ 0: phi[1, 3] :¼ 0: phi[2, 1] :¼ exp(I*omega*theta): phi[2, 2] :¼ 0: phi[2, 3] :¼ 0: phi[3, 1] :¼ 0: phi[3, 2] :¼ 0: phi[3, 3] :¼ 1: r[1] :¼ 0: r[2] :¼ 1: LP[1, 1, 1] :¼ 0:LP[1, 1, 2] :¼ 0:LP[1, 1, 3] :¼ 0:LP[1, 2, 1] :¼ 0:LP[1, 2, 2] :¼ 0:LP[1, 2, 3] :¼ 0:LP[1, 3, 1] :¼ 0: LP[1, 3,2] :¼ 0: LP[1, 3, 3] :¼ 0: LP[2, 1, 1] :¼ Pi/2:LP[2, 1, 2] :¼ 0:LP[2, 1, 3] :¼ 0:LP[2, 2, 1] :¼ 0:LP[2, 2, 2] :¼ 1:LP[2, 2, 3] :¼ 0:LP[2, 3, 1] :¼ 0:LP[2, 3, 2] :¼ 0: LP[2, 3, 3] :¼ 0: NL[1] :¼ x[1, 4]*x[2, 1] + 2*(x[2, 1]**2) x[1, 3]*x[2, 1] + 3*x[1, 3]*x[2, 3] + (x[2, 2]**2) + 5*(x[1, 3]**2): NL[2] :¼ (x[2, 1]**2) + (x[2, 2]**2) x[1, 2]*x[2, 1]: NL[3] :¼ x[1, 5]*x[1, 3] + (x[1, 1]**2) (x[1, 2]**2) + 2*(x[1, 3]**2): m0 :¼ 2: m2 :¼ 3: The reduced equation on a center manifold up to the 3rd order is 1 ðp5 þ l1p4 þ 8p3 þ 8l1p2 þ 16p 16l1Þl1n1 dn1=dt ¼ pn2 4 ; 2 ð4 þ p2 Þ3 8 þ
pl12 ð4 þ p2 Þn2 ð4 þ p2 Þ3
8 ð15p5 þ 48l1p4 þ 120p3 þ 304l1p2 þ 240p 512l1Þn12 15 ð4 þ p2 Þ3
32 l1ð3p4 35p3 þ 24p2 20p þ 48Þn1n2 32 l1ð3p2 28Þn22 þ 15 15 ð4 þ p2 Þ2 ð4 þ p2 Þ3 ðp5 þ 2l1p4 þ 8p3 þ 16l1p2 þ 16p 32l1Þn1n3 þ4 ð4 þ p2 Þ3 pð4 þ p2 Þl1n2n3 þ 16 3 ð4 þ p2 Þ
þ4 þ
ð8p3 3l2p3 þ 16l1p2 þ 32p 12l2p 64l1Þn32 ð4 þ p2 Þ
2
128 ð3p2 20p þ 12Þn12 n2 128 ð3p2 28Þn1n22 2 2 15 15 ð4 þ p2 Þ ð4 þ p2 Þ
64 ð9p2 4Þn13 15 ð4 þ p2 Þ2
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
2 ð45p5 þ 192p4 þ 360p3 þ 1216p2 þ 720p 2048Þn12 n3 3 15 ð4 þ p2 Þ
þ
8 ð57p4 140p3 þ 456p2 80p þ 912Þn1n2n3 3 15 ð4 þ p2 Þ
2 ð45p3 þ 48p2 þ 180p 448Þn22 n3 ð65p4 þ 8p2 1040Þn1n32 4 2 3 15 ð4 þ p2 Þ ð4 þ p2 Þ
8
pð4 þ p2 Þn2n32 3
8
ð3p3 þ 8p2 þ 12p 32Þn33 2
ð4 þ p2 Þ ð4 þ p2 Þ 1 ðp5 þ l1p4 þ 8p3 þ 8l1p2 þ 16p 16l1Þl1n1 dn2=dt ¼ pn1 8 3 2 ð4 þ p2 Þ p l12 ð4 þ p2 Þn2 16 3 ð4 þ p2 Þ 16 ð15p5 þ 48l1p4 þ 120p3 þ 304l1p2 þ 240p 512l1Þn12 þ 3 15 pð4 þ p2 Þ 64 ð3p4 35p3 þ 24p2 20p þ 48Þl1n1n2 64 ð3p2 28Þl1n22 þ 3 2 15 15 pð4 þ p2 Þ pð4 þ p2 Þ 5 4 3 2 ðp þ 2l1p þ 8p þ 16l1p þ 16p 32l1Þn1n3 þ8 3 ð4 þ p2 Þ p l1ð4 þ p2 Þn2n3 þ 32 3 ð4 þ p2 Þ ð8p3 3l2p3 þ 16l1p2 þ 32p 12l2p 64l1Þn32 þ8 2 ð4 þ p2 Þ p 128 ð9p2 4Þn13 256 ð3p2 20p þ 12Þn12 n2 256 ð3p2 28Þn1n22 þ 2 2 15 pð4 þ p2 Þ2 15 15 pð4 þ p2 Þ pð4 þ p2 Þ
4 ð45p5 þ 192p4 þ 360p3 þ 1216p2 þ 720p 2048Þn12 n3 3 15 ð4 þ p2 Þ p
þ
16 ð57p4 140p3 þ 456p2 80p þ 912Þn1n2n3 3 15 ð4 þ p2 Þ p
4 ð45p3 þ 48p2 þ 180p 448Þn22 n3 ð65p4 þ 8p2 1040Þn1n32 8 2 3 15 ð4 þ p2 Þ p pð4 þ p2 Þ
16
ð4 þ p2 Þn2n32 3
16
ð3p3 þ 8p2 þ 12p 32Þn33 2
ð4 þ p2 Þ ð4 þ p2 Þ p 2 ð4 þ p Þl1n1n2 dn3=dt ¼ l2n3 4 2 ð4 þ p2 Þ ðp5 þ 2l1p4 8p3 16l1p2 16p 32l1Þn22 þ 2n32 2 ð4 þ p2 Þ p 16 ð3p2 10p þ 12Þn12 n2 16 ð3p2 28Þn1n22 þ þ 15 pð4 þ p2 Þ 15 pð4 þ p2 Þ 3 2 8 ð9p 40p þ 36Þn2 ð4 þ p2 Þn1n2n3 ð16 8p2 þ p4 Þn22 n3 þ þ 4 þ 2 2 2 15 pð4 þ p2 Þ ð4 þ p2 Þ ð4 þ p2 Þ p þ 32
ðp2 4p þ 4Þn2n32 . pð4 þ p2 Þ
245
246
R. Qesmi et al. / Applied Mathematics and Computation 181 (2006) 220–246
The normal form in polar coordinate: order = 3 l12 ðp4 þ 12p2 32Þrr l1ðp4 þ 12p2 32Þrrz þ4 2 4 6 64 þ 48p þ 12p þ p 64 þ 48p2 þ 12p4 þ p6 88 3 32 2 32 3 4 2 2 rr 5 p þ 5 p 5 p þ 128 ð130p 24p þ 2112Þrrz 5 ; þ þ 2 4 6 2 4 64 þ 48p þ 12p þ p pð16 þ 8p þ p Þ
dr=dt ¼ 2
1 l12 ð3p2 4Þ l1ð3p2 4Þz þ 32 dh=dt ¼ p 16 2 4 6 2 pð64 þ 48p þ 12p þ p Þ pð64 þ 48p2 þ 12p4 þ p6 Þ 16 3 208 2 64 2 rr 5 p 15 p 5 p þ 1088 ð256p4 48p2 þ 4160Þz2 15 þ þ pð64 þ 48p2 þ 12p4 þ p6 Þ pð16 þ 8p2 þ p4 Þ l1ð16 8p2 þ p4 Þrr2 ð16 8p2 þ p4 Þrr2 z þ . dz=dt ¼ pð16 þ 8p2 þ p4 Þ pð16 þ 8p2 þ p4 Þ References [1] M. Ait Babram, O. Arino, M.L. Hbid, Approximation scheme of a system manifold for functional differential equations, J. Math. Anal. Appl. 213 (1997) 554–572. [2] M.Ait Babram, O. Arino, M.L. Hbid, Computational scheme of a center manifold for neutral functional differential equations, J. Math. Anal. Appl. 258 (2001) 396–414. [3] O. Diekman, S.A. Van Gils, S.M. Verduyn Lunel, H.O. Whalter, Delay equations. Functional-, complex-, and nonlinear analysis, Applied Mathematical Sciences, Vol. 110, Springer-Verlag, New York, 1995. [4] T. Faria, On a planar system modeling a neuron network with memory, J. Differen. Equat. 168 (2000) 129–149. [5] T. Faria, L.T. Magalha`es, Normal forms for retarded functional differential equations with parameters and applications to Hopf bifurcation, J. Differen. Equat. 122 (1995) 181–200. [6] T. Faria, L.T. Magalha`es, Normal forms for retarded functional differential equations with parameters and applications to Hopf bifurcation, J. Differen. Equat. 122 (1995) 181–200. [7] K. Gopalsamy, I. Leung, Delay induced periodicity in a neural netlet of excitation and inhibition, Physica D 89 (1996) 395–426. [8] J. Guckenheimer, P.J. Holmes, Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields, Springer-Verlag, New York, Heidelberg, Berlin, 1983. [9] J.K. Hale, S.M. Verduyn Lunel, Introduction to Functional Differential Equations, Springer-Verlag, New York, 1993. [10] Yu.A. Kuznetsov, Elements of Applied Bifurcation Theory, Springer-Verlag, New York, 1994. [11] L. Olien, J. Be´lair, Bifurcations, stability, and monotonicity properties of a delayed neural network model, Physica D 102 (1997) 349– 363. [12] R. Qesmi, M. Aitbabram, M.L. Hbid, A Maple program for computing a terms of a center manifolds, and elements of bifurcations for a class of retarded functional differential equations with Hopf singularity, Appl. Math. Comput. 175 (2005) 42–78. [13] S. Wiggins, Introduction to applied nonlinear dynamical systems and chaos, Texts in Applied Mathematics, 2, Springer-Verlag, New York, 1990.