• 検索結果がありません。

SINGULAR POINTS AND LIE ROTATED VECTOR FIELDS

N/A
N/A
Protected

Academic year: 2022

シェア "SINGULAR POINTS AND LIE ROTATED VECTOR FIELDS"

Copied!
7
0
0

読み込み中.... (全文を見る)

全文

(1)

Vol. 24, No. 3 (2000) 179–185 S016117120001005X

© Hindawi Publishing Corp.

SINGULAR POINTS AND LIE ROTATED VECTOR FIELDS

JIE WANG and CHEN CHEN (Received 2 April 1999)

Abstract.This paper gives the definition of Lie rotated vector fields in the plane and the conditions of movement of singular points on Lie rotated vector fields with variable parameters.

Keywords and phrases. Lie rotated vector fields, Lie bracket, one parameter group, singu- lar points.

2000 Mathematics Subject Classification. Primary 34C05.

1. Introduction. Many engineering problems are usually run into a class of non- linear equations that contain variable parameters. In order to study whole orbits or whole phase diagrams of vector fields that contain parameters, it is a complicated and interesting problem how the whole orbit or whole phase diagram change as param- eter is changed. It is extremely complicated for general containing parameter vector fields to change in the plane, but for some special containing parameter rotated vector fields, their change has regular rule as parameter is changed. These are many results in this respects [3, 4, 5, 6, 7].

In Section 2, we present the basic definitions of Lie rotated vector fields. We define Lie rotated vector fields using one parameter group approach. In accordance with the strict definition of rotated vector field, the singular points ofX(µ)must be kept fixed, but in this paper, the singular points ofX(µ)can be moved as parameterµis changed. In Section 3, we discuss the motion of singular points on Lie rotated vector fields. In the section, we require the singular points of X(µ) to be strictly moved as parameterµ is changed, and permit the moved singular points to disappear or decompose, which do not coincide with the singular points of original vector field. We give some conditions and properties corresponding to the vector fieldY. In this paper, we give some examples to illustrate the concept and notion of Lie rotated vector fields.

2. Lie rotated vector fields. We consider vector fields on the planex=(x1,x2)

R2,

X=

X1(x),X2(x)

, Y=

Y1(x),Y2(x)

. (2.1)

For the vector fields (2.1), we define

X∧Y=X1Y2−X2Y1, X,Y =X1Y1+X2Y2. (2.2) If X and Y are vector fields, then [X,Y ]is a vector field which is operated by Lie

(2)

bracket, i.e.,

[X,Y ]= Z1,Z2

, (2.3)

whereZ1andZ2are expressed as,

Z1= X,∇Y1−Y ,∇X1, Z2= X,∇Y2−Y ,∇X2, (2.4) respectively, whereis gradient operator.

Let the plane vector fieldsX(µ)=(X1(x,µ),X2(x,µ))be defined by the following differential equations:

dx1

dt =X1(x,µ), dx2

dt =X2(x,µ), (2.5)

whereX1andX2are functions ofxand parameterµ∈I⊂R, and the singular points are isolated.

Definition2.1. Let the plane vector field X(µ) be determined by (2.5), where X1,X2∈C3(R2×I,R),I= {µ| |µ|< δ}is a real interval,δis a given positive number.

If vector fieldY exists which is defined by the following differential equations:

dx1

dt =Y1(x), dx2

dt =Y2(x), (2.6)

whereY1andY2∈C3(R2,R). At all ordinary points ofX(0), such that the following relation holds

L(0)def= X(0)∧

Xµ(0)+[X(0),Y ]

>0(<0), (2.7) whereXµ(0)is the derivative of the vector fieldX(µ)atµ=0, then X(µ), µ∈I, is called Lie rotated vector fields.

Remark2.2. If the vector fieldX(µ)is defined onD×I, whereD⊂R2, such that X(0)satisfies relation (2.7) at all ordinary points ofX(0)onD, thenX(µ), µ∈I, is called Lie rotated vector fields onD.

Lemma2.3. Letψs be a one parameter transform group which is produced byC1 vector fieldY,s∈R, and letXbeC1vector field. Ifsis fixed, andϕp(t)is an integral curve ofXthrough the pointp,ϕp(0)=p, thenψs◦ϕp(t)is an integral curve ofψsX through the pointψs(p). IfX|p=0, thensX)|ψs(p)=0.

Proof. The proof follows from [1] and [2]. In fact, ifϕp(t)is an integral curve of Xthrough the pointp, then

ψs◦ϕp(t)

t=0s(p) (2.8)

and

ψs◦ϕp(t)

t· d

dtt

sϕp(t)◦ϕp(t)t· d

dtt

sϕp(t)·Xϕp(t)= ψsX

ψs◦ϕp(t). (2.9)

(3)

It follows thatψs◦ϕp(t)is an integral curve ofψsXthrough the pointψs(p).

Next, due to

ψsX|q=Dψs

ψ−s(q)

·X

ψ−s(q)

, q∈R2. (2.10)

Setq=ψs(p), note that we already suppose X|p=0, again note that ψs is a one parameter transform which is produced byY, then

ψsX|ψs(p)=Dψs(p)·X(p)=Dψs(p)·X|p=0, (2.11) i.e.,ψs(p)is a singular point ofψsX.

Lemma2.4. Letψs be a one parameter transform group which is produced byC1 vector fieldY,s∈R, fixs, then the index of isolated singularity ofC1vector fieldXis not changed under theψs transform.

Proof. In fact, by the condition of the lemma, it is known thatψsis a differentiable homeomorphism, then the lemma follows from [8, Theorem 4.2].

Next, ifX(µ) is a Lie rotated vector field, then Y is a corresponding vector field which satisfies (2.7), andψs is a one parameter transform group which is produced byY,s∈R.

Lemma2.5. LetX(µ)be a Lie rotated vector field, for allε >0, there existδ=δ(ε), such that when|µ|< δ,ψsX(µ)constitutes a rotated vector field.

Proof. Let the singular points ofψsX(µ), µ=0, on the planeR2bepµ1,...,pµk

and the singular points ofX(0)on the planeR2arep1,...,pm,∀ε >0, 0< ε1, let Sε(pµi)orSε(pj) (1≤i≤k, 1≤j≤m)be open neighborhood pµi (1≤i≤k)and pj(1≤j≤m), and radiusε, such thatSε(p)∩Sε(q)= ∅, wherepandq∈ {pµi}∪{pj} (1≤i≤k, 1≤j≤m),p=q. Letψs be a one parameter transform group which is produced byC1vector fieldY,s∈R. By the limit definition of Lie bracket, we have

ψsX(µ)=ψ0X(µ)+ s 1!

d

dts=0ψsX(µ)+s2 2!

d2

dt2s=0ψsX(µ)+···

=X(µ)+s

1![X(µ),Y ]+s2

2![[X(µ),Y ],Y ]+···.

(2.12)

Next, we notice thatX(µ)can be unfolded as X(µ)=X(0)+µ

1!Xµ(0)+µ2

2!Xµ(0)+···, (2.13) since

[X(µ),Y ]=[X(0),Y ]+µ 1!

Xµ(0),Y 2 2!

Xµ(0),Y +···. (2.14)

Lets=µ, it follows from (2.12), (2.13), and (2.14) that ψµX(µ)=X(0)+µ

Xµ(0)+[X(0),Y ] +1

2µ2

Xµ(0)+2

Xµ(0),Y +[[X(0),Y ],Y ]

+···. (2.15)

(4)

At the ordinary points of R2\k

i=1Sε(pµi)m

j=1Sε(pj)

, for given ε >0, we sooner or later can findδ11(ε) >0, such that when|µ|< δ1, we have

ψµX(µ)∧

∂µ

ψµX(µ)

=X(0)∧

Xµ(0)+[X(0),Y ] +µX(0)∧

Xµ(0)+2

Xµ(0),Y +[[X(0),Y ],Y ] +···

=L(0)+O(µ) >0(<0)

(2.16) and letϑ(µ)be the crossing angle ofψµX(µ)and the x1axis, for givenε >0, we sooner or later can findδ22(ε), such that when|µ|< δ2, at the ordinary points of R2\k

i=1Sε(pµi)m

j=1Sε(pj)

µX(µ)isX(0)whenµ=0,ϑ(0)is the crossing angle ofX(0)and thex1axis), so

0<|ϑ(µ)−ϑ(0)|< π. (2.17) Takeδ=min{δ12}, then when|µ|< δ,ψµX(µ)constitutes a rotated vector field.

Remark2.6. In accordance with the strict definition of rotated vector field, the singular points must be kept fixed, but the singular points ofψµX(µ)in Lemma 2.5 can be moved as parameterµ is changed. In the unmistakable circumstance, when

|µ|< δ, we callψµX(µ)a rotated vector.

In the above lemma,δneeds not be a quite small positive number, i.e., 0< δ1 need not be set up. For the sake of distinctness, we cite an example to illustrate this equation.

Example2.7. LetX(µ)=(x2,−x1+µx2), if we takeY=(−x2/2,0), then at all the ordinary points ofX(0), we have

X(0)= 1 2

x21+x22>0, (2.18) that is,X(µ)is a Lie rotated vector field.

Now we consider the range ofµ, because ψµX(µ)=

1 2µx1+

1−1

4µ2

x2,−x1+1 2µx2

(2.19) so

ψµX(µ)∧

∂µ

ψµX(µ)

=1 2

x12+x22

1

2µx1x2+1

8µ2x22. (2.20) Formula (2.16) is compared with formula (2.20), we can find thatO(µ)in formula (2.16) is replaced byO(µ)in formula (2.20),

O(µ)= −1

2µx1x2+1

8µ2x22 (2.21)

(5)

yet go a step further calculating, we have ψµX(µ)∧

∂µ

ψµX(µ)

=1 2x22+1

8

µx22x12 (2.22)

which is larger than zero at the ordinary points ofX(0)and ψµX(µ) for allµ∈R, but the range ofµ that satisfies formula (2.17) is|µ|<4, thus we takeδ=4, when

|µ|< δ=4,ψµX(µ)constitutes rotated vector field.

3. The motion of singular points. LetX(µ)be a Lie rotated vector field, we require the singular points ofX(µ)to be strictly moved as parameterµis changed, and permit the singular points that have been moved disappear or decompose, but require the singular points that have been decomposed to be at most limited in number, which do not coincide with the singular points of the original vector field.

Ifpis a singular point ofX(µ), we nameJµ(p)for index of singular pointpofX(µ), under the same circumstances,J0(p0)for index of singular pointp0ofX(µ),Jµ(q) for index of singular pointqofψµX(µ) (µ=0).

Theorem3.1. LetX(µ)be a Lie rotated vector field,X(0)|p0=0, and letY|p0=0. If the singular pointp0ofX(0)disappears or decomposes aspi(1≤i≤k)inX(µ) (µ= 0), thenJ0(p0)=0, andJµ(pi)=0(µ=0, 1≤i≤k).

Proof. First of all, we prove thatJ0(p0)=0. In fact, because ofX(µ)|p0=0(µ=0), utilize Lemma 2.3 and conditionY|p0=0, we know thatψµX(µ)|p0=0(µ=0), it fol- lows from Lemma 2.5, for givenδ >0, when|µ|< δ,ψµX(µ)constitutes a rotated vector field. Takeη >0 as quite small positive number, such that Sη(p0)does not contain the singular points ofψµX(µ) (µ=0), and only contains the isolate singu- lar pointp0 ofX(0). It is easy to know thatJµ(p0)=0 about ∂Sη(p0). By (2.17) of Lemma 2.5, it follows thatJ0(p0)=0 when|µ|< δ.

Using the same method, we prove Jµµ(pi))= 0 = 0, 1 ≤i≤ k) and by Lemma 2.4, we findJµ(pi)=0(µ=0,1≤i≤k).

Corollary3.2. LetX(µ) be a Lie rotated vector field,X(0)|p0 =0, ifY|p0=0, and moved the singular pointspi−µ(p0) (µ=0, 1≤i≤k), thenJ0(p0)=0and Jµ(pi)=0(µ=0,1≤i≤k).

Proof. SinceX(0)|p0 =0, let the singular point of X(µ) (µ=0)disappears or decomposes intop1,...,pk points which do not coincide with singular point p0 of X(0), i.e.,X(µ)|pi =0(1≤i≤k), yet because ofX(µ)|p0=0(µ=0)and Y|p0=0.

By Lemma 2.3, we haveψµX(µ)|ψµ(p0)=0 andψµX(µ)|ψµ(pi)=0, but by condition ψµ(pi)=p0, we know thatψµX(µ)|p0=0, as in the proof of Theorem 3.1, we can prove thatJ0(p0)=0 andJµ(pi)=0(µ=0,1≤i≤k).

Corollary3.3. LetX(µ) be a Lie rotated vector field,X(0)|p0 =0, ifY|p0=0, but for some i0 (1≤i0 k), set up ψµ(pi0)=p0 =0), then J0(p0)=jµ(pi0), Jµ(pi)=0(µ=0,1≤i≤kandi=i0).

Example3.4. LetX(µ)=

x22,−x1

, and let Y=

3x1−αx2,2x2

(3.1)

(6)

when|µ|< δ, we takeα >0 andα1, on the range ofD= {(x1,x2)|x2< α−1} ⊂R2, at all ordinary points∈DofX(0), set up

L(0)=αx12+x22−αx23>0, (3.2) that is,X(µ)constitutes a Lie rotated vector field onD, the singular points ofX(µ)are strictly moved as parameterµis changed. We note thatY|p0=0,p0=(0,0)is singular point ofX(0), by Theorem 3.1, we can find thatJ0(p0)=0 andJµ(pi)=0(µ=0), wherepi=(µ,0).

Theorem3.5. LetX(µ)be a Lie rotated vector field,X(0)|p0=0,p0is elementary.

(1)IfY|p0=0, thenp0cannot be moved as parameterµis changed.

(2)IfY|p0=0, thenp0can be moved as parameterµis changed, and the moved point is the singular pointψ−µ(p0)ofX(µ) (µ=0).

Proof. (1) We noteJ0(p0)= ±1=0, it is proved immediately from Theorem 3.1.

(2) First of all, we prove thatp0 is indeed moved asµ is changed, suppose that it is not real, i.e.,p0 is not moved asµ is changed, then that X(µ)|p0 =0 (µ=0), by Lemma 2.3, we know thatψµX(µ)|ψµ(p0)=0. Becausep0is isolate singular point ofX(0), we take ¯δ >0 and ample smallη >0, it follows thatψµ(p0)∈Sη(p0). When 0<|µ|<δ < δ, then for¯ ∂Sη(p0), we haveJµ(p0)=0 (sinceψµX(µ)|p0=0), where µ=0. ButJ0(p0)= ±1=0, this is a contradiction from Lemma 2.5. Thus we have provedp0is indeed moved asµis changed, and by Corollaries 3.2 and 3.3, it follows thatp0is moved as the singular pointψ−µ(p0)ofX(µ) (µ=0)whenµ is changed.

Lemma3.6. LetX(µ)be a Lie rotated vector field,X(0)|p0=0, and there is an elliptic region at the singular pointp0.

(1)IfY|p0=0, then the singular pointp0cannot be moved when parameterµ=0.

(2)IfY|p0=0, then when parameterµ=0, singular pointp0 is moved, andp0be moved as singular pointψ−µ(p0)ofX(µ).

Proof. (1) We already know thatY|p0, suppose the original equation is not real, then when µ=0, singular pointp0 is moved, thus we letp0 moved as the singu- lar point pµ ofX(µ), X(µ)|pµ =0, pµ=p0, µ=0. From Lemma 2.3, we know that ψµX(µ)|ψµ(pµ)=0, and byY|p0=0, we know thatψµ(pµ)=p0(µ=0). LetΩbe an elliptic region at the singular pointp0ofX(0), for arbitrary fixedµ (0<|µ|< δ), it is sure to have some elliptic trajectoryr ofX(0), which does not contain the point ofψµ(pµ)onr and inr. Sincer has single direction, and there is no singular point ofψµX(µ)onrand inr. By Lemma 2.5, we can know that positive half trajectory or negative half trajectory ofψµX(µ)which pass through the pointpwill wander about without a home to go to, wherepis any point which passes the inner region ofr, this is a contradiction.

(2) Now we knowY|p0=0, yet use reduction to absurdity. Suppose, whenµ=0, sin- gular pointp0is not moved, i.e., establishX(µ)|p0=0, namely we haveψµX(µ)|ψµ(pµ)= 0, andψµ(p0)=p0. The method of the proof is completely alike as part (1), we can prove it is a contradiction. Thus let µ =0, singular point p0 is moved as singular point pµ (p0 =pµ) of X(µ), i.e., ψµX(µ)|ψµ(pµ) =0 (µ= 0). If ψµ(pµ)= p0, the

(7)

method of the proof is alike as in part (1), yet it is a contradiction, thus only establish ψµ(pµ)=p0, orpµ−µ(p0).

Lemma3.7. LetX(µ)be a Lie rotated vector field,X(0)|p0=0, and whenµ=0,p0

is moved as the singular pointpµ(pµ=p0)ofX(µ)asµis changed. IfY|p0=0, then for singular pointpµ(orp0), at least there are a positive half trajectory and a negative half trajectory ofX(µ)(orX(0)) to get into it.

Proof. We only prove the circumstance of pointpµ(the proof is completely alike as the circumstance of pointp0).

From Lemma 3.6, we know that there is no elliptic region which links with the singu- lar pointp0ofX(0), the same do the singular pointpµofX(µ), and from Theorem 3.1, we know that the index ofpµofX(µ)is zero. Takepµ as circular center, make the circumference of a circlelwith radius rather small, and let that hyperbolic region of pointpµwhich intersects with the circumference of a circlelhash. By the Bendixson’s formula in§6of Chapter 3 of [8], we can immediately findh=2.

From Lemmas 3.6and 3.7, we have the following theorem.

Theorem3.8. LetX(µ)be a Lie rotated vector field,X(0)|p0=0, and letY|p0=0, then some singulars while can be moved as parameterµis changed inX(µ)only contain two hyperbolic regions and their index is zero.

Acknowledgement. The authors express grateful thanks to Professor Yaoxian Wang for his help and direction in the work.

References

[1] W. M. Boothby,An Introduction to Differentiable Manifolds and Riemannian Geometry, Academic Press [A subsidiary of Harcourt Brace Jovanovich, Publishers], New York, London, 1975. Zbl 333.53001.

[2] S. S. Chen and W. H. Chen,Lectures in Differential Geometry, Beijing Univ. Press, Beijing, 1983.

[3] Y. Y. Chen,Limit cycles and quasirotational vector fields, Acta Math. Sinica32(1989), no. 6, 786–792. MR 91c:58113.

[4] G. F. D. Duff,Limit-cycles and rotated vector fields, Ann. of Math. (2)57(1953), 15–31.

MR 14,751c. Zbl 050.09103.

[5] Z. E. Ma,The motion of singular closed orbit rotated vector fields, J. Xi’an Jiaotong Univ.4 (1978), 49–65.

[6] L. M. Perko,Rotated vector fields and the global behavior of limit cycles for a class of qua- dratic systems in the plane, J. Differential Equations18(1975), 63–86. MR 51#10752.

Zbl 297.34024.

[7] Y. Q. Ye, S. L. Cai, L. S. Chen, K. C. Huang, D. J. Luo, Z. E. Ma, E. N. Wang, M. S. Wang, and X. A. Yang,Theory of Limit Cycles, American Mathematical Society, Providence, R.I., 1986. MR 88e:58080.

[8] Z. F. Zhang, T. R. Ding, W. Z. Huang, and Z. X. Dong,The Qualitative Theory of Differential Equation, Science Publishing House, Beijing, 1985.

Jie Wang and Chen Chen: School of Electric Power, Shanghai Jiaotong University, Shanghai,200030, China

参照

関連したドキュメント

We consider some applications of the singular integral equation of the second kind of Fox1. Some new solutions to Fox’s integral equation are discussed in relation to

ONE REMARK ON VARIATIONAL PROPERTIES OF GEODESICS IN PSEUDORIEMANNIAN AND GENERALIZED FINSLER SPACES.. JOSEF MIKEŠ † , IRENA HINTERLEITNER ‡ and ALENA

Here, we show that the complementability of L 1 (µ, X) in cabv(µ, X) together with that one of X in the bidual X ∗∗ is equivalent to the complementability of L 1 (µ, X) in its

The asymptotic behavior of the singular and entropy numbers is established for the Erdelyi–K¨ober and Hadamard integral operators (see, e.g., [15]) acting in weighted L 2 spaces...

Prignet; Equivalence between entropy and renormalized solutions for parabolic equations with smooth measure data, NoDEA Nonlinear Differential Equations Appl... Friedman;

In this paper, we focus on the existence and some properties of disease-free and endemic equilibrium points of a SVEIRS model subject to an eventual constant regular vaccination

SINGULAR AND MAXIMAL RADON TRANSFORMS 521 Coupling this result with the uniqueness of the exponential representa- tion, we conclude that the vector fields in that representation

But the proof of Theorem 1.2 is exactly the proof of [10, Theorem 4 ], the difference is that such a proof uses the next Lemma 2.3 that it works when the polynomial differential