• 検索結果がありません。

SHARP INEQUALITIES FOR PERIODIC FUNCTIONS ∗

N/A
N/A
Protected

Academic year: 2022

シェア "SHARP INEQUALITIES FOR PERIODIC FUNCTIONS ∗ "

Copied!
9
0
0

読み込み中.... (全文を見る)

全文

(1)

SHARP INEQUALITIES FOR PERIODIC FUNCTIONS

Jing-wen Li

, Gen-qiang Wang

Received 4 September 2004

Abstract

Sharp inequalities for periodic functions are established which can help to improve many existence criteria for solutions of differential equations.

1 Introduction

Let CT, where T >0, be the space of all real continuous T-periodic functions of the form x : R → R and endowed with the usual linear structure as well as the norm x 0 = max0tT|x(t)|. For any x ∈ C(1)(R, R)∩CT and any ξ ∈ [0, T], by the fundamental theorem of Calculus,

x 0≤|x(ξ)|+ ] T

0 |x (s)|ds. (1)

In particular, let CT0 be the set of all real functions of the form y ∈ CT such that y(ξy) = 0 for someξy ∈[0, T]. Then for anyy∈C(1)(R, R)∩CT0,

y 0≤ ] T

0 |y (s)|ds. (2)

Such inequalities have been used, among many things, for finding a priori bounds for T-periodic solutions of differential equations. By means of such a priori bounds, we may then look for T-periodic solutions by means of fixed point theorems such as the continuation theorems (see e.g. [1]) which are popular (see for examples [1-18]).

However, since (1) and (2) were applied tofind the a priori bounds in these references, and since they are not sharp inequalities (as will be seen below), the corresponding existence criteria cannot be sharp neither.

In [19], it is noted that (1) can easily be extended to x 0≤|x(ξ)|+1

2 ] T

0 |x (s)|ds, (3)

Mathematics Subject Classifications: 26D15, 34K15, 34C25.

Department of Mathematics, Shaoyang Univeristy, Shaoyang, Hunan 422000, P. R. China

Department of Computer Science, Guangdong Polytechnic Normal University, Guangzhou, Guang- dong 510665, P. R. China

75

(2)

which can be used for deriving existence criteria for periodic solutions. The major objective of this paper is, among other things, to show further that the inequalities (3) is sharp. To illustrate their use, we will show how some of the existence criteria can be improved in straightforward manners.

2 Sharp Inequalities

We begin by improving the inequality (1).

THEOREM 1. Supposex=x(t)∈C(1)(R, R)∩CT andξ ∈[0, T]. Then x 0≤|x(ξ)|+1

2 ] T

0 |x (s)|ds, (4)

where the constant factor 1/2 is the best possible.

PROOF. Let x=x(t)∈C(1)(R, R)∩CT andξ∈[0, T].Then for anyt∈[ξ,ξ+T], we have

x(t) =x(ξ) + ] t

ξ

x (s)ds (5)

and

x(t) =x(ξ+T) + ] t

ξ+T

x (s)ds=x(ξ)− ] ξ+T

t

x (s)ds. (6)

From (5) and (6), we see that for any t∈[ξ,ξ+T], 2x(t) = 2x(ξ) +

] t ξ

x (s)ds− ] ξ+T

t

x (s)ds, (7)

that is

x(t) =x(ξ) +1 2

+] t ξ

x (s)ds− ] ξ+T

t

x (s)ds ,

. (8)

Thus for anyt∈[ξ,ξ+T]

|x(t)|≤|x(ξ)|+1 2

] ξ+T

ξ |x (s)|ds, (9)

so that

x 0 = max

ξtξ+T|x(t)|≤|x(ξ)|+1 2

] ξ+T

ξ |x (s)|ds.

≤ |x(ξ)|+1 2

] T

0 |x (s)|ds. (10)

(3)

Now we assert that ifαis a constant andα<1/2,then there arex∈C(1)(R, R)∩ CT andξ ∈[0, T] such that

x 0>|x(ξ)|+α ] T

0 |x (s)|ds. (11)

Indeed, let x(t) = 2−cosT tandξ= 0.Then x 0= 3 and

|x(ξ)|+α ] T

0 |x (s)|ds= 1 +α2π T

] T 0

sin2π

T s

ds= 1 + 4α< x 0

as required. This shows that the constant 1/2 in (4) is the best possible. The proof is complete.

THEOREM 2. Lety∈C(1)(R, R)∩CT0.Then y 0≤ 1

2 ] T

0 |y (s)|ds, (12)

and the constant factor 1/2 is the best possible.

PROOF. Let y∈C(1)(R, R)∩CT0.Theny∈C(1)(R, R)∩CT and there isξ∈[0, T] such thaty(ξ) = 0.From Theorem 1, we have

y 0≤ 1 2

] T

0 |y (s)|ds. (13)

Now we assert that if β be a constant and β <1/2,then there is y∈C(1)∩CT0 such that

y 0>β ] T

0 |y (s)|ds. (14)

Indeed, let y(t) = 1−cosT t.Theny(0) = 0,soy∈C(1)(R, R)∩CT0, y 0= 2 and β

] T

0 |y (s)|ds=β2π T

] T 0

sin2π

T s

ds= 4β< y 0 (15)

as required. Thus the constant factor 1/2 in (12) is the best possible. The proof is complete.

Next let x ∈ C(n)(R, R)∩CT where n N 2. For i = 1,2, .., n−1, note that x(i1)(0) =x(i1)(T),so that isξ∈[0, T] such thatx(i)(ξ) = 0.Thus from Theorem 2, we have

COROLLARY 1. Letx∈C(n)(R, R)∩CT wherenN2.Then

x(i)

0≤1 2

] T 0

x(i+1)(s)ds, i= 1,2, ..., n−1. (16)

(4)

3 Applications

By means of the sharp inequalities derived above, we can improve many existence criteria for periodic solutions of delay differential equations in the literature. We will demonstrate our ideas by improving the results in several recent papers.

First, in [3], the authors consider the existence of 2π-periodic solutions of Rayleigh equations of the form

x (t) +f(x (t)) +g(x(t−τ(t))) =p(t), (17) where f andg are real continuous functions defined onR,f(0) = 0,τ andpare real continuous functions defined onRwith period 2π, andU

0 p(t)dt= 0.

We will show that the condition 4π[r1+ (2π+ 1)r2]<1 in Theorem 1 of [3] can be replaced by the weaker condition 2π[r1+ (π+ 1)r2]<1.More precisely, we have the following existence criteria.

THEOREM 3. Suppose there exist constantsr1, r2 N 0, K > 0 and D > 0 such that

[A1]|f(y)|≤r1|y|+K fory∈R,

[A2]xg(x)>0 and|g(x)|> r1|x|+K for|x|> D,and [A3] limx→−∞g(x)x ≤r2.

Then for 2π[r1+ (π+ 1)r2]<1, (17) has a 2π-periodic solution.

PROOF. We let

x (t) +λf(x (t)) +λg(x(t−τ(t))) =λp(t), (18) where λ ∈ (0,1). In view of the proof of Theorem 1 in [3], it suffices to prove that for any 2π-periodic solutionx(t) of (18), there exist constantsM0 andM1,which are independent fromx(t) andλ, such that

x 0≤M0 and x 0≤M1. (19)

First of all, as in the proof of Theorem 1 in [3], we may show that there is at∈[0,2π]

such that

|x(t)|≤ x 0+D. (20)

Then by (4) and (20), we get x 0≤|x(t)|+1

2 ]

0 |x (s)|ds <(π+ 1) x 0+D. (21) In view of the condition 2π[r1+ (π+ 1)r2]<1, we mayfind a positive numberεsuch that

2π[r1+ (π+ 1) (r2+ε)]<1. (22) From condition [A3], there is aρ> D such that forx <−ρ,

g(x)

x < r2+ε. (23)

(5)

Let

E1={t|t∈[0,2π], x(t−τ(t))>ρ}, E2={t|t∈[0,2π], x(t−τ(t))<−ρ},

E3={t|t∈[0,2π],|x(t−τ(t))|≤ρ},

andgρ= max|x|≤ρ|g(x)|.As in the proof of Theorem 1 in [3], we have ]

E2

|g(x(t−τ(t)))|dt≤2π(r2+ε) x 0, (24) ]

E3

|g(x(t−τ(t)))|dt≤2πgρ, (25) ]

E1

|g(x(t−τ(t)))|dt ≤ ]

E2

|g(x(t−τ(t)))|dt +

]

E3

|g(x(t−τ(t)))|dt+ ]

0 |f(x (t))|dt, (26) and

]

0 |x (s)|ds ≤ ]

0 |f(x (t))|dt+ ]

E1

|g(x(t−τ(t)))|dt +

]

E2

|g(x(t−τ(t)))|dt+ ]

E3

|g(x(t−τ(t)))|dt+ 2π p 0(27). By (26) and (27), we see that

]

0 |x (s)|ds ≤ 2π p 0+ 2 ]

0 |f(x (t))|dt +2

]

E2

|g(x(t−τ(t)))|dt+ 2 ]

E3

|g(x(t−τ(t)))|dt. (28) From (18), (21), (24), (25), (27) and condition [A1], we have

]

0 |x (s)|ds ≤ 2 ]

0 |f(x (t))|dt+ 2π(r2+ε) x 0+ 2πgρ

+ 2π p 0

≤ 2{2πr1 x 0+ 2πK+ 2π(r2+ε) [(π+ 1) x 0+D] + 2πgρ}+ 2π p 0

≤ 4π[r1+ (π+ 1) (r2+ε)] x 01

≤ 2π[r1+ (π+ 1) (r2+ε)]

]

0 |x (s)|ds+σ1

= σ

]

0 |x (s)|ds+σ1, (29)

(6)

where the fourth inequality follows from (16), and σ = 2π[r1+ (π+ 1) (r2+ε)] and σ1= 4π(r2+ε)D+ 4πgρ+ 4πK+ 2π p 0.It follows that

]

0 |x (s)|ds≤D1, (30)

where D1=1σ1

σ.By (16), we have x 0≤ 1

2 ]

0 |x (s)|ds≤M1 (31)

where M1= 12D1. In view of (21) and (31), we have

x 0<(π+ 1) x 0+D≤M0, (32)

where M0= (π+ 1)M1+D.The proof is complete.

We remark that the same reasoning shows that the condition 4π[r1+(2π+ 1)r2]<1 in Theorem 2 of [3] can be replaced by the weaker condition 2π[r1+ (π+ 1)r2]<1.

In [4], the authors studied the existence ofT-periodic solutions of equations of the form

x (t) +f(t, x(t), x(t−τ0(t)))x (t) +β(t)g(x(t−τ1(t))) =p(t), (33) where f is a real continuous functions defined on R3 with positive period T, g is a real continuous function defined on R, and β,τ01 as well as p are real continuous functions defined onRwith periodT.By replacing appropriate inequalities in [4] with ours, it is not difficult to see that the conditions f1 < T1 andr < 1βf1T

1T2 in Theorem 1 of [4] can be replaced by the weaker conditions f1< T2 andr <2

2f1T β1T2

, and the conditionr < βσ

1T in Theorem 2 of [4] by the weaker conditionr < β

1T.

Similar principles can also be applied to otherfirst order delay differential equations.

For insatnce, in [20], the authors studied the existence of T-periodic solutions of equations of the form

x(t) =f(t, x(t), x(t−τ1(t)), ..., x(t−τm(t))), (34) where f = f(t, u0, ..., um) is a real continuous function defined on Rm+2 and is T- periodic intforfixedu0, ..., um. We assume as in [20] thata0, ..., am1, ...,τm, p∈CT

andτ1, ...,τm, pare nonnegative functions andτ0(t)≡0 onR.

THEOREM 4. Suppose that (i) there exists ρ0 >0 such that f(t, u0, ..., um)>0 (< 0) for u0, ..., um > ρ0, and f(t, u0, ..., um)< 0 (respectively> 0) for u0, ..., um <

−ρ0, and (ii)

|f(t, u0, u1, ..., u3)|≤|a0(t)| |u0|+|a1(t)| |u1|+...+|am(t)| |um|+p(t) (35) fort∈R.If

+m [

i=0

] T

0 |ai(t)|dt ,

<2, (36)

(7)

then the equation (34) has at least oneT-periodic solution.

PROOF. Wefirst remark that the condition (36) is much weaker than the origianl condition

T +m

[

i=0

0maxtT|ai(t)| ,

<1.

We let

x (t) =λf(t, x(t), x(t−τ1(t)), ..., x(t−τm(t))), (37) whereλ∈(0,1).In view of the proof of Theorem 1 in [20]. It suffices to prove that for any T-periodic solution x(t) of (37), there exist constants R1, which is independent from x(t) andλ, such that

x 0≤ρ1. (38)

From (37), we see that ] T

0

f(t, x(t), x(t−τ1(t)), ..., x(t−τm(t)))dt= 0. (39) In view of our assumption (i) and (39) and the fact that x is T-periodic, there is a t0∈[0, T] such that

|x(t0)|≤ρ0. (40)

It is easy to see from (4) and (40) that x 0≤ρ0+1

2 ] T

0 |x (s)|ds. (41)

From (35), (37) and (41), we get ] T

0 |x (t)|dt ≤ ] T

0 |f(t, x(t), x(t−τ1(t)), ..., x(t−τm(t)))|dt

≤ [m i=0

] T

0 |ai(t)| |x(t−τi(t))|dt+ ] T

0

p(t)dt

≤ [m

i=0

] T 0 |ai(t)|

# ρ0+1

2 ] T

0 |x (s)|ds

$

dt+T p 0

= 1

2

# m [

i=0

] T

0 |ai(t)|dt

$ ] T

0 |x (s)|ds+T p 00 [m i=0

] T

0 |ai(t)|dt

= 1

2

# m [

i=0

] T

0 |ai(t)|dt

$ ] T

0 |x (s)|ds+C1, (42)

(8)

for some positive constantC1.By (36), we see that there is some positive constantC2

such that

] T

0 |x (t)|dt≤C2. (43)

In view of (41) and (43), we know that

x 0≤ρ0+1

2C2. (44)

The proof is complete.

To close this note, we mention that in [21], the authors studied the existence of T-periodic solutions of equations of the form

x(t) =r(t)−a(t)x(t−σ)−b(t)x(t−τ), (45) whereτandσare positive constants,b∈CT∩C(1)(R, R) such thata(t+σ)−b(t+τ) = 0 for t ∈ R.By means of the same principle illustrated above, it is not difficult to see that the condition

0maxtT|b(t)|+T max

0tT|a(t)|<1

in the main Theorem of [21] can be replaced by the weaker condition max0tT|b(t)|+

1 2

UT

0 a(t)dt <1.

References

[1] R. E. Gaines and J. L. Mawhin, Coincidence Degree and Nonlinear Differential Equations, Lecture Notes in Math., 586, Berlin, New York: Springer-Verlag,1977.

[2] G. Q. Wang and S. S. Cheng, A priori bounds for periodic solutions of a delay Rayleigh equation, Applied Mathematics Letters, 12(1999),41-44.

[3] S. P. Lu, W. G. Ge and Z. X. Zheng, Periodic solutions for a kind of Rayleigh equation with a deviating argument (in Chinese), Acta Math Sinica,47(2)(2004), 299-304.

[4] S. P. Lu and W. G. Ge. X. Zheng, Periodic solutions of the scond order differential equation with deviating arguments (in Chinese), Acta Math Sinica, 45(4)(2002), 811-818.

[5] G. Q. Wang and S. S. Cheng, Periodic solutions ofn-th delay Rayleigh equations, Annales Polonici Mathematici, 78(3)(2002), 261-266.

[6] G. Q. Wang and S. S. Cheng, A priori bounds for periodic solutions of a delay Rayleigh equation with damping, Tamkang J. Math., 34(3)(2003), 41-44.

(9)

[7] G. Q. Wang and J. R. Yan, On existence of periodic solutions of the Rayleigh equation of retarded type,.Internat.J.Math.& Math. Sci., 23(1)(2000), 65-68.

[8] G. Q. Wang and J. R. Yan, Existence theorem of periodic positive solutions for the Rayleigh equation of retarded type, Portugaliae Math., 57(2)(2000), 154-160.

[9] G. Q. Wang and J. R. Yan, Existence of periodic solution for n-th order nonlinear delay differential equation, Far East J. Appl. Math., 3(1)(1999), 129-134.

[10] G. Q. Wang and J. R. Yan, Existence of periodic solutions for second order nonlin- ear neutral delay equations (in Chinese), Acta Math Sinica, 47(2)(2004), 379-384.

[11] G. Q. Wang and J. R. Yan, Existence of periodic solutions forn-th order nonlinear non-automous equations with multiple variable lags (in Chinese), Acta Mathemat- ica Scientia, 23(4)(2003), 431-435.

[12] Y. Li, Periodic solution of a neutral delay equation, J. Math. Anal. Appl., 214(1977), 11-21.

[13] Y. Li, Periodic positive solution of a neutral delay model (in Chinese), Acta Math Sinica, 39(6)(1996), 789-795.

[14] K. Deimling, Nonlinear Functional Analysis, Springer-Verlag, 1985.

[15] F. Zanolin, Periodic solutions for differential systems of Rayleigh type (in Chinese), Rend. Istit. Mat. Univ. Trieste 12(1/2)(1980), 69-77.

[16] S. Invernizzi and F. Zanolin, Periodic solutions of a differential delay equation of Rayleigh type, Rend. Istit. Mat. Univ. Padova 61(1979), 115-124.

[17] F. Liu, Existence of periodic solutions to a class of second-order nonlinear differ- ential equations, Acta Math. Sinica, 33(2)(1990), 260-269.

[18] P. Omari and G. Villari, Periodic solutions of Rayleigh equations with damping of definite sign, Atti. Accad. Naz. Lincer Cl. Sci. Fis. Mat. Natur.1(1)(1990), 29-35.

[19] J. W. Li and S. S. Cheng, Periodic solutions of a second order forced sublinear differential equation with delay, preprint.

[20] X. B. Cao, On the existence of periodic solutions for nonlinear system with multiple delays, Applied Math. Mech., (English Ed.) 24(1)(2003), 117—122

[21] Y. K. Li, On a periodic neutral delay logarithmic population model (in Chinese), J. System Sci. Math. Sci., 19(1)(1999), 34-38.

参照

関連したドキュメント

We summarize the conditions discovered for the existence of new ergodic type solutions (asymptotically almost periodic, pseudo almost periodic,. .) of differential equa- tions

Wei, Existence of positive periodic solutions of nonautonomous differential equations with delay, Chinese Annals of Mathematics, 20A(6) (1999) 715-720.. Kuang, Delay

In our work, we show firstly an adequate fixed point theorem for vectorial version with two components (see theorem 2.7) which extend some existing results even in the case of

It is well known that in general Burkholder’s function (that is, the special function leading to a given martingale inequality) is not unique, see e.g.. Sometimes it is of interest

Also in this case all

Furthermore, the following analogue of Theorem 1.13 shows that though the constants in Theorem 1.19 are sharp, Simpson’s rule is asymptotically better than the trapezoidal

Using an “energy approach” introduced by Bronsard and Kohn [11] to study slow motion for Allen-Cahn equation and improved by Grant [25] in the study of Cahn-Morral systems, we

By applying the Schauder fixed point theorem, we show existence of the solutions to the suitable approximate problem and then obtain the solutions of the considered periodic