Periodic Solutions of Nonautonomous Ordinary Differential Equations
Periodic Solutions of Nonautonomous Ordinary Differential Equations
Periodic Solutions of Nonautonomous Ordinary Differential Equations
DOI 10.1007/s00605-009-0138-7
I. Kiguradze · A. Lomtatidze
Received: 4 September 2008 / Accepted: 23 June 2009 / Published online: 7 July 2009
© Springer-Verlag 2009
Abstract For higher order ordinary differential equations, new sufficient conditions
on the existence and uniqueness of periodic solutions are established. Results obtained
cover the case when the right-hand side of the equation is not of a constant sign with
respect to an independent variable.
Introduction
In the present paper, for a higher order nonautonomous ordinary differential equation
we investigate the problem on the existence of a periodic solution with a prescribed
Communicated by A. Jüngel.
For the first author, the research was supported by the Georgian National Science Foundation under the
project GNSF/ST06/3-002. For the second author, the research was supported by the Ministry of
Education of the Czech Republic under the project MSM0021622409 and by the Academy of Sciences of
the Czech Republic, Institutional Research Plan No. AV0Z10190503.
I. Kiguradze (B)
A. Razmadze Mathematical Institute, 1, Aleksidze Str., 0193 Tbilisi, Georgia
e-mail: [email protected]
A. Lomtatidze
Department of Mathematics and Statistics, Faculty of Sciences,
Masaryk University, Kotlářská 2, 611 37 Brno, Czech Republic
e-mail: [email protected]
123
236 I. Kiguradze, A. Lomtatidze
period. In Sect. 1, the optimal, in a certain sense, conditions are found guaranteeing
the existence of a unique ω-periodic solution of the linear differential equation
where δ : R × [0, +∞[ → [0, +∞[ is a sublinear function with respect to the second
variable. Moreover, in spite of previously known results (see [2–7], [11–20] and the
references therein), we do not restrict signs of the functions p1 and p2 .
Throughout the paper, we assume that n ≥ 2 and ω > 0. We also use the following
notation.
1 1
[x]+ = (|x| + x) , [x]− = (|x| − x) .
2 2
+∞
1
ζ (x) = for x > 1.
kx
k=1
L ω is the space of all ω-periodic real functions which are Lebesgue integrable on
[0, ω].
L 2ω is the space of all ω-periodic real functions which are square Lebesgue integra-
ble on [0, ω].
Cω , resp. ACω , is the space of continuous, resp. absolutely continuous, ω-periodic
functions u : R → R,
123
Periodic solutions of nonautonomous ordinary differential equations 237
ω
1
lim δ(t, )dt = 0.
→+∞
0
ω
If p ∈ L ω and 0 p(t)dt = 0, then
ω 2 ω
| p(t)|dt
γ0 ( p) = 1 + ω
0 , γ ( p) = γ0 ( p) | p(t)|dt.
p(t)dt
0
0
ω
If p1 , p2 ∈ L ω and 0 p2 (t)dt = 0, then
ω 2 ω
p 0 (t)dt
η0 ( p 1 , p 2 ) = 1 + ω
0 , η( p1 , p2 ) = η0 ( p1 , p2 ) p0 (t)dt,
p2 (t)dt
0
0
where
1
p0 (t) = (| p1 (t)| + | p2 (t)| + || p1 (t)| − | p2 (t)||).
2
ω
1
c0 = u(t)dt
ω
0
1 Linear problem
where p, q ∈ L ω .
123
238 I. Kiguradze, A. Lomtatidze
The following lemma is well-known from the general theory of linear boundary
value problems (see, e.g., [9, Theorem 1.1]).
Lemma 1.1 Equation (1.1) has a unique ω-periodic solution iff the corresponding
homogeneous equation
u (n) = p(t)u (1.10 )
has no nontrivial ω-periodic solution.
and
ω
ζ (2) ω 2−1 () 2
u − c0 C
2
≤ u (t) dt. (1.4)
ω
π 2π
0
for some c, t0 ∈ R.
123
Periodic solutions of nonautonomous ordinary differential equations 239
+∞
u(t) = c0 + h k (t) for t ∈ R (1.9)
k=1
and
+∞
2π ω
u () (t) = khk t + for t ∈ R,
ω 4k
k=1
where
2kπ 2kπ
h k (t) = c1k sin t + c2k cos t.
ω ω
Hence, by virtue of Parseval’s equality, we get
ω
ω 2
+∞
|u(t) − c0 | dt =
2
c1k + c2k
2
(1.10)
2
0 k=1
and
ω
+∞
() 2 ω 2π 2 2 2
u (t) dt = k c1k + c2k
2
. (1.11)
2 ω
0 k=1
Inequality (1.3) now immediately follows from (1.10) and (1.11). Moreover, equality
(1.5) holds if and only if
i.e., when
2π t 2π t
u(t) ≡ c0 + c11 sin + c21 cos .
ω ω
However, the latter identity is equivalent to (1.6) for a suitable choice of c and t0 .
Now we will prove inequality (1.4). Choose t0 ∈ [0, ω] such that
u − c0 Cω = |u(t0 ) − c0 |.
+∞
u − c0 C
2
ω
≤ ζ (2) k 2 h 2k (t0 ). (1.12)
k=1
123
240 I. Kiguradze, A. Lomtatidze
+∞
u − c0 C
2
ω
= ζ (2) k 2 h 2k (t0 )
k=1
c
h k (t0 ) = for k = 1, 2, . . . . (1.13)
k 2
2kπ t0 2kπ t0 2
h 2k (t0 ) = c1k
2
+ c2k
2
− c1k cos − c2k sin .
ω ω
Hence, from (1.11) and (1.12) we get (1.4). Moreover, equality (1.7) holds if and only
if (1.13) and
2kπ 2kπ
c1k cos t0 − c2k sin t0 = 0 for k = 1, 2, . . . (1.14)
ω ω
c 2kπ c 2kπ
c1k = sin t0 , c2k = 2 cos t0 , for k = 1, 2, . . . ,
k 2 ω k ω
ω
ζ (2) ω 2−1 () 2
u − c0 C
2
< u (t) dt. (1.16)
ω
π 2π
0
Proof Assume the contrary that (1.16) does not hold. Then, by virtue of (1.15) and
Lemma 1.2, identity (1.8) is fulfilled with c = 0. Hence,
2−1
+∞
2π 1 2kπ
u (2−1) (t) = c(−1) sin (t − t0 ) for 0 < t − t0 < ω.
ω k ω
k=1
123
Periodic solutions of nonautonomous ordinary differential equations 241
Therefore,
2−1
(2−1) π 2π
u (t) = c(−1) (ω − t + t0 ) for 0 < t − t0 < ω,
ω ω
ω
p(t)u(t)dt = 0, (1.17)
0
ω ω
(m) 2
u (t) dt = (−1)m
p(t)u 2 (t)dt for n = 2m, (1.18)
0 0
and
ω
p(t)u 2 (t)dt = 0 for n = 2m + 1. (1.19)
0
while if
ω
p(t)dt = 0, (1.21)
0
uC
2
ω
≤ γ0 ( p)u − c0 C
2
ω
(1.22)
holds.
123
242 I. Kiguradze, A. Lomtatidze
Now assume that (1.21) holds. By virtue of (1.17), for any c0 ∈ R we have
ω ω
c0 p(t)dt = − p(t) (u(t) − c0 ) dt.
0 0
Hence,
ω
| p(t)|dt
|c0 | ≤ 0 ω u − c0 Cω .
p(t)dt
0
ω
p(t) ≡ 0, (−1) m
p(t)dt ≥ 0. (1.23)
0
and
ω
n−1
π 2π
(−1)m p(s) +
ds ≤ (1.25)
ζ (n) ω
0
be fulfilled. Then, (1.1) has one and only one ω-periodic solution.
Proof By virtue of Lemma 1.1, it is sufficient to show that the homogeneous equation
(1.10 ) has no nontrivial ω-periodic solution.
Assume the contrary that u is a nontrivial ω-periodic solution of (1.10 ). Then, by
virtue of Lemma 1.4, u (m) (t) ≡ 0 and relations (1.17) and (1.18) hold. Denote by c0
the mean value of the function u. Then, in view of (1.17), it easily follows from (1.18)
that
ω ω ω
(m) 2
0 < u (t) dt = (−1)m
p(t) (u(t) − c0 ) dt − (−1) c0
2 m 2
p(t)dt,
0 0 0
123
Periodic solutions of nonautonomous ordinary differential equations 243
ω ω
(m) 2
0 < u (t) dt ≤ (−1)m p(t) + (u(t) − c0 )2 dt. (1.26)
0 0
ω ω n ω
(m) 2
|u(t) − c0 | dt <
2
u (t) dt,
2π
0 0
or there exist c, t0 ∈ R such that (1.6) is fulfilled and, moreover, c = 0. In both cases,
by virtue of (1.24) and (1.26), we obtain a contradiction
ω ω 2
(m) 2
u (t) dt < (u (m) (t) dt.
0 0
ω
n−1
(m) 2 π 2π
u (t) dt ≤ u − c0 C
2
.
ζ (n) ω ω
0
ω
ζ (n) ω n−1
u − c0 C
2
< |u (m) (t)|2 dt.
ω
π 2π
0
Remark 1.2 Condition (1.23) in Theorem 1.1 cannot be replaced by the condition
ω
p(t) ≡ 0, (−1) m
p(t)dt > −ε, (1.27)
0
123
244 I. Kiguradze, A. Lomtatidze
where
ω n 1 1
0 < ε < min , .
2π ω 2ζ (n)
Then conditions (1.24) and (1.25) hold, but instead of (1.23) condition (1.27) is ful-
filled. However, the function u is a nontrivial ω-periodic solution of (1.10 ). Therefore,
by virtue of Lemma 1.1, equation (1.1) either has no ω-periodic solution or has infi-
nitely many ω-periodic solutions.
Remark 1.3 Condition (1.24) is optimal and cannot be weakened. Indeed, if p(t) ≡
n
(−1)m 2πω , then for any c1 , c2 ∈ R\{0} the function
2π t 2π t
u(t) = c1 sin + c2 cos
ω ω
Remark 1.4 It follows from Theorem 1.1 that the second order equation
ω
p(t) ≡ 0, p(t)dt ≤ 0, (1.29)
0
and
2
2
2π 2π
p(t) ≥ − for t ∈ R, p(t) ≡ − .
ω ω
This result belongs to Mawhin [21] and Mawhin and Ward [22]. For n = 2, condition
(1.25) is not a new as well, since as it is shown by Lasota and Opial [17] condition
(1.29), together with
ω
16
[ p(t)]− dt ≤ ,
ω
0
123
Periodic solutions of nonautonomous ordinary differential equations 245
Theorem 1.1 implies a result stated in [18] and Theorem 1.1 established in [13]. Note
that in [18] the condition
ω
n−1
2 2π
| p(t)|dt ≤
π ω
0
π2
ζ (n) ≤ ζ (2) = for n ≥ 2.
6
Theorem 1.2 Let n = 2m,
ω
(−1) m
p(t)dt < 0, (1.30)
0
and
ω
n−1
π 2π
γ0 ( p) (−1)m p(t) + dt ≤ . (1.31)
ζ (n) ω
0
ω ω
(m) 2
0 < u (t) dt < (−1)m p(t) + u 2 (t)dt
0 0
ω
≤ γ0 ( p)u − c0 C
2
ω
(−1)m p(t) + dt.
0
ω
ζ (n) ω n−1 (m) 2
u − c0 C
2
≤ u (t) dt.
ω
π 2π
0
ω
ζ (n) ω n−1
γ0 ( p) (−1)m p(t) + dt > 1,
π 2π
0
123
246 I. Kiguradze, A. Lomtatidze
which contradicts (1.31). Thus, (1.10 ) has no nontrivial ω-periodic solution. Therefore,
according to Lemma 1.1, (1.1) has one and only one ω-periodic solution.
ω
1
v − c0 Cω ≤ |v (t)|dt, (1.32)
2
0
Thus
t
1 t1
v − c0 Cω = v (s)ds ≤ |v (s)|ds,
t0 t0
t +ω
0 0 +ω
t
v − c0 Cω = v (s)ds ≤
|v (s)|ds.
t1 t1
0 +ω
t ω
2v − c0 Cω ≤ |v (s)|ds = |v (s)|ds.
t0 0
ω
σ p(t)dt < 0, (1.33)
0
and
ω
2n−2
1 2π
γ ( p) [σ p(t)]+ dt ≤ . (1.34)
ζ (2n − 2) ω
0
123
Periodic solutions of nonautonomous ordinary differential equations 247
ω
2ζ (2n − 2) ω 2n−2 (n−1) 2
0 < u − c0 C
2
≤ u (t) dt
ω
ω 2π
0
ω 2n−2
< 2ζ (2n − 2) u (n−1) C
2
ω
, (1.35)
2π
ω ω
(n−1) 1 (n) 1
u Cω ≤ u (t) dt = | p(t)u(t)|dt.
2 2
0 0
ω
u (n−1) C
2
ω
≤ | p(t)|dt,
4
0
where
ω
= | p(t)|u 2 (t)dt.
0
ω
ζ (2n − 2) ω 2n−2
0 < u − c0 C
2
ω
< | p(t)|dt. (1.36)
2 2π
0
ω ω
= (2[σ p(t)]+ − σ p(t)) u (t)dt = 22
[σ p(t)]+ u 2 (t)dt
0 0
ω ω
≤ 2
2uC ω
[σ p(t)]+ dt ≤ 2γ0 ( p)u − c0 C
2
ω
[σ p(t)]+ dt.
0 0
ω 2n−2 ω
γ ( p)ζ (2n − 2) [σ p(t)]+ dt > 1,
2π
0
123
248 I. Kiguradze, A. Lomtatidze
which contradicts (1.34). Thus, (1.10 ) has no nontrivial ω-periodic solution. Therefore,
by virtue of Lemma 1.1, (1.1) has one and only one ω-periodic solution.
2 Nonlinear problem
f (t + ω, x1 , . . . , xn ) ≡ f (t, x1 , . . . , xn ).
(1.10 ) have no nontrivial ω-periodic solution. Then, (2.1) has at least one ω-periodic
solution.
For σ = 1, this lemma is proved in [13]. For σ = −1, the lemma can be proved
analogously.
Theorem 2.1 Let n = 2m and, on the set R × Rn , the inequalities
n
p1 (t)|x1 | − δ t, |xk | ≤ (−1)m f (t, x1 , x2 , . . . , xn ) sgn x1
k=1
n
≤ p2 (t)|x1 | + δ t, |xk | (2.4)
k=1
123
Periodic solutions of nonautonomous ordinary differential equations 249
be fulfilled, where p1 , p2 ∈ L ω ,
ω
p1 (t) ≡ 0, p1 (t)dt ≥ 0, (2.5)
0
and
ω
n−1
π 2π
[ p2 (t)]+ dt ≤ (2.7)
ζ (n) ω
0
Proof By virtue of Lemma 2.1 with σ = (−1)m , it is sufficient to show that for any
p ∈ L ω , satisfying the condition
the result close to Theorem 2.1 is contained in the paper by Mawhin [21] and Mawhin
and Ward [22].
Theorem 2.2 Let, on the set R×Rn , inequalities (2.2) be fulfilled, where p1 , p2 ∈ L ω ,
ω
p2 (t)dt < 0, (2.9)
0
ω
n−1
π 2π
η0 ( p 1 , p 2 ) [ p2 (t)]+ dt ≤ , (2.10)
ζ (n) ω
0
123
250 I. Kiguradze, A. Lomtatidze
ω
n−2
1 2π
η( p1 , p2 ) [ p2 (t)]+ dt ≤ . (2.11)
ζ (2n − 2) ω
0
1
| p(t)| ≤ max {| p1 (t)|, | p2 (t)|} = (| p1 (t)| + | p2 (t)| + || p1 (t)| − | p2 (t)||) .
2
Hence,
γ0 ( p) ≤ η0 ( p1 , p2 ) (2.12)
and
γ ( p) ≤ η( p1 , p2 ). (2.13)
Let us now pass to the case where f (t, x1 , . . . , xn ) ≡ f (t, x1 ), and thus (2.1) has
the form
f (t + ω, x) ≡ f (t, x).
123
Periodic solutions of nonautonomous ordinary differential equations 251
Proof From (2.15) it follows (2.2), where f (t, x1 , . . . , xn ) ≡ f (t, x1 ) and δ(t, ) ≡
| f (t, 0)|. Therefore, by virtue of Theorems 2.1 and 2.2, (2.14) has at least one
ω-periodic solution. To complete the proof of the theorem it remains to show that
this equation has no more then one ω-periodic solution.
Let u 1 and u 2 be any ω-periodic solutions of (2.14). Then the function u(t) =
u 2 (t) − u 1 (t) is an ω-periodic solution of (1.10 ), where
⎧
⎨ f (t,u 2 (t))− f (t,u 1 (t)) if u 2 (t) = u 1 (t),
u 2 (t)−u 1 (t)
p(t) =
⎩σ p (t) if u 2 (t) = u 1 (t).
1
Remark 2.2 In the case when the functions p1 and p2 are not of constant signs, Theo-
rems 2.1–2.3 are new even for the second order equation (see, e.g., [3]). Note also that
if p1 and p2 are of constant signs, then Theorems 2.1–2.3 imply Theorems 2.1–2.4
established in [13].
References
1. Baslandze, S.R., Kiguradze, I.T.: On the unique solvability of a periodic boundary value problem
for third-order linear differential equations (in Russian). Differ. Uravn. 42, 153–158 [English transl.:
Differ. Equ. 42, 165–171 (2006)]
2. Bates, F.W., Ward, J.R. Jr.: Periodic solutions of higher order systems. Pac. J. Math. 84, 275–282 (1979)
3. De Coster, C., Habets, P.: Two-Point Boundary Value Problems: Lower and Upper Functions.
Mathematics in Science and Engineering, vol. 205. Elsevier, Amsterdam (2006)
4. Fučik, S., Mawhin, J.: Periodic solutions of some nonlinear differential equations of higher order. Cas.
Pest. Mat. 100, 276–283 (1975)
5. Gaines, R.E., Mawhin, J.L.: Coincidence Degree, and Nonlinear Differential Equations. Lecture Notes
in Mathematics, vol. 568. Springer, Berlin (1977)
6. Gegelia, G.T.: On bounded and periodic solutions of even-order nonlinear ordinary differential equa-
tions (in Russian). Differ. Uravn. 22, 390–396 [English transl.: Differ. Equ. 22, 278–283 (1986)]
7. Gegelia, G.T.: On periodic solutions of ordinary differential equations. In: Qualitative Theory of Differ-
ential Equations (Szeged 1988), Colloq. Math. Soc. János Bolyai, vol. 53, pp. 211–217. North-Holland,
Amsterdam (1990)
8. Hardy, G.H., Littlewood, J.E., Pólya, G.: Inequalities. Cambridge University Press, London (1951)
9. Kiguradze, I.: Boundary value problems for systems of ordinary differential equations (in Russian).
In: Current Problems in Mathematics. Newest Results, vol. 30, pp. 3–103, Itogi Nauki i Tekhniki,
Akad Nauk SSSR, Vsesoyuz. Inst. Nauchn. i Tekhn. Inform, Moscow [English transl.: J. Sov. Math.
43, 2259–2339 (1988)]
10. Kiguradze, I.: On bounded and periodic solutions of linear higher order differential equations (in
Russian). Mat. Zametki 37, 48–62 (1985)
11. Kiguradze, I.: On periodic solutions of n-th order ordinary differential equations. Nonlinear
Anal. 40, 309–321 (2000)
123
252 I. Kiguradze, A. Lomtatidze
12. Kiguradze, I.T. (2008) On periodic problem at resonance for higher order nonautonomous differential
equations (in Russian). Differ. Uravn. 44 (to appear)
13. Kiguradze, I.T., Kusano, T.: On periodic solutions of higher-order nonautonomous ordinary differential
equations (in Russian). Differ. Uravn. 35, 72–78 [English transl.: Differ. Equ. 35, 71–77 (1999)]
14. Kiguradze, I., Kusano, T.: On conditions for the existence and uniqueness of a periodic solution of
nonautonomous differential equations (in Russian). Differ. Uravn. 36, 1301–1306 [English transl.:
Differ. Equ. 36, 1436–1442 (2000)]
15. Kiguradze, I., Kusano, T.: On periodic solutions of even-order ordinary differential equations. Ann.
Mat. Pura Appl. 180, 285–301 (2001)
16. Kipnis, A.A.: On periodic solutions of higher order nonlinear differential equations (in Russian). Prikl.
Mat. Mekh. 41, 362–365 [English transl.: J. Appl. Math. Mech. 41, 355–358 (1977)]
17. Lasota, A., Opial, Z.: Sur les solutions péridoques des équations différentielles ordinaires. Ann. Polon.
Math. 16, 69–94 (1964)
18. Lasota, A., Szafraniec, F.H.: Sur les solutions périodicues d’une équation différentielle ordinaire
d’ordere n. Ann. Polon. Math. 18, 339–344 (1966)
19. Omari, P., Zanolin, F.: On forced nonlinear oscillations in nth order differential systems with geometric
conditions. Nonlinear Anal. 8, 723–748 (1984)
20. Mawhin, J.: L 2 -estimates and periodic solutions of some nonlinear differential equations. Boll. Un.
Mat. Italy (4), 10, 341–354 (1974)
21. Mawhin, J.: The periodic boundary value problem for some second order ordinary differential equa-
tions. In: Gregus (ed.) Equadiff 5 (Bratislava 1981), pp. 256–259. Teubner, Leipzig (1982)
22. Mawhin, J., Ward, J.R.: Nonuniform nonresonance conditions at the two first eigenvalues for periodic
solutions of forced Lienard and Duffing equations. Rocky Mt. J. Math. 12, 643–654 (1982)
123