Open Access

Oscillation criteria of fractional differential equations

Advances in Difference Equations20122012:33

DOI: 10.1186/1687-1847-2012-33

Received: 27 November 2011

Accepted: 15 March 2012

Published: 15 March 2012

Abstract

In this article, we are concerned with the oscillation of the fractional differential equation

r ( t ) D - α y η ( t ) - q ( t ) f t ( v - t ) - α y ( v ) d v = 0 for t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equa_HTML.gif

where D - α y https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq1_HTML.gif is the Liouville right-sided fractional derivative of order α (0,1) of y and η > 0 is a quotient of odd positive integers. We establish some oscillation criteria for the equation by using a generalized Riccati transformation technique and an inequality. Examples are shown to illustrate our main results. To the best of author's knowledge, nothing is known regarding the oscillatory behavior of the equation, so this article initiates the study.

MSC (2010): 34A08; 34C10.

Keywords

oscillation fractional derivative fractional differential equation

1 Introduction

The goal of this article is to obtain several oscillation theorems for the fractional differential equation
r ( t ) D - α y η ( t ) - q ( t ) f t ( v - t ) - α y ( v ) d v = 0 for t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equa_HTML.gif
(1.1)
where α (0, 1) is a constant, η > 0 is a quotient of odd positive integers, D - α y https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq1_HTML.gif is the Liou-ville right-sided fractional derivative of order α of y defined by ( D - α y ) ( t ) : = 1 Γ ( 1 - α ) d d t t ( v - t ) - α y ( v ) d v https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq2_HTML.gif for t + := (0, ∞), here Γ is the gamma function defined by Γ ( t ) : = 0 v t - 1 e - v d v https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq3_HTML.gif for t +, and the following conditions are assumed to hold:
  1. (A)

    r and q are positive continuous functions on [t0, ∞) for a certain t0 > 0 and f : is a continuous function such that f(u)/(u η ) ≥ K for a certain constant K > 0 and for all u ≠ 0.

     

By a solution of (1.1) we mean a nontrivial function y C(+,) such that t ( v - t ) - α y ( v ) d v C 1 + , , r ( t ) D - α y η ( t ) C 1 + , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq4_HTML.gif and satisfying (1.1) for t > 0. Our attention is restricted to those solutions of (1.1) which exist on + and satisfy sup{|y(t)| : t > t*} > 0 for any t* ≥ 0. A solution y of (1.1) is said to be oscillatory if it is neither eventually positive nor eventually negative. Otherwise it is nonoscillatory. Equation (1.1) is said to be oscillatory if all its solutions are oscillatory.

The theory of fractional derivatives goes back to Leibniz's note in his list to L'Hospital [1], dated 30 September 1695, in which the meaning of the derivative of order 1/2 is discussed. Leibniz's note led to the appearance of the theory of derivatives and integrals of arbitrary order, which by the end of nineteenth century took more or less finished form due primarily to Liouville, Grünwald, Letnikov, and Riemann. Recently, there have been several books on the subject of fractional derivatives and fractional integrals, such as the books [26].

For three centuries the theory of fractional derivatives developed mainly as a pure theoretical field of mathematics useful only for mathematicians. However, in the last few decades many authors pointed out that fractional derivatives and fractional integrals are very suitable for the description of properties of various real problems.

The mathematical modeling and simulation of systems and processes, based on the description of their properties in term of fractional derivatives, naturally leads to differential equations of fractional order and to the necessity to solve such equations. Fractional differential equations are generalizations of classical differential equations of integer order and have gained considerable popularity and importance during the past three decades or so, due mainly to their demonstrated applications in numerous seemingly diverse and widespread fields of science and engineering. Nowadays the number of scientific and engineering problems involving fractional calculus is already very large and still growing. It was found that various, especially interdisciplinary applications can be elegantly modeled with the help of the fractional derivatives. Fractional differentials and integrals provide more accurate models of systems under consideration. Some of the areas of present applications of fractional calculus include fluid flow, rheology, dynamical processes in self-similar and porous structures, diffusive transport Akin to diffusion, electrical networks, probability and statistics, control theory of dynamical systems, viscoelasticity, electrochemistry of corrosion, chemical physics, optics, and signal processing, economics, and so on; for example, see [712] and the references cited therein.

Many articles have investigated some aspects of fractional differential equations, such as the existence and uniqueness of solutions to Cauchy type problems, the methods for explicit and numerical solutions, and the stability of solutions, and we refer to [1320]. However, to the best of author's knowledge very little is known regarding the oscillatory behavior of fractional differential equations up to now. In particular, nothing is known regarding the oscillation properties of (1.1) up to now. To develop the qualitative theory of fractional differential equations, it is of great interest to study the oscillation of (1.1). In this article, we establish several oscillation criteria for (1.1) by applying a generalized Riccati transformation technique and an inequality. Our results are essential new. We also provide several examples to illustrate the results.

2 Preliminaries and lemmas

In this section, we present the definitions of fractional integrals and fractional derivatives, which are used throughout this article. More details can be found in [26]. We also give several lemmas, which are useful in establishing our results.

There are several kinds of definitions of fractional integrals and fractional derivatives, such as the Riemann-Liouville definition, the Caputo definition, the Liouville definition, the Grünwald-Letnikov definition, the Erdélyi-Kober definition and the Hadamard definition. We adopt the Liouville right-sided definition on the half-axis + for the purpose of this article.

Definition 2.1 (Kilbas et al. [5]) The Liouville right-sided fractional integral of order β > 0 of a function g : + on the half-axis + is given by
I - β g ( t ) : = 1 Γ ( β ) t ( v - t ) β - 1 g ( v ) d v for t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ1_HTML.gif
(2.1)

provided the right-hand side is pointwise defined on + , where Γ is the gamma function.

Definition 2.2 (Kilbas et al. [5]) The Liouville right-sided fractional derivative of order β > 0 of a function g : + on the half-axis + is given by
D - g β ( t ) : = ( - 1 ) β d β d t β I - β - β g ( t ) = ( - 1 ) β 1 Γ β - β d β d t β t ( v - t ) β - β - 1 g ( v ) d v for t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ2_HTML.gif
(2.2)

provided the right-hand side is pointwise defined on +, where β := min{z : zβ} is the ceiling function.

Lemma 2.1 Let y be a solution of (1.1) and
G ( t ) : = t ( v - t ) - α y ( v ) d v f o r α ( 0 , 1 ) a n d t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ3_HTML.gif
(2.3)
then
G ( t ) = - Γ ( 1 - α ) D - α y ( t ) f o r α ( 0 , 1 ) a n d t > 0 . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ4_HTML.gif
(2.4)
Proof From (2.3) and (2.2), for α (0, 1) and t > 0 we obtain
G ( t ) = Γ ( 1 - α ) 1 Γ ( 1 - α ) d d t t ( v - t ) - α y ( v ) d v = - Γ ( 1 - α ) ( - 1 ) α 1 Γ α - α d α d t α t ( v - t ) α - α - 1 y ( v ) d v = - Γ ( 1 - α ) D - α y ( t ) . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equb_HTML.gif

The proof is complete.

Lemma 2.2 (Hardy et al. [21]) If X and Y are nonnegative, then
m X Y m - 1 - X m ( m - 1 ) Y m f o r m > 1 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equc_HTML.gif

where the equality holds if and only if X = Y.

3 Main results

Theorem 3.1 Suppose that (A) and
t 0 r - 1 / η ( t ) d t = https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ5_HTML.gif
(3.1)
hold. Furthermore, assume that there exists a positive function b C1[t0, ∞) such that
lim sup t t 0 t K b ( s ) q ( s ) - r ( s ) b + ( s ) η + 1 ( η + 1 ) η + 1 Γ ( 1 - α ) b ( s ) η d s = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ6_HTML.gif
(3.2)

where b + ( s ) : = max b ( s ) , 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq5_HTML.gif. Then every solution of (1.1) is oscillatory.

Proof Suppose that y is a nonoscillatory solution of (1.1). Without loss of generality, we may assume that y is an eventually positive solution of (1.1). Then there exists t1 [t0, ∞) such that
y ( t ) > 0 and G ( t ) > 0 for t t 1 , , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ7_HTML.gif
(3.3)
where G is defined as in (2.3). Therefore, it follows from (1.1) that
r ( t ) D - α y η ( t ) = q ( t ) f ( G ( t ) ) > 0 for t t 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ8_HTML.gif
(3.4)
Thus r ( t ) D - α y η ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq6_HTML.gif is strictly increasing on [t1,∞) and is eventually of one sign. Since r(t) > 0 for t [t0, ∞) and η > 0 is a quotient of odd positive integers, we see that D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually of one sign. We now claim
D - α y ( t ) < 0 for t t 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ9_HTML.gif
(3.5)
If not, then D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually positive and there exists t2 [t1, ∞) such that D - α y ( t 2 ) > 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq8_HTML.gif. Since r ( t ) D - α y η ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq6_HTML.gif is strictly increasing on [t1, ∞), it is clear that r ( t ) D - α y η ( t ) r ( t 2 ) D - α y η ( t 2 ) : = c 1 > 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq9_HTML.gif for t [t2, ∞). Therefore, from (2.4) we have
- G ( t ) Γ ( 1 - α ) = D - α y ( t ) c 1 1 / η r - 1 / η ( t ) for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equd_HTML.gif
Integrating both sides of the last inequality from t2 to t, we get
t 2 t r - 1 / η ( s ) d s - G ( t ) - G ( t 2 ) c 1 1 / η Γ ( 1 - α ) < G ( t 2 ) c 1 1 / η Γ ( 1 - α ) for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Eque_HTML.gif
Letting t → ∞, we see t 2 r - 1 / η ( s ) d s G ( t 2 ) c 1 1 / η Γ ( 1 - α ) < https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq10_HTML.gif. This contradicts (3.1). Hence, (3.5) holds. Define the function w by the generalized Riccati substitution
w ( t ) = b ( t ) - r ( t ) D - α y η ( t ) G η ( t ) for t t 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ10_HTML.gif
(3.6)
Then we have w(t) > 0 for t [t1,∞). From (3.6), (1.1), (2.4), and (A), it follows that
w ( t ) = b ( t ) - r ( t ) D - α y η ( t ) G η ( t ) + b ( t ) - r ( t ) D - α y η ( t ) G η ( t ) b + ( t ) - r ( t ) D - α y η ( t ) G η ( t ) + b ( t ) - r ( t ) D - α y η ( t ) G η ( t ) + r ( t ) D - α y η ( t ) η G η - 1 ( t ) G ( t ) G 2 η ( t ) = b + ( t ) w ( t ) b ( y ) + b ( t ) - q ( t ) f ( G ( t ) ) G η ( t ) + r ( t ) D - α y η ( t ) η - Γ ( 1 - α ) D - α y ( t ) G η + 1 ( t ) b + ( t ) w ( t ) b ( t ) - K b ( t ) q ( t ) - η Γ ( 1 - α ) b ( t ) r ( t ) w ( t ) b ( t ) r ( t ) 1 + 1 / η = - K b ( t ) q ( t ) + b + ( t ) b ( t ) w ( t ) - η Γ ( 1 - α ) b ( t ) r ( t ) - 1 / η w 1 + 1 / η ( t ) for t t 1 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ11_HTML.gif
(3.7)
where b + https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq11_HTML.gif is defined as in Theorem 3.1. Taking
m = 1 + 1 η , X = η Γ ( 1 - α ) 1 / m w ( t ) b ( t ) r ( t ) 1 / ( η + 1 ) and Y = b + ( t ) / b ( t ) η b ( t ) r ( t ) 1 / m m η η Γ ( 1 - α ) η / m , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equf_HTML.gif
from (3.7) and Lemma 2.2 we conclude that
w ( t ) - K b ( t ) q ( t ) + r ( t ) b + ( t ) η + 1 ( η + 1 ) η + 1 Γ ( 1 - α ) b ( t ) η for t t 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equg_HTML.gif
Integrating both sides of the last inequality from t1 to t, we obtain
t 1 t K b ( s ) q ( s ) - r ( s ) b + ( s ) η + 1 ( η + 1 ) η + 1 Γ ( 1 - α ) b ( s ) η d x w ( t 1 ) - w ( t ) < w ( t 1 ) for t t 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equh_HTML.gif

Letting t → ∞, we get lim sup t t 1 t K b ( s ) q ( s ) - r ( s ) b + ( s ) η + 1 ( η + 1 ) η + 1 Γ ( 1 - α ) b ( s ) η d s w ( t 1 ) < , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq12_HTML.gif which contradicts (3.2). The proof is complete.

Theorem 3.2 Suppose that (A) and (3.1) hold. Furthermore, suppose that there exist a positive function b C1[t0, ∞) and a function H C ( D , ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq13_HTML.gif, where D : = ( t , s ) : t s t 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq14_HTML.gif, such that
H ( t , t ) = 0 f o r t t 0 , H ( t , s ) > 0 f o r ( t , s ) D 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equi_HTML.gif
where D 0 : = ( t , s ) : t > s t 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq15_HTML.gif, and H has a nonpositive continuous partial derivative H s ( t , s ) : = H ( t , s ) s https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq16_HTML.gif on D 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq17_HTML.gif with respect to the second variable and satisfies
lim sup t 1 H ( t , t 0 ) t 0 t 1 [ b ( s ) q ( s ) H ( t , s ) b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 α ) H ( t , s ) ] η ] d s = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ12_HTML.gif
(3.8)

where h + ( t , s ) : = max 0 , H s ( t , s ) + H ( t , s ) b + ( s ) b ( s ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq18_HTML.gif for ( t , s ) D 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq19_HTML.gif, here b + https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq11_HTML.gif is defined as in Theorem 3.1. Then all solutions of (1.1) are oscillatory.

Proof Suppose that y is a nonoscillatory solution of (1.1). Without loss of generality, we may assume that y is an eventually positive solution of (1.1). We proceed as in the proof of Theorem 3.1 to get that (3.7) holds. Multiplying (3.7) by H(t, s) and integrating from t1 to t - 1, for t [t1 + 1, ∞) we obtain
t 1 t - 1 K b ( s ) q ( s ) H ( t , s ) d s - t 1 t - 1 H ( t , s ) w ( s ) d s + t 1 t - 1 H ( t , s ) b + ( s ) b ( s ) w ( s ) d s - t 1 t - 1 H ( t , s ) η Γ ( 1 - α ) [ b ( s ) r ( s ) ] - 1 / η w 1 + 1 / η ( s ) d s . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ13_HTML.gif
(3.9)
Using the integration by parts formula, for t [t1 + 1, ∞) we get
- t 1 t - 1 H ( t , s ) w ( s ) d s = - - H ( t , s ) w ( s ) s = t 1 s = t - 1 + t 1 t - 1 H s ( t , s ) w ( s ) d s < H ( t , t 1 ) w ( t 1 ) + t 1 t - 1 H s ( t , s ) w ( s ) d s . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ14_HTML.gif
(3.10)
Substituting (3.10) in (3.9), for t [t1 + 1, ∞) we have
K t 1 t - 1 b ( s ) q ( s ) H ( t , s ) d s H ( t , t 1 ) w ( t 1 ) + t 1 t - 1 H s ( t , s ) + H ( t , s ) b + ( s ) b ( s ) w ( s ) - η Γ ( 1 - α ) H ( t , s ) [ b ( s ) r ( s ) ] 1 / η w 1 + 1 / η ( s ) d s H ( t , t 1 ) w ( t 1 ) + t 1 t - 1 h + ( t , s ) w ( s ) - η Γ ( 1 - α ) H ( t , s ) [ b ( s ) r ( s ) ] 1 / η w 1 + 1 / η ( s ) d s , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ15_HTML.gif
(3.11)
where h+ is defined as in Theorem 3.2. Taking
m = 1 + 1 η , X = η Γ ( 1 - α ) H ( t , s ) 1 / m w ( s ) [ b ( s ) r ( s ) ] 1 / ( η + 1 ) and Y = h + η ( t , s ) [ b ( s ) r ( s ) ] 1 / m m η η Γ ( 1 - α ) H ( t , s ) η / m , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equj_HTML.gif
by using Lemma 2.2 in (3.11) we obtain for t [t1 + 1, ∞),
t 1 t - 1 b ( s ) q ( s ) H ( t , s ) d s K - 1 H ( t , t 1 ) w ( t 1 ) + K - 1 t 1 t - 1 b ( s ) r ( s ) h + η ( t , s ) ( η + 1 ) η + 1 [ Γ ( 1 - α ) H ( r , s ) ] η d s . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ16_HTML.gif
(3.12)
Since H s ( t , s ) 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq20_HTML.gif for t > st0, we have 0 > H(t,t1) ≤ H(t,t0) for t > t1t0. Therefore, from (3.12) we get for t [t1 + 1, ∞),
t 1 t - 1 b ( s ) q ( s ) H ( t , s ) - b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 - α ) H ( t , s ) ] η d s K - 1 H ( t , t 1 ) w ( t 1 ) K - 1 H ( t , t 0 ) w ( t 1 ) . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ17_HTML.gif
(3.13)
Since 0 < H(t, s) ≤ H(t, t0) for t > st0, we have 0 < H ( t , s ) H ( t , t 0 ) 1 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq21_HTML.gif for t > st0. Hence, it follows from (3.13) that
1 H ( t , t 0 ) t 0 t - 1 b ( s ) q ( s ) H ( t , s ) - b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 - α ) H ( t , s ) ] η d s = 1 H ( t , t 0 ) t 0 t 1 b ( s ) q ( s ) H ( t , s ) - b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 - α ) H ( t , s ) ] η d s + 1 H ( t , t 0 ) t 1 t - 1 b ( s ) q ( s ) H ( t , s ) - b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 - α ) H ( t , s ) ] η d s 1 H ( t , t 0 ) t 0 t 1 b ( s ) q ( s ) H ( t , s ) d s + 1 H ( t , t 0 ) K - 1 H ( t , t 0 ) w ( t 1 ) t 0 t 1 b ( s ) q ( s ) d s + K - 1 w ( t 1 ) for t t 1 + 1 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equk_HTML.gif
Letting t → ∞, we have
lim sup t 1 H ( t , t 0 ) t 0 t 1 [ b ( s ) q ( s ) H ( t , s ) b ( s ) r ( s ) h + η ( t , s ) K ( η + 1 ) η + 1 [ Γ ( 1 α ) H ( t , s ) ] η ] d s t 0 t 1 b ( s ) q ( s ) d s + K 1 w ( t 1 ) < , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equl_HTML.gif

which is a contradiction to (3.8). The proof is complete.

Next, we consider the case
t 0 r - 1 / η ( t ) d t < , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ18_HTML.gif
(3.14)

which yields (3.1) doesn't hold. In this case, we have the following results.

Theorem 3.3 Suppose that (A) and (3.14) hold and that there exists a positive function b C1[t0,∞) such that (3.2) holds. Furthermore, assume that for every constant Ct0,
C 1 r ( t ) C t q ( s ) d s 1 / η d t = . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ19_HTML.gif
(3.15)

Then every solution y of (1.1) is oscillatory or satisfies lim t t ( v - t ) - α y ( v ) d v = 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq22_HTML.gif.

Proof Assume that y is a nonoscillatory solution of (1.1). Without loss of generality, assume that y is an eventually positive solution of (1.1). Proceeding as in the proof of Theorem 3.1, we obtain that (3.3) and (3.4) hold. Then there are two cases for the sign of D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif. The proof when D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually negative is similar to that of Theorem 3.1 and hence is omitted.

Next, assume that D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually positive. Then there exists t2t1 such that D - α y ( t ) > 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq23_HTML.gif for tt2. From (2.4) we get G'(t) < 0 for tt2. Thus we get limt→∞G(t) := L ≥ 0 and G(t) ≥ L. We now claim L = 0. Assume not, i.e., L > 0, then from (3.4) we get
r ( t ) D - α y η ( t ) = q ( t ) f ( G ( t ) ) q ( t ) K G η ( t ) K L η q ( t ) for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equm_HTML.gif
Integrating both sides of the last inequality from t2 to t, we have
r ( t ) D - α y η ( t ) r ( t 2 ) D - α y η ( t 2 ) + K L η t 2 t q ( s ) d s > K L η t 2 t q ( s ) d s for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equn_HTML.gif
Hence, from (2.4) we get
- G ( t ) Γ ( 1 - α ) = D - α y ( t ) > K 1 / η L 1 r ( t ) t 2 t q ( s ) d s 1 / η for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equo_HTML.gif
Integrating both sides of the last inequality from t2 to t, we obtain
G ( t ) G ( t 2 ) - Γ ( 1 - α ) K 1 / η L t 2 t 1 r ( u ) t 2 u q ( s ) d s 1 / η d u for t t 2 , . https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equp_HTML.gif

Letting t → ∞, from (3.15) we get limt→∞G(t) = -∞. This contradicts (3.3). Therefore, we have L = 0, i.e., limt→∞G(t) = 0. In view of (2.3), we see that the proof is complete.

Theorem 3.4 Suppose that (A) and (3.14) hold. Let b(t) and H(t,s) be defined as in Theorem 3.2 such that (3.8) holds. Furthermore, assume that for every constant Ct0, (3.15) holds. Then every solution y of (1.1) is oscillatory or satisfies lim t t ( v - t ) - α y ( v ) d v = 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq22_HTML.gif.

Proof Assume that y is a nonoscillatory solution of (1.1). Without loss of generality, assume that y is an eventually positive solution of (1.1). Proceeding as in the proof of Theorem 3.1, we obtain that (3.3) and (3.4) hold. Then there are two cases for the sign of D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif. The proof when D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually negative is similar to that of Theorem 3.2 and hence is omitted. The proof when D - α y ( t ) https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq7_HTML.gif is eventually positive is similar to that of the proof of Theorem 3.3 and thus is omitted. The proof is complete.

Remark 3.1 From Theorems 3.1-3.4, we can derive many different sufficient conditions for the oscillation of (1.1) with different choices of the functions b and H.

4 Examples

Example 4.1 Consider the fractional differential equation
t η - 1 D - α y η ( t ) - 1 t 2 t ( v - t ) - α y ( v ) d v η = 0 , t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ20_HTML.gif
(4.1)
where α (0,1), η > 0 is a quotient of odd positive integers and (η + 1)η+1[Γ(1 - α)] η > 1. In (4.1), r(t) = tη-1, q ( t ) = 1 t 2 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq24_HTML.gif and f(u) = u η . Take t0 > 0 and K = 1. Since
t 0 r - 1 / η ( t ) d t = t 0 1 t 1 - 1 η d t = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equq_HTML.gif
we find that (A) and (3.1) hold. We will apply Theorem 3.1, and it remains to satisfy the condition (3.2). Taking b(s) = s, we obtain
lim sup t t 0 t [ K b ( s ) q ( s ) r ( s ) [ b + ' ( s ) ] η + 1 ( η + 1 ) η + 1 [ Γ ( 1 α ) b ( s ) ] η ] d s = lim sup t t 0 t 1 s [ 1 1 ( η + 1 ) η + 1 [ Γ ( 1 α ) ] η ] d s = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equr_HTML.gif

which implies that (3.2) holds. Therefore, by Theorem 3.1 every solution of (4.1) is oscillatory.

Example 4.2 Consider the fractional differential equation
t η + 1 D - α y η ( t ) - 2 t 3 + exp t ( v - t ) - α y ( v ) d v t ( v - t ) - α y ( v ) d v η = 0 , t > 0 , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equ21_HTML.gif
(4.2)
where α (0,1) and η > 0 is a quotient of odd positive integers. In (4.2), r(t) = tη+1, q(t) = 2t and f(u) = (3 + e u )u η . Take t0 > 0 and K = 3. Since
t 0 r - 1 / η ( t ) d t = t 0 1 t 1 + 1 η d t < , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equs_HTML.gif
we see that (A) and (3.14) hold 3.3, it remains. To apply Theorem to satisfy the conditions (3.2) and (3.15). Take b(s) = 1, then we get
lim sup t t 0 t [ K b ( s ) q ( s ) r ( s ) [ b + ' ( s ) ] η + 1 ( η + 1 ) η + 1 [ Γ ( 1 α ) b ( s ) ] η ] d s = lim sup t t 0 t 6 s d s = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equt_HTML.gif
which implies that (3.2) holds. For every constant Ct0, we can find 0 < M < 1 and t M C such that t - CMt for t [t M , ∞). Thus, we conclude
C 1 r ( t ) C t q ( s ) d s 1 / η d t = C 1 t η + 1 C t 2 s d s 1 / η d t = C t 2 - C 2 t η + 1 1 / η d t = C ( t + C ) ( t - C ) t η + 1 1 / η d t ( 2 C M ) 1 η t M 1 t d t = , https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_Equu_HTML.gif

which yields that (3.15) holds. Hence, by Theorem 3.3 every solution y of (4.2) is oscillatory or satisfies lim t t ( v - t ) - α y ( v ) d v = 0 https://static-content.springer.com/image/art%3A10.1186%2F1687-1847-2012-33/MediaObjects/13662_2011_Article_146_IEq22_HTML.gif.

Declarations

Acknowledgements

This study was supported by Hunan Provincial Natural Science Foundation of P. R. China (Grant No. 11JJ3010) and by the Science and Technology Program of Hunan Province of P. R. China (Grant No. 2010FJ6021). D.-X. Chen wish to thank the anonymous referees for their valuable suggestions and comments, which helped the author to improve the previous manuscript of the article.

Authors’ Affiliations

(1)
College of Science, Hunan Institute of Engineering

References

  1. Leibniz GW: Mathmatische Schiften. Grorg Olms Verlagsbuchhandlung, Hildesheim; 1962.
  2. Oldham KB, Spanier J: The Fractional Calculus. Academic Press, San Diego; 1974.
  3. Miller KS, Ross B: An Introduction to The Fractional Calculus and Fractional Differential Equations. John Wiley & Sons, New York; 1993.
  4. Podlubny I: Fractional Differential Equations. Academic Press, San Diego; 1999.
  5. Kilbas AA, Srivastava HM, Trujillo JJ: Theory and Applications of Fractional Differential Equations. Elsevier Science BV, Amsterdam; 2006.
  6. Das S: Functional Fractional Calculus for System Identification and Controls. Springer, New York; 2008.
  7. Bagley RL, Torvik PJ: A theoretical basis for the application of fractional calculus to viscoelasticity. J Rheol 1983, 27: 201–210. 10.1122/1.549724View Article
  8. Mainardi F: Fractional calculus: Some basic problems in continuum and statistical mechanics. In Fractals and Fractional Calculus in Continuum Mechanics. Edited by: Carpinteri, A, Mainardi, F. Springer, New York; 1997:291–348.View Article
  9. Mandelbrot B: Some noises with 1/f spectrum, a bridge between direct current and white noise. IEEE Trans Inf Theory 1967, 13: 289–298.View Article
  10. Rossikhin YA, Shitikova MV: Applications of fractional calculus to dynamic problems of linear and nonlinear hereditary mechanics of solids. Appl Mech Rev 1997, 50: 15–67. 10.1115/1.3101682View Article
  11. Baillie RT: Long memory processes and fractional integration in econometrics. J Econ 1996, 73: 5–59.MathSciNetView Article
  12. Magin RL: Fractional calculus in bioengineering. Crit Rev Biomed Eng 2004, 32: 1–377. 10.1615/CritRevBiomedEng.v32.10View Article
  13. Saadatmandi A, Dehghan M: A new operational matrix for solving fractional-order differential equations. Comput Math Appl 2010, 59: 1326–1336. 10.1016/j.camwa.2009.07.006MathSciNetView Article
  14. Zhou Y, Jiao F, Li J: Existence and uniqueness for p -type fractional neutral differential equations. Nonlinear Anal 2009, 71: 2724–2733. 10.1016/j.na.2009.01.105MathSciNetView Article
  15. Galeone L, Garrappa R: Explicit methods for fractional differential equations and their stability properties. J Comput Appl Math 2009, 228: 548–560. 10.1016/j.cam.2008.03.025MathSciNetView Article
  16. Trigeassou JC, Maamri N, Sabatier J, Oustaloup A: A Lyapunov approach to the stability of fractional differential equations. Signal Process 2011, 91: 437–445. 10.1016/j.sigpro.2010.04.024View Article
  17. Deng W: Smoothness and stability of the solutions for nonlinear fractional differential equations. Nonlinear Anal 2010, 72: 1768–1777. 10.1016/j.na.2009.09.018MathSciNetView Article
  18. Ghoreishi F, Yazdani S: An extension of the spectral Tau method for numerical solution of multi-order fractional differential equations with convergence analysis. Comput Math Appl 2011, 61: 30–43. 10.1016/j.camwa.2010.10.027MathSciNetView Article
  19. Edwards JT, Ford NJ, Simpson AC: The numerical solution of linear multi-term fractional differential equations: systems of equations. J Comput Appl Math 2002, 148: 401–418. 10.1016/S0377-0427(02)00558-7MathSciNetView Article
  20. Muslim M: Existence and approximation of solutions to fractional differential equations. Math Comput Model 2009, 49: 1164–1172. 10.1016/j.mcm.2008.07.013MathSciNetView Article
  21. Hardy GH, Littlewood JE, Pólya G: Inequalities. 2nd edition. Cambridge University Press, Cambridge; 1988.

Copyright

© Chen; licensee Springer. 2012

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://​creativecommons.​org/​licenses/​by/​2.​0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.