Professional Documents
Culture Documents
Basic Theory of Fractional Differential Equations
Basic Theory of Fractional Differential Equations
Basic Theory of Fractional Differential Equations
www.elsevier.com/locate/na
Abstract
In this paper, the basic theory for the initial value problem of fractional differential equations involving Riemann–Liouville
differential operators is discussed employing the classical approach. The theory of inequalities, local existence, extremal solutions,
comparison result and global existence of solutions are considered.
c 2007 Elsevier Ltd. All rights reserved.
1. Introduction
Although the tools of fractional calculus have been available and applicable to various fields of study, the
investigation of the theory of fractional differential equations has only been started quite recently [1,2,5,8]. The
differential equations involving Riemann–Liouville differential operators of fractional order 0 < q < 1, appear to
be important in modelling several physical phenomena [3,4,6,9,10] and therefore seem to deserve an independent
study of their theory parallel to the well-known theory of ordinary differential equations.
In this paper, we shall study systematically the basic theory of fractional differential equations involving
Riemann–Liouville differential operators of order 0 < q < 1. Instead of deducing the basic existence and uniqueness
results from the fixed point theory, we follow the classical proofs of differential equations [7] so that one can compare
and contrast the differences as well as understand the intricacies that might result in the investigation. We shall start
with the fundamental theory of inequalities, which provide necessary comparison results that are useful in further
study of qualitative and quantitative properties of solutions of fractional differential equations. We then prove Peano’s
local existence result and consider the existence of extremal solutions. As an application of the comparison result
developed, we discuss a global existence result in the end. We hope to continue to develop further theory later.
Consider the initial value problem (IVP) for fractional differential equations given by
D q x = f (t, x), x(0) = x0 , (1)
∗ Corresponding author.
E-mail addresses: lakshmik@fit.edu (V. Lakshmikantham), asv5357@louisiana.edu (A.S. Vatsala).
where f ∈ c([0, T ], R), D q x is the fractional derivative of x and q is such that 0 < q < 1. Since f is assumed
continuous, the IVP (1) is equivalent to the following Volterra fractional integral
Z t
1
x(t) = x0 + (t − s)q−1 f (s, x(s))ds, 0 ≤ t ≤ T, (2)
Γ (q) 0
that is, every solution of (2) is also a solution of (1) and vice versa. Here and elsewhere Γ denotes the Gamma function.
Let us first discuss a fundamental result relative to the fractional integral inequalities.
Proof. Suppose that the conclusion (4) is not true. Then, because of the continuity of the functions involved and (3),
it follows that there exists a t1 such that 0 < t1 ≤ T and
v(t1 ) = w(t1 ), v(t) < w(t), 0 < t ≤ t1 . (5)
Let us suppose that the inequality (ii) is strict. Then using the nondecreasing nature of f and (5), we get
Z t1
1
w(t1 ) > w(0) + (t1 − s)q−1 f (s, w(s))ds
Γ (q) 0
Z t1
1
≥ v(0) + (t1 − s)q−1 f (s, v(s))ds ≥ v(t1 )
Γ (q) 0
which is a contradiction in view of (5). Hence the conclusion (4) is valid and the proof is complete.
The next result is for nonstrict inequalities, which requires a one-sided Lipschitz type condition.
Theorem 2.2. Assume that the conditions of Theorem 2.1 hold with nonstrict inequalities (i) and (ii). Suppose further
that
L
f (t, x) − f (t, y) ≤ (x − y), (6)
1 + tq
whenever x ≥ y and L > 0. Then, v(0) ≤ w(0) and L < Γ (q + 1) implies
v(t) ≤ w(t), 0 ≤ t ≤ T. (7)
Proof. Set w (t) = w(t) + (1 + t q ), for small > 0 so that we have,
w (0) = w(0) + > w(0) and w (t) > w(t), 0 ≤ t ≤ T. (8)
Rt
Now w (t) ≥ w(0) + Γ (q) 0 (t − s)
1 q−1 f (s, w(s))ds + (1 + t ). In view of (8), using one-sided Lipschitz condition
q
1
Z t
w (t) > w (0) + (t − s)q−1 f (s, w (s))ds, (10)
Γ (q) 0
in view of the condition L < Γ (q + 1). We now apply Theorem 2.1 to the inequalities (i), (9) and (10) to get
v(t) < w (t), 0 ≤ t ≤ T . since > 0 is arbitrary, we conclude that (7) is true and we are done.
In this section, we shall consider the local existence and the existence of extremal solutions for the IVP (1). Let us
first discuss Peano’s type existence result.
Theorem 3.1. Assume that f ∈ [R0 , R], where R0 = [(t, x) : 0 ≤ t ≤ a and |x − x0 | ≤ b], and let | f (t, x)| ≤ M
on R0 . Then the IVP for fractional differential Eq. (1) possesses at least one solution x(t) on 0 ≤ t ≤ α, where
1
α = min(a, [ M
b
Γ (q + 1)] q ), 0 < q < 1, Γ being the Gamma function as before.
Proof. Let x0 (t) be a continuous function on [−δ, 0], δ > 0 such that x0 (0) = x0 , |x0 (t) − x0 | ≤ b. For 0 < ≤ δ,
we define a function x (t) = x0 (t) on [−δ, 0] and
Z t
1
x (t) = x0 + (t − s)q−1 f (s, x (s − ))ds (12)
Γ (q) 0
on [0, α1 ], where α1 = min(α, ). We observe that
Z t Z t
1 M
|x (t) − x0 | ≤ (t − s)q−1 | f (s, x (s − ))|ds ≤ (t − s)q−1 ds
Γ (q) 0 Γ (q) 0
Mα q
= ≤ b, (13)
Γ (q + 1)
because of the choice of α1 . If α1 < α, we can employ (12) to extend as a continuous function on [−δ, α2 ],
α2 = min(α, 2) such that |x (t) − x0 | ≤ b holds. Continuing this process, we can define x (t) over [−δ, α] so
that |x (t) − x0 | ≤ b is satisfied on [−δ, α]. Furthermore, letting 0 ≤ t1 ≤ t2 ≤ α, we see that
1 t1
Z Z t2
|x (t1 ) − x (t2 )| = (t1 − s)q−1
f (s, x (s − ))ds − (t2 − s)q−1
f (s, x (s − ))ds
Γ (q) 0
0
1 t1
Z Z t2
q−1
− (t2 − s) ] f (s, x (s − ))ds +
q−1
(t2 − s)q−1
f (s, x (s − ))ds
= [(t1 − s)
Γ (q) 0
t1
M t1
Z Z t2
q−1
(t q−1
(t q−1
≤ [(t1 − s) − 2 − s) ]ds + 2 − s) ds
Γ (q) 0
t1
M q q 2M
= [2(t2 − t1 )q + t1 − t2 ] ≤ (t2 − t1 )q < , (14)
Γ (q + 1) Γ (q + 1)
1
provided that |t2 − t1 | < δ0 = [ Γ2M
(q+1) q
] . It then follows from (13) and (14) that the family {x (t)} forms an
equicontinuous and uniformly bounded functions. As application of Ascoli–Arzela’s theorem shows the existence of
a sequence {n } such that 1 > 2 > · · · > n → 0 as n → ∞, and x(t) = limn→∞ xn (t) exists uniformly on
2680 V. Lakshmikantham, A.S. Vatsala / Nonlinear Analysis 69 (2008) 2677–2682
[−δ, α]. Since f is uniformly continuous. we obtain that f (t, xn (t − n )) tends to uniformly to f (t, x(t)) as n → ∞,
and, hence term by term integration of (12) with = n , α1 = α yields
Z t
1
x(t) = x0 + (t − s)q−1 f (s, x(s))ds.
Γ (q) 0
This proves that x(t) is a solution of the IVP (1) and the proof is complete.
Using Theorems 2.1 and 3.1, we can now prove the existence of extremal solutions for the IVP (1).
Theorem 3.2. Under the hypothesis of Theorem 3.1, there exists extremal solutions for the IVP (1) on the interval
1
(q+1) q
0 ≤ t ≤ α0 , α0 = min(a, [ bΓ2M+b ] ), provided f (t, x) is nondecreasing in x for each t.
Proof. We shall prove the existence of maximal solution only, since the case of minimal solution is very similar. Let
0 < ≤ b2 and consider the fractional differential equation with an initial condition
exists on [0, α0 ]. Clearly η(0) = x0 . The uniform continuity of f , gives argument as before (as in Theorem 3.1), that
η(t) is a solution of IVP (1).
Next we show that η(t) is the required maximal solution of (1), on 0 ≤ t ≤ α0 . Let x(t) be any solution of (1) on
0 ≤ t ≤ α0 . Then we have
x0 < x0 + = x(0, ),
V. Lakshmikantham, A.S. Vatsala / Nonlinear Analysis 69 (2008) 2677–2682 2681
1
Z t
x(t) < x0 + (t − s)q−1 [ f (s, x(s)) + ]ds,
Γ (q) 0
Z t
1
x(t, ) ≥ x0 + + (t − s)q−1 [ f (s, x(s, )) + ]ds.
Γ (q) 0
By Theorem 2.1, we get x(t) < x(t, ) on [0, α0 ] for every > 0. The uniqueness of the maximal solution shows that
x(t, ) tends uniformly to η(t) on [0, α0 ] as → 0. The proof is therefore complete.
4. Global existence
Theorem 4.1. Assume that m ∈ c([0, T ], R+ ), g ∈ c([0, T ], R+ ), g(t, u) is nondecreasing in u for each t and
Z t
1
m(t) ≤ m(0) + (t − s)q−1 g(s, m(s))ds, 0 ≤ t ≤ T. (17)
Γ (q) 0
Let η(t) be the maximal solution of
D q u = g(t, u), u(0) = u 0 ≥ 0, (18)
existing on [0, T ) such that m(0) ≤ u(0). Then we have
m(t) ≤ η(t), 0 ≤ t ≤ T. (19)
Proof. In view of the definition of the maximal solution η(t), it is enough to prove, to conclude (19), that
m(t) < u(t, ), 0 ≤ t ≤ T, (20)
where u(t, ) is any solution of
D q u = g(t, u) + , u(0) = u o + , > 0. (21)
Now it follows from (21) that
1
Z t
u(t, ) > u 0 + + (t − s)q−1 g(s, u(s, ))ds.
Γ (q) 0
Then applying Theorem 2.1, we get immediately (20) and since lim→0 u(t, ) = η(t) uniformly on each 0 ≤ t ≤
T0 < T , the proof is complete.
We are now in position to prove global existence result.
Theorem 4.2. Assume that f ∈ c([0, ∞) × R, R), g ∈ c([0, ∞) × R+ , R+ ), g(t, u) is nondecreasing in u for each t
and
| f (t, x)| ≤ g(t, |x|). (22)
Suppose further that we have local existence of solutions x(t, x0 ) of
D q x = f (t, x), x(0) = x0 , (23)
and the maximal solution η(t) of
D q u = g(t, u), u(0) = u 0 ≥ 0,
exists on [0, ∞). Then the largest integral of existence of any solution x(t, x0 ) of (23) such that |x0 | ≤ u 0 is [0, ∞).
Proof. Let x(t, x0 ) be any solution of (23) such that |x0 | ≤ u 0 ; which exists on [0, β) for 0 < β < ∞ and the value
of β cannot be increased further. Set m(t) = |x(t, x0 )| for 0 ≤ t < β. Then using the assumption (22) we get
Z t
1
m(t) ≤ |x0 | + (t − s)q−1 g(s, m(s))ds.
Γ (q) 0
2682 V. Lakshmikantham, A.S. Vatsala / Nonlinear Analysis 69 (2008) 2677–2682
References
[1] M. Caputo, Linear models of dissipation whose Q is almost independent, II, Geophy. J. Roy. Astronom. 13 (1967) 529–539.
[2] W.G. Glöckle, T.F. Nonnenmacher, A fractional calculus approach to self similar protein dynamics, Biophys. J. 68 (1995) 46–53.
[3] K. Diethelm, N.J. Ford, Analysis of fractional differential equations, J. Math. Anal. Appl. 265 (2002) 229–248.
[4] K. Diethelm, N.J. Ford, Multi-order fractional differential equations and their numerical solution, Appl. Math. Comput. 154 (2004) 621–640.
[5] K. Diethelm, A.D. Freed, On the solution of nonlinear fractional differential equations used in the modeling of viscoplasticity, in: F. Keil,
W. Mackens, H. Vob, J. Werther (Eds.), Scientific Computing in Chemical Engineering II: Computational Fluid Dynamics, Reaction
Engineering, and Molecular Properties, Springer, Heidelberg, 1999, pp. 217–224.
[6] V. Kiryakova, Generalized fractional calculus and applications, in: Pitman Res. Notes Math. Ser., vol. 301, Longman-Wiley, New York, 1994.
[7] V. Lakshmikantham, S. Leela, Differential and Integral Inequalities, vol. I, Academic Press, New York, 1969.
[8] R. Metzler, W. Schick, H.G. Kilian, T.F. Nonnenmacher, Relaxation in filled polymers: A fractional calculus approach, J. Chem. Phys. 103
(1995) 7180–7186.
[9] I. Podlubny, Fractional Differential Equations, Academic Press, San Diego, 1999.
[10] S.G. Samko, A.A. Kilbas, O.I. Marichev, Fractional Integrals and Derivatives, Theory and Applications, Gordon and Breach, Yverdon, 1993.