Topological Methods in Nonlinear Analysis

Volume 46, No. 1, 2015, 69–92

c 2015 Juliusz Schauder Centre for Nonlinear Studies

Nicolaus Copernicus University

THE BLOW-UP AND GLOBAL EXISTENCE

OF SOLUTIONS OF CAUCHY PROBLEMS
FOR A TIME FRACTIONAL DIFFUSION EQUATION

Quan-Guo Zhang — Hong-Rui Sun

Abstract. In this paper, we investigate the blow-up and global existence

of solutions to the following time fractional nonlinear diffusion equations
(
C D α u − 4u = |u|p−1 u, x ∈ RN , t > 0,
0 t
u(0, x) = u0 (x), x ∈ RN ,
where 0 < α < 1, p > 1, u0 ∈ C0 (RN ) and C α
0 Dt u =
1−α 1−α
(∂/∂t)0 It (u(t, x) − u0 (x)), 0 It denotes left Riemann–Liouville frac-
tional integrals of order 1− α. We prove that if 1 < p < 1 + 2/N , then every
nontrivial nonnegative solution blow-up in finite time, and if p ≥ 1 + 2/N
and ku0 kLqc (RN ) , qc = N (p − 1)/2 is sufficiently small, then the problem
has global solution.

1. Introduction
This paper is concerned with the blow-up and global existence of solutions
to the following Cauchy problems for time fractional diffusion equation
C α
(1.1) 0 Dt u − 4u = |u|p−1 u, x ∈ RN , t > 0,
(1.2) u(0, x) = u0 (x), x ∈ RN ,

2010 Mathematics Subject Classification. 26A33, 35K15.

Key words and phrases. Fractional differential equation, blow-up, global existence, Cauchy
problems.
Supported by the program for New Century Excellent Talents in University(NECT-12-
0246) and FRFCU(lzujbky-2013-k02) and FRFCU(lzujbky-2015-75).

69
70 Q.G. Zhang — H.R. Sun
n o
where 0 < α < 1, p > 1, u0 ∈ C0 (RN ) = u ∈ C(RN ) lim u(x) = 0 and
|x|→∞
C α
0 Dt u = (∂/∂t)0 It1−α (u(t, x) − u0 (x)), 0 It1−α denotes left Riemann–Liouville
fractional integrals of order 1 − α and is defined by
Z t
1
I
0 t
1−α
u = (t − s)−α u(s) ds.
Γ(1 − α) 0
When α = 1, the problem (1.1)–(1.2) reduces to the semilinear heat equation

(1.3) ut − 4u = |u|p−1 u, x ∈ RN , t > 0

with (1.2). In his pioneering article [9], Fujita showed that if 1 < p < 1 + 2/N
and u0 6≡ 0, then every solution of (1.3)–(1.2) blows up in a finite time. If
p > 1 + 2/N , then for initial values bounded by a sufficiently small Gaussian,
that is for τ > 0, there is ε = ε(τ ) > 0 such that if 0 ≤ u0 (x) ≤ εGτ (x), then the
solution of (1.3)–(1.2) is global. The critical case p = 1+2/N was later proved to
be in the blow-up category [10], [13] (see also [27], [1]). In [27], Weissler proved
that if the initial value u0 is small enough in Lqc (RN ), qc = N (p − 1)/2 > 1,
then the solution of (1.3)–(1.2) exists globally.
In [12], Kirane, Laskyi and Tatar studied the following evolution problem
C α
(1.4) 0 Dt u + (−4)β/2 u = h(x, t)|u|1+ep , x ∈ RN , t > 0,
1−α
with (1.2), where 0 < α < 1, 0 < β ≤ 2, C α
0 Dt u = (∂/∂t)0 It (u(t, x) − u0 (x)),
σ ρ N
pe > 0, h satisfies h(x, t) ≥ Ch |x| t for x ∈ R , t > 0, Ch > 0, and σ,
ρ satisfy some conditions. (−4)β/2 u = F −1 (|ξ|β F (u)), where F denotes
Fourier transform and F −1 denotes its inverse. They obtained that if 0 <
pe ≤ (α(β + σ) + βρ)/(αN + β(1 − α)), then the problem (1.4)–(1.2) admits no
global weak nonnegative solution other than the trivial one.
In [4], Cazenave, Dickstein and Weissler considered the following heat equa-
tion with nonlinear memory,
Z t
(1.5) ut − 4u = (t − s)−γ |u|p−1 uds, x ∈ RN , t > 0
0

with (1.2), where p > 1, 0 ≤ γ < 1 and u0 ∈ C0 (RN ).

Let pγ = 1 + 2(2 − γ)/(N − 2 + 2γ)+ , (N − 2 + 2γ)+ = max{0, N − 2 + 2γ}
and p∗ = max{1/γ, pγ } ∈ (0, ∞]. They obtained that if p ≤ p∗ , u0 ≥ 0, u0 6≡ 0,
then the solution u of (1.5)–(1.2) blows up in finite time and if p > p∗ and
u0 ∈ Lqsc (RN ), qsc = N (p − 1)/(4 − 2γ) with ku0 kLqsc (RN ) sufficiently small,
then the solution exists globally.
In [8], Fino and Kirane discussed the following equation
Z t
(1.6) β/2
ut + (−4) u = (t − s)−γ |u|p−1 u ds, x ∈ RN , t > 0
0
 Blow-up and Global Existence for Fractional equation 71

with (1.2), 0 < β ≤ 2, they got the blow-up and global existence results by
using the test function method. The method based on rescalings of a compactly
support test function to prove the blow-up results which is used by Mitidieri and
Pohozaev [19] to show the blow-up results.
Fractional Cauchy problems are useful to model anomalous diffusion, de-
scribe Hamiltonian chaos, etc. see [21], [25], [28], [18] and references therein. We
refer to several works on the mathematical treatments for time fractional diffu-
sion equation. Eidelman and Kochubei [7] studied an evolution equation with
time fractional and a uniformly elliptic operator with variable coefficients in the
spatial variables, where the fundamental solution was constructed and investi-
gated. In [14], Kochubei considered the Cauchy problem for a linear evolution
systems of partial differential equations with the fractional derivative in the time
variable and constructed and investigated the Green matrix of the Cauchy prob-
lem. In [24], [17], the existence and properties of solutions for a time fractional
equation in a bounded domain were considered by applying eigenfunction ex-
pansions. Recently, there are many papers about the existence and properties of
solutions for the fractional abstract Cauchy problem, see for example [26], [15],
[29], [20] and the references therein.
Motivated by the above results, in this paper, we study the problem (1.1)–
(1.2). Fujita exponent is determined. We will show that
(i) For u0 ∈ C0 (RN ), u0 ≥ 0, and u0 6≡ 0, if 1 < p < 1 + 2/N, then the
solution of (1.1)–(1.2) blows up in finite time.
(ii) For u0 ∈ C0 (RN ) ∩ Lqc (RN ), where qc = N (p − 1)/2, if p ≥ 1 + 2/N and
ku0 kLqc is sufficiently small, then (1.1)–(1.2) has a global solution.
Compare with the classical results of heat equation (1.3)–(1.2), the major
difference between the time fractional equation (1.1)–(1.2) and the heat equation
(1.3)–(1.2) is that in critical case, that is p = 1 + 2/N , the solution of (1.1)–(1.2)
could exist globally.
In [4], the authors inferred that for (1.6)–(1.2), the Fujita critical exponent is
not the one which would be predicted from the scaling properties of the equation,
which is different from the heat equation (1.3)–(1.2). But for (1.1), despite it
is also nonlocal about t, we can also obtain the Fujita critical exponent by the
scaling properties of the equation (1.1). In fact, if u(t, x) is a solution of (1.1)
with the initial value u0 (x), then, for every λ > 0, λ2α/(p−1) u(λ2 t, λα x) is also
a solution of (1.1) with initial value λ2α/(p−1) u0 (λα x). Since

(1.7) kλ2α/(p−1) u0 (λα · )kLq (RN ) = λ2α/(p−1)−αN /q ku0 kLq (RN ) ,

it follows that the invariant Lebesgue norm for (1.7) is given by qc = N (p − 1)/2.
If qc > 1, then p > 1 + 2/N . Therefore, one predicts 1 + 2/N is the Fujita critical
72 Q.G. Zhang — H.R. Sun

exponent. Our main results show 1 + 2/N is the Fujita critical exponent for the
problem (1.1)–(1.2).
This paper is organized as follows. In Section 2, some preliminaries are
presented. In Section 3, the local existence and uniqueness of mild solution of
the problem (1.1)–(1.2) are established. In Section 4, we show that the blow-up
and global existence of the solutions to the problem (1.1)–(1.2).

2. Preliminaries
In this section, we present some preliminaries that will be used in the next
sections.
First, we list some properties of two special functions. The Mittag–Leffler
function is defined for complex z ∈ C in [11], [23]
∞
X zk
Eα,β (z) = , α, β ∈ C, Re(α) > 0
Γ(αk + β)
k=0

and its Riemann–Liouville fractional integral satisfies

1−α α−1
0 It (t Eα,α (λtα )) = Eα,1 (λtα ) for λ ∈ C, 0 < α < 1.

We also need the following Wright type function which was considered by
Mainardi [16] (see also [23])
∞
X (−z)k
(2.1) φα (z) =
k!Γ(−αk + 1 − α)
k=0
∞
1 X (−z)k Γ(α(k + 1)) sin(π(k + 1)α)
=
π k!
k=0

for 0 < α < 1. φα is an entire function and has the following properties.
Z ∞
(a) φα (θ) ≥ 0 for θ ≥ 0 and φα (θ) dθ = 1.
0
Z ∞
Γ(1 + r)
(b) φα (θ)θr dθ = for r > −1.
0 Γ(1 + αr)
Z ∞
(c) φα (θ)e−zθ dθ = Eα,1 (−z), z ∈ C.
0
Z ∞
(d) α θφα (θ)e−zθ dθ = Eα,α (−z), z ∈ C.
0
If C α 1 1
0 Dt f ∈ L (0, T ), g ∈ C ([0, T ]) and g(T ) = 0, then we have the following
formula of integration by parts
Z T Z T
C α
(2.2) g 0 Dt f dt = (f (t) − f (0)) C α
t DT g dt,
0 0
 Blow-up and Global Existence for Fractional equation 73

C α d 1−α
where t DT g = − dt t IT g,
Z T
1
1−α
t IT g = (s − t)−α g(s) ds.
Γ(1 − α) t

We need know the Caputo fractional derivative of the following function,

which will be used in the next sections. For given T > 0 and n > 0, if we let

(1 − t/T )n , t ≤ T,
ϕ(t) =
0, t > T,

then
 n−α
Γ(n + 1) t
C α
t DT ϕ(t) = T −α 1 − , t ≤ T,
Γ(n + 1 − α) T
(see for example [11]).
We denote A = 4 and it generates a semigroup {T (t)}t≥0 on C0 (RN ) with
domain
D(A) = {u ∈ C0 (RN ) | 4u ∈ C0 (RN )}.
Then T (t) is an analytic and contractive semigroup on C0 (RN ) [3], [22] and, for
t > 0, x ∈ RN ,
Z
1 2
(2.3) T (t)u0 = G(t, x − y)u0 (y) dy, G(t, x) = N/2
e−|x| /(4t) ,
RN (4πt)
and T (t) is a contractive semigroup on Lq (RN ) for q ≥ 1 [5], and

(2.4) kT (t)u0 kLp (RN ) ≤ (4πt)−(N/2)(1/q−1/p) ku0 kLq (RN )

for u0 ∈ Lq (RN ), q ≤ p ≤ +∞.

Define the operators Pα (t) and Sα (t) as
Z ∞
(2.5) Pα (t)u0 = φα (θ)T (tα θ)u0 dθ, t ≥ 0,
0
Z ∞
(2.6) Sα (t)u0 = α θφα (θ)T (tα θ)u0 dθ, t ≥ 0.
0

Consider the following linear time fractional equation


C Dα u − 4u = f (t, x), x ∈ RN , t > 0,
0 t
(2.7)
u(0, x) = u0 (x), x ∈ RN ,

where u0 ∈ C0 (RN ) and f ∈ L1 ((0, T ), C0 (RN )). If u is a solution of (2.7), then

by [2] (see also [26]), we get
Z t
u(t, x) = Pα (t)u0 + (t − s)α−1 Sα (t − s)f (s, x) ds,
0

where Pα (t) and Sα (t) are given by (2.5) and (2.6), respectively.
74 Q.G. Zhang — H.R. Sun

We denote
Z ∞
K(t, x) = φα (θ)G(tα θ, x) dθ, x ∈ RN \ {0}, t > 0.
0

Note that for given t > 0 and x ∈ RN \ {0}, G(tα θ, x) → 0 as θ → 0, so K is

R∞ R
well defined. Since 0 φα (θ) dθ = 1, RN G(t, x) dx = 1, we know that

kK(t, · )kL1 (RN ) = 1 for t > 0.

Lemma 2.1. The operator {Pα (t)}t>0 has the following properties
(a) If u0 ≥ 0, u0 6≡ 0, then Pα (t)u0 > 0 and kPα (t)u0 kL1 (RN ) = ku0 kL1 (RN ) .
(b) If 1 ≤ p ≤ q ≤ +∞ and 1/r = 1/p − 1/q < 2/N , then
Γ(1 − N/(2r))
(2.8) kPα (t)u0 kLq (RN ) ≤ (4πtα )−N/(2r) ku0 kLp (RN ) .
Γ(1 − αN /(2r))
Proof. (a) follows from T (t)u0 > 0, φα ≥ 0 and kK(t, · )kL1 (RN ) = 1.
For (b), by (2.4) and the properties of φα , we have
Z +∞ Z +∞
α
φα (θ)T (t θ)u0 dθ ≤ φα (θ)(4πtα θ)−N/(2r) ku0 kLp (RN ) dθ
0 Lq (RN ) 0
Z +∞
α −N/(2r)
= (4πt ) φα (θ)θ−N/(2r) dθku0 kLp (RN )
0
Γ(1 − N/(2r))
= (4πtα )−N/(2r) ku0 kLp (RN ) .
Γ(1 − αN /(2r))
Hence, we derive (2.8) holds. 

Lemma 2.2. For the operator {Sα (t)}t>0 , we have the following results.
(a) If u0 ≥ 0 and u0 6≡ 0, then
1
Sα (t)u0 > 0 and kSα (t)u0 kL1 (RN ) = ku0 kL1 (RN ) .
Γ(α)
(b) For 1 ≤ p ≤ q ≤ +∞, let 1/r = 1/p − 1/q, if 1/r < 4/N , then
Γ(2 − N/(2r))
(2.9) kSα (t)u0 kLq (RN ) ≤ α(4πtα )−N/(2r) ku0 kLp (RN ) .
Γ(1 + α − αN /(2r))
Proof. The proof is similar to that of Lemma 2.1, so we omit it. 

Lemma 2.3. For u0 ∈ C0 (RN ), we have Pα (t)u0 ∈ D(A) for t > 0, and
C α
0 Dt Pα (t)u0 = APα (t)u0 , t > 0,
C
kAPα (t)u0 kL∞ (RN ) ≤ ku0 kL∞ (RN ) , t > 0,
tα
for some constant C > 0.
 Blow-up and Global Existence for Fractional equation 75

Proof. Let X = C0 (RN ). First, we prove if u0 ∈ X, then Pα (t)u0 ∈ D(A).

In fact, for u0 ∈ X,
Z ∞ Z 1
α
Pα (t)u0 = φα (θ)T (t θ)u0 dθ = (φα (θ) − φα (0))T (tα θ)u0 dθ
0 0
Z 1 Z ∞
+ φα (0) T (tα θ)u0 dθ + φα (θ)T (tα θ)u0 dθ.
0 1
R1
Clearly, 0
T (tα θ)u0 dθ ∈ D(A). Note that there exists positive constant C such
that
ku0 kX
kAT (tα θ)u0 kX ≤ C , t > 0, θ > 0,
tα θ
we get that
Z ∞
φα (θ)T (tα θ)u0 dθ ∈ D(A).
1
Next, we show that
Z 1
(φα (θ) − φα (0))T (tα θ)u0 dθ ∈ D(A).
0

In fact, for every h > 0,

 Z 1 Z 1 
1
T (h) (φα (θ) − φα (0))T (tα θ)u0 dθ − (φα (θ) − φα (0))T (tα θ)u0 dθ
h 0 0
1 1
Z
= (φα (θ) − φα (0))(T (tα θ + h) − T (tα θ))u0 dθ.
h 0
Since
(T (tα θ + h) − T (tα θ))u0 C φα (θ) − φα (0)
≤ ku0 kX , ≤ C,
h X tα θ θ
for some constant C > 0 independent of θ and h, so, by dominated convergence
theorem, we know
Z 1
(φα (θ) − φα (0))T (tα θ)u0 dθ ∈ D(A).
0

Note that
Z 1
APα (t)u0 = A (φα (θ) − φα (0))T (tα θ)u0 dθ
0
Z 1 Z ∞
+ φα (0)A T (tα θ)u0 dθ + A φα (θ)T (tα θ)u0 dθ
0 1
Z 1
= (φα (θ) − φα (0))AT (tα θ)u0 dθ
0
Z ∞
φα (0)(T (tα )u0 − u0 )
+ + φα (θ)AT (tα θ)u0 dθ.
tα 1
76 Q.G. Zhang — H.R. Sun

Therefore
C
(2.10) kAPα (t)u0 kX ≤ ku0 kX
tα
for some positive constant C.
By dominated convergence theorem, we obtain that for u0 ∈ X,
d
Pα (t)u0 = tα−1 ASα (t)u0 , t > 0.
dt
Furthermore, if u0 ∈ D(A), then
d
Pα (t)u0 = tα−1 Sα (t)Au0 , t > 0.
dt
Since
Z ∞
t
sα−1
Z
1−α α−1 1
0 It (t Sα (t)Au0 ) = α
αθφα (θ)T (sα θ)Au0 dθ ds,
Γ(1 − α)
0 (t − s) 0
Z ∞ Z
1
α
αθφα (θ)T (s θ)Au0 dθ = Eα,α (λsα )(λ − A)−1 Au0 dλ,
0 2πi Γ0
where Γ0 is a path composed from two rays ρeiτ , ρ ≥ 1, π/2 < τ < π and ρe−iτ
and a curve eiβ , −τ ≤ β ≤ τ,
1−α α−1
0 It (t Eα,α (λtα )) = Eα,1 (λtα ),
so,
1−α α−1
(2.11) 0 It (t Sα (t)Au0 ) = Pα (t)Au0 = APα (t)u0 .
Therefore, we get C α
0 Dt Pα (t)u0 = APα (t)u0 if u0 ∈ D(A), t > 0. Next, we prove
that the conclusion also holds if u0 ∈ X.
In fact, if u0 ∈ X, then we can find {u0,n } ⊂ D(A) such that u0,n → u0
in X. By (2.11) and Lemma 2.1, we know
C α
0 Dt Pα (t)u0,n = APα (t)u0,n and kPα (t)u0,n kX ≤ ku0,n kX .
We denote un = Pα (t)u0,n . Then, there exists u ∈ X such that for every
T > 0, un → u uniformly in X for t ∈ [0, T ] as n → ∞. Since
T 1−α
k0 It1−α un kX ≤ kun kL∞ ((0,T ),X) , t ∈ [0, T ],
(1 − α)Γ(1 − α)
so we know 0 It1−α un → 0 It1−α u in X. By (2.10),
C
kC α
0 Dt un kX ≤ ku0,n kX , for some constant C > 0, t > 0.
tα
C α
Hence, for every δ > 0, there exists w ∈ C([δ, ∞), X) such that 0 Dt un → w
uniformly in X on t ∈ [δ, ∞).
Note that for t ∈ [δ, ∞),
d
C α
0 Dt un = ( I 1−α (Pα (t)u0,n − u0,n )) = Aun ,
dt 0 t
 Blow-up and Global Existence for Fractional equation 77

so
d 1−α
Iw= (u − u0 ) = C α
0 Dt u, t ∈ [δ, ∞).
dt 0 t
Since A is closed, we have w = Au, that is C α
0 Dt u = Au = APα (t)u0 , t ∈ [δ, ∞).
C α
By arbitrariness of δ, we have 0 Dt u = APα (t)u0 , t > 0. 

Lemma 2.4. Assume that f ∈ Lq ((0, T ), C0 (RN )), q > 1. Let

Z t
w(t) = (t − s)α−1 Sα (t − s)f (s) ds,
0

then
Z t
1−α
0 It w = Pα (t − s)f (s) ds.
0

Furthermore, if qα > 1, then w ∈ C([0, T ], C0 (RN )).

Proof. Let X = C0 (RN ). By Fubini theorem and (2.11), we have

Z t Z s
1−α 1 −α
I
0 t w = (t − s) (s − τ )α−1 Sα (s − τ )f (τ ) dτ ds
Γ(1 − α) 0 0
Z tZ t
1
= (t − s)−α (s − τ )α−1 Sα (s − τ )f (τ ) ds dτ
Γ(1 − α) 0 τ
Z t Z t−τ
1
= (t − s − τ )−α sα−1 Sα (s)f (τ ) ds dτ
Γ(1 − α) 0 0
Z t
= Pα (t − τ )f (τ ) dτ.
0

For every h > 0 and t + h ≤ T , we have w(t + h) − w(t) = I1 + I2 , where

Z t+h Z ∞
I1 = α θφα (θ)(t + h − τ )α−1 T ((t + h − τ )α θ)f (τ ) dθ dτ.
t 0
Z tZ ∞
I2 = α θφα (θ)[(t + h − τ )α−1 T ((t + h − τ )α θ)
0 0
− (t − τ )α−1 T ((t − τ )α θ)]f (τ ) dθ dτ,

By Hölder inequality, we have

Z t+h Z ∞
(2.12) kI1 kX ≤ α θφα (θ)(t + h − τ )α−1 kf (τ )kX dθ dτ
t 0
Z t+h
1
= (t + h − τ )α−1 kf (τ )kX dτ
Γ(α) t
 Z t+h (q−1)/q
1 q(α−1)/(q−1)
≤ kf kLq ((0,T ),X) (t + h − τ ) dτ
Γ(α) t
 (q−1)/q
1 q−1
= kf kLq ((0,T ),X) h(qα−1)/q .
Γ(α) qα − 1
78 Q.G. Zhang — H.R. Sun

Note that, for 0 < τ < t,

k(t + h − τ )α−1 T ((t + h − τ )α θ)f (τ ) − (t − τ )α−1 T ((t − τ )α θ)f (τ )kX

≤ 2(t − τ )α−1 kf (τ )kX
and there exists constant C > 0 such that
k[(t + h − τ )α−1 T ((t + h − τ )α θ) − (t − τ )α−1 T ((t − τ )α θ)]f (τ )kX
≤ |(t + h − τ )α−1 − (t − τ )α−1 |kT ((t + h − τ )α θ)f (τ )kX
+ (t − τ )α−1 k(T ((t + h − τ )α θ) − T ((t − τ )α θ))f (τ )kX
≤ C(t − τ )α−2 hkf (τ )kX .
Therefore
Z tZ ∞  
1 h
kI2 kX ≤ C αθφα (θ) min , dθ kf (τ )kX dτ
0 0 (t − τ )1−α (t − τ )2−α
Z t  q/(q−1) (q−1)/q
C 1 h
≤ min , dτ
Γ(α) 0 (t − τ )1−α (t − τ )2−α
· kf kLq ((0,T ),X) .
Observe that
Z t  q/(q−1)
1 h
min , dτ
0 (t − τ )1−α (t − τ )2−α
Z t  q/(q−1)
1 h
= min , dτ
0 τ 1−α τ 2−α
Z ∞  q/(q−1)
1 h
≤ min , dτ
0 τ 1−α τ 2−α
Z h Z ∞
= τ q(α−1)/(q−1) dτ + hq/(q−1) τ q(α−2)/(q−1) dτ
0 h
q(q − 1)
= hqα−1/(q−1) ,
(qα − 1)(q + 1 − qα)
so,
(2.13) kI2 kX ≤ Ckf kLq ((0,T ),X) h(qα−1)/q .
Hence, (2.12)–(2.13) imply that the conclusion hold. 

Remark 2.5. For α = 1, the conclusion of Lemma 2.4 also holds (see Theo-
rem 3.1 of Chapter 4 in [22]).

3. Local existence
In this section, we give the local existence and uniqueness of mild solution
of the problem (1.1)–(1.2). First, we give the definition of mild solution of (1.1)–
(1.2).
 Blow-up and Global Existence for Fractional equation 79

Definition 3.1. Let u0 ∈ C0 (RN ), T > 0. We call that u ∈ C([0, T ], C0 (RN ))

is a mild solution of the problem (1.1)–(1.2) if u satisfies the following integral
equation
Z t
u(t) = Pα (t)u0 + (t − s)α−1 Sα (t − s)|u|p−1 u(s) ds, t ∈ [0, T ].
0

For the problem (1.1)–(1.2), we have the following local existence result.

Theorem 3.2. Given u0 ∈ C0 (RN ), then there exists a maximal time Tmax =
T (u0 ) > 0 such that the problem (1.1)–(1.2) has a unique mild solution u in
C([0, T ), C0 (RN )) and either Tmax = +∞ or Tmax < +∞ and kukL∞ ((0,t),C0 (RN ))
→ +∞ as t → Tmax . If, in addition, u0 ≥ 0, u0 6≡ 0, then u(t) > 0 and
u(t) ≥ Pα (t)u0 for t ∈ (0, Tmax ). Moreover, if u0 ∈ Lr (RN ) for some r ∈ [1, ∞),
then u ∈ C([0, Tmax ), Lr (RN )).

Proof. For given T > 0 and u0 ∈ C0 (RN ), let

ET = {u | u ∈ C([0, T ], C0 (RN )), kukL∞ ((0,T ),L∞ (RN )) ≤ 2ku0 kL∞ (RN ) },

d(u, v) = max ku(t) − v(t)kL∞ (RN ) for u, v ∈ ET .

t∈[0,T ]

Since C([0, T ], C0 (RN )) is a Banach space, (ET , d) is a complete metric space.

We define the operator G on ET as
Z t
G(u)(t) = Pα (t)u0 + (t − s)α−1 Sα (t − s)|u(s)|p−1 u(s) ds, u ∈ ET ,
0

then G(u) ∈ C([0, T ], C0 (RN )) in view of Lemma 2.4. If u ∈ ET , then by

Lemma 2.1(b) and Lemma 2.2(b), for t ∈ [0, T ],
Z t
1
kG(u)(t)kL∞ (RN ) ≤ ku0 kL∞ (RN ) + (t − s)α−1 ku(s)kpL∞ (RN ) ds
Γ(α) 0
2p T α
≤ ku0 kL∞ (RN ) + ku0 kpL∞ (RN ) .
αΓ(α)
Hence, we can choose T small enough such that
2p T α
ku0 kp−1
L∞ (RN )
≤ 1,
αΓ(α)
so we get kG(u)kL∞ ((0,T ),L∞ (RN )) ≤ 2ku0 kL∞ (RN ) . Furthermore, for u, v ∈ ET ,
we have for t ∈ [0, T ]

kG(u)(t) − G(v)(t)kL∞ (RN )

Z t
1
≤ (t − s)α−1 k|u(s)|p−1 u(s) − |v(s)|p−1 v(s)kL∞ (RN ) ds
Γ(α) 0
p−1
4(p−1) p T α ku0 kL ∞ (RN )
≤ ku − vkL∞ ((0,T ),L∞ (RN )) .
αΓ(α)
80 Q.G. Zhang — H.R. Sun

We can choose T small enough such that

p 4(p−1) T α ku0 kp−1
L∞ (RN ) 1
≤ ,
αΓ(α) 2
then kG(u)(t) − G(v)(t)kL∞ (RN ) ≤ ku − vkL∞ ((0,T ),C0 (RN )) /2. Therefore, G is
contractive on ET . So, G has a fixed point u ∈ ET by the contraction mapping
principle.
Now, we prove the uniqueness. Let u, v ∈ C([0, T ], C0 (RN )) be the mild
solutions of (1.1)–(1.2) for some T > 0, then there exists positive constant C > 0
such that

ku(t) − v(t)kL∞ (RN ) = kG(u(t) − G(v)(t)kL∞ (RN )

Z t
≤C (t − s)α−1 ku(s) − v(s)kL∞ (RN ) ds.
0
Hence, by Gronwall’s inequality, we know u = v.
Next, using the uniqueness of solution, we conclude that the existence of
solution on a maximal interval [0, Tmax ), where

Tmax = sup{T > 0 | there exists a mild solution u

of (1.1)–(1.2) in C([0, T ], C0 (RN ))}.
Assume that Tmax < +∞ and there exists M > 0 such that for t ∈ [0, Tmax ),
ku(t)kL∞ (RN ) ≤ M.
Next, we will verify that lim u(t) exists in C0 (RN ). In fact, for 0 < t <
−
t→Tmax
τ < Tmax , by the proof of Lemma 2.4, there exists constant C > 0 such that
ku(t) − u(τ )kL∞ (RN ) ≤ kPα (t)u0 − Pα (τ )u0 kL∞ (RN )
Z t
+ (t − s)α−1 Sα (t − s)|u(s)|p−1 u(s)
0

− (τ − s)α−1 Sα (τ − s)|u(s)|p−1 u(s) ds

L∞ (RN )
Z τ
+ (τ − s)α−1 Sα (τ − s)|u(s)|p−1 u(s) ds
t L∞ (RN )
p Z τ
M
≤ kPα (t)u0 − Pα (τ )u0 kL∞ (RN ) + (τ − s)α−1 ds
Γ(α) t
Z t
p
+ CM min{(t − s)α−1 , (t − s)α−2 (τ − t)} ds
0
Mp 1
≤ kPα (t)u0 − Pα (τ )u0 kL∞ (RN ) + (τ − t)α + CM p (τ − t)α .
αΓ(α) α(1 − α)
Since Pα (t)u0 is uniformly continuous in [0, Tmax ], so lim u(t) exists.
−
t→Tmax
 Blow-up and Global Existence for Fractional equation 81

We denote uTmax = lim u(t) and define u(Tmax ) = uTmax . Hence, u ∈

−
t→Tmax
C([0, Tmax ], C0 (RN )) and then, by Lemma 2.4,

Z t
(t − s)α−1 Sα (t − s)|u(s)|p−1 u(s) ds ∈ C([0, Tmax ], C0 (RN )).
0

For h > 0, δ > 0, let

Eh,δ = {u ∈ C([Tmax , Tmax + h], C0 (RN )) | u(Tmax ) = uTmax , d(u, uTmax ) ≤ δ},

where d(u, v) = max ku(t) − v(t)kL∞ (RN ) for u, v ∈ Eh,δ .

t∈[Tmax ,Tmax +h]
Via C([Tmax , Tmax + h], C0 (RN )) is a Banach space, we know (Eh,δ , d) is
a complete metric space.
We define the operator G on Eh,δ as

Z Tmax
G(v)(t) = Pα (t)u0 + (t − τ )α−1 Sα (t − τ )|u|p−1 u(τ ) dτ
0
Z t
+ (t − τ )α−1 Sα (t − τ )|v|p−1 v(τ ) dτ, v ∈ Eh,δ .
Tmax

Clearly, G(v) ∈ C([Tmax , Tmax + h], C0 (RN )) and G(v)(Tmax ) = uTmax .

If v ∈ Eh,δ , then for t ∈ [Tmax , Tmax + h],

kG(v)(t) − uTmax kL∞ (RN )

≤ kPα (t)u0 − Pα (Tmax )u0 kL∞ (RN ) + kI3 kL∞ (RN ) + kI4 kL∞ (RN ) ,

where
Z Tmax
I3 = (t − τ )α−1 Sα (t − τ )|u(τ )|p−1 u(τ )
0
− (Tmax − τ )α−1 Sα (Tmax − τ )|u(τ )|p−1 u(τ ) dτ,
Z t
I4 = (t − τ )α−1 Sα (t − τ )|v|p−1 v(τ ) dτ.
Tmax

Taking h small enough such that

δ
kPα (t)u0 − Pα (Tmax )u0 kL∞ (RN ) ≤ for t ∈ [Tmax , Tmax + h],
3
δ
kI3 kL∞ (RN ) ≤ ,
3
82 Q.G. Zhang — H.R. Sun
Z t
kI4 kL∞ (RN ) ≤ (t − τ )α−1 Sα (Tmax − τ )(|v|p−1 v(τ )
Tmax

− |uTmax |p−1 uTmax ) dτ

L∞ (RN )
Z t
+ (t − τ )α−1 Sα (Tmax − τ )|uTmax |p−1 uTmax dτ
Tmax L∞ (RN )
Z t Z t
1
≤ Cδ (t − τ )α−1 dτ + kuTmax kpL∞ (RN ) (t − τ )α−1 dτ
Tmax Γ(α) Tmax

Cδ kuTmax kpL∞ (RN ) δ

= (t − Tmax )α + (t − Tmax )α ≤
α Γ(α + 1) 3
for t ∈ [Tmax , Tmax + h]. Then, we have kG(v)(t) − uTmax kL∞ (RN ) ≤ δ, t ∈
[Tmax , Tmax + h].
Next, we will prove that G is contractive on Eh,δ for h small enough. In fact,
for w, v ∈ Eh,δ , t ∈ [Tmax , Tmax + h],

kw(t) − v(t)kL∞ (RN )

Z t
≤ (t − τ )α−1 kSα (t − τ )(|w|p−1 w(τ ) − |v|p−1 v(τ ))kL∞ (RN ) dτ
Tmax
≤ kw − vkL∞ ((Tmax ,Tmax +h),L∞ (RN )) (kwkL∞ ((Tmax ,Tmax +h),L∞ (RN ))
Z t
p−1 p
+ kvkL∞ ((Tmax ,Tmax +h),L∞ (RN )) ) (t − τ )α−1 dτ
Γ(α) Tmax
2p−1 p
≤ (δ + kuTmax kL∞ (RN ) )p−1 (t − Tmax )α d(w, v).
Γ(α + 1)
Choosing h small enough such that
2p−1 p 1
(δ + kuT kL∞ (RN ) )p−1 hα ≤ .
Γ(α + 1) 2
Then, G is contractive on Eh,δ . So, we know G has a fixed point v ∈ Eh,δ . Since
v(Tmax ) = G(v(Tmax )) = u(Tmax ), if we let

u(t), t ∈ [0, T
max ),
u
e(t) =
v(t), t ∈ [Tmax , Tmax + h],

e ∈ C([0, Tmax + h], C0 (RN )) and

then u
Z t
u
e(t) = Pα (t)u0 + (t − τ )α−1 Sα (t − τ )|ũ|p−1 u
e(τ ) dτ.
0

Therefore, u
e(t) is a mild solution of (1.1)–(1.2), which contradicts with the defi-
nition of Tmax .
If u0 ∈ Lr (RN ) for some 1 ≤ r < ∞, then repeating the above argument,
we get the conclusion. Moreover, if u0 ≥ 0, then we can obtain the nonnegative
 Blow-up and Global Existence for Fractional equation 83

solution of (1.1) applying the above argument in the set ET+ = {u ∈ ET | u ≥ 0}.
Then, we know u(t) ≥ Pα (t)u0 > 0 on t ∈ (0, Tmax ). 

4. Blow-up and global existence

In this section, we prove the blow-up results and global existence of solutions
of (1.1)–(1.2). First, we give the definition of weak solution of (1.1)–(1.2).

Definition 4.1. We call u ∈ Lp ((0, T ), L∞ N ∞ N

loc (R )), for u0 ∈ Lloc (R ) and
T > 0, is a weak solution of (1.1) if
Z Z T
(|u|p−1 uϕ + u0 C α
t DT ϕ) dt dx
RN 0
Z Z T Z Z T
C α
= u(−4ϕ) dt dx + u t DT ϕ dt dx
RN 0 RN 0
2,1
for every ϕ ∈ Cx,t (RN × [0, T ]) with supp ϕ ⊂⊂ RN and ϕ( · , T ) = 0.
x

Lemma 4.2. Assume u0 ∈ C0 (RN ), let u ∈ C([0, T ], C0 (RN )) be a mild

solution of (1.1)–(1.2), then u is also a weak solution of (1.1)–(1.2).

Proof. Assuming that u ∈ C([0, T ], C0 (RN )) is a mild solution of (1.1)–

(1.2), we have
Z t
u − u0 = Pα (t)u0 − u0 + (t − τ )α−1 Sα (t − τ )|u|p−1 u dτ.
0

Note that by Lemma 2.4,

Z t  Z t
1−α α−1 p−1
I
0 t (t − τ ) Sα (t − τ )|u| u(τ ) dτ = Pα (t − s)|u|p−1 u(s) ds,
0 0

so, we know
Z t
1−α 1−α
0 It (u − u0 ) = 0 It (Pα (t)u0 − u0 ) + Pα (t − τ )|u|p−1 u(τ ) dτ.
0
2,1
Then, for every ϕ ∈ Cx,t (RN × [0, T ]) with supp ϕ ⊂⊂ RN and ϕ(x, T ) = 0, we
x
get
Z
1−α
(4.1) 0 It (u − u0 )ϕ dx = I5 (t) + I6 (t),
RN

where
Z Z Z t
1−α
I5 (t) = 0 It (Pα (t)u0 − u0 )ϕ dx, I6 (t) = Pα (t − s)|u|p−1 u ds ϕ dx.
RN RN 0

By Lemma 2.3,
Z Z
dI5 1−α
(4.2) = A(Pα (t)u0 )ϕ dx + 0 It (Pα (t)u0 − u0 )ϕt dx.
dt RN RN
84 Q.G. Zhang — H.R. Sun

For every h > 0, t ∈ [0, T ) and t + h ≤ T , we have

1 t+h
Z Z
1
(I6 (t + h) − I6 (t)) = Pα (t + h − s)|u|p−1 u dsϕ(t + h, x) dx
h h 0 R N

1 t
Z Z
− Pα (t − s)|u|p−1 udsϕ(t, x) dx = I7 + I8 + I9 ,
h 0 RN
where
Z Z t+h Z ∞
1
I7 = φα (θ)T ((t + h − s)α θ)|u|p−1 u(s) dθ ds ϕ(t + h, x) dx,
h RN t 0
Z Z tZ ∞
1
I8 = φα (θ)(T ((t+h−s)α θ)−T ((t−s)α θ))|u|p−1 u(s) dθ ds ϕ(t, x) dx,
h RN 0 0
Z Z tZ ∞
1
I9 = φα (θ)T ((t+h−s)α θ)|u|p−1 u(s) dθ ds(ϕ(t+h, x)−ϕ(t, x)) dx.
h RN 0 0
By dominated convergence theorem, we conclude that
Z
I7 → |u|p−1 uϕ dx as h → 0,
RN
Z Z tZ ∞
I9 → φα (θ)T ((t − s)α θ)|u|p−1 u(s) dθ ds ϕt dx
RN 0 0
Z Z t
= Pα (t − s)|u|p−1 u(s)dsϕt dx as h → 0.
RN 0
Since
Z Z tZ ∞ Z 1
I8 = αθφα (θ)(t + τ h − s)α−1
RN 0 0 0
· A(T ((t + τ h − s)α θ))|u|p−1 u(s) dτ dθ dsϕ dx
Z Z tZ ∞ Z 1
= A αθφα (θ)(t + τ h − s)α−1
RN 0 0 0
· T ((t + τ h − s)α θ)|u|p−1 u(s) dτ dθ ds ϕ dx
Z Z tZ ∞ Z 1
= αθφα (θ)(t + τ h − s)α−1
RN 0 0 0
· T ((t + τ h − s)α θ)|u|p−1 u(s) dτ dθ ds Aϕ dx,
by dominated convergence theorem, we know
Z Z t
I8 → (t − s)α−1 Sα (t − s)|u|p−1 u(s) ds Aϕ dx as h → 0.
RN 0

Hence, the right derivative of I6 on [0, T ) is

Z Z Z t
|u|p−1 uϕ dx + (t − s)α−1 Sα (t − s)|u|p−1 u(s) ds Aϕ dx
RN RN 0
Z Z t
+ Pα (t − s)|u|p−1 u(s) ds ϕt dx
RN 0
 Blow-up and Global Existence for Fractional equation 85

and it is continuous in [0, T ). Therefore,

Z Z Z t
dI6 p−1
(4.3) = |u| uϕ dx + (t − s)α−1 Sα (t − s)|u|p−1 u(s) ds Aϕ dx
dt RN RN 0
Z Z t
+ Pα (t − s)|u|p−1 u(s) ds ϕt dx
RN 0
Z Z Z t
= |u|p−1 uϕ dx + (t − s)α−1 Sα (t − s)|u|p−1 u(s) ds Aϕ dx
RN RN 0
Z Z t 
1−α α−1 p−1
+ 0 It (t − s) Sα (t − s)|u| u(s) ds ϕt dx,
RN 0

for t ∈ [0, T ). It follows from (4.1)–(4.3) that

Z T Z Z T
d 1−α dI5 dI6
0= I
0 t (u − u0 )ϕ dx dt = + dt
0 dt RN 0 dt dt
Z TZ
= Pα (t)u0 4ϕ dx dt
0 RN
Z T Z Z T Z
− (u − u0 )C α
t DT ϕ dx dt + |u|p−1 uϕ dx dt
0 RN 0 RN
Z T Z Z t
+ (t − s)α−1 Sα (t − s)|u|p−1 u ds 4ϕ dx dt
0 RN 0
Z T Z Z T Z Z T Z
= u4ϕ dx dt − (u − u0 )C α
t DT ϕ dx dt + |u|p−1 uϕ dx dt.
0 RN 0 RN 0 RN

Hence, we get the conclusion. 

We say the solution u of the problem (1.1)–(1.2) blows up in a finite time T

if lim ku(t, · )kL∞ (RN ) = +∞.
t→T
Now, we give a blow-up result of the problem (1.1)–(1.2).

Theorem 4.3. Let u0 ∈ C0 (RN ) and u0 ≥ 0, if

Z
u0 (x)χ(x)dx > 1,
RN

where
Z √ −1 √
N 2 +|x|2 N 2 +|x|2
χ(x) = e− dx e− ,
RN

then the mild solutions of (1.1)–(1.2) blow up in a finite time.

Proof. We take ψ ∈ C0∞ (R) such that


1, |x| ≤ 1,
ψ(x) =
0, |x| ≥ 2,
86 Q.G. Zhang — H.R. Sun

and 0 ≤ ψ(x) ≤ 1, x ∈ R. Let ψn (x) = ψ(x/n), n = 1, 2, . . . By Lemma 4.2,

a mild solution of (1.1)–(1.2) is also a weak solution of it. So, using the def-
inition of weak solution of (1.1)–(1.2), taking ϕn (x, t) = χ(x)ψn (x)ϕ1 (t) for
ϕ1 ∈ C 1 ([0, T ]) with ϕ1 (T ) = 0 and ϕ1 ≥ 0, we have
Z Z T Z Z T
p
(4.4) u ϕn dx dt + u0 C α
t DT ϕn dx dt
RN 0 RN 0
Z Z T
= (−u4ϕn + uC α
t DT ϕn ) dx dt.
RN 0

Since 4(χψn ) = (4χ)ψn + 2∇χ · ∇ψn + (4ψn )χ and 4χ ≥ −χ, by (4.4) and
the dominated convergence theorem, let n → ∞, we have
Z Z T Z Z T
p
(4.5) u χϕ1 dx dt + u0 χC α
t DT ϕ1 dx dt
RN 0 RN 0
Z Z T
≤ (uχϕ1 + uχC α
t DT ϕ1 ) dx dt.
RN 0

Hence, by Jensen’s inequality and (4.5), we have

Z T Z p Z Z T
uχ dx ϕ1 dt + u0 χC α
t DT ϕ1 dx dt
0 RN RN 0
Z Z T
≤ (uχϕ1 + uχC α
t DT ϕ1 ) dx dt.
RN 0
R
So, if we denote f (t) = RN uχ dx, then
Z T Z T
p
(4.6) (f − f )ϕ1 dt ≤ (f − f (0))C α
t DT ϕ1 dt.
0 0
αe
We take ϕ1 = where ψe ∈
t IT ψ(t) C01 ((0, T ))
and ψe ≥ 0, then (4.6) implies
Z T Z T Z T
α p p αe
0 It (f − f )ψ dt = (f − f )t IT ψ(t) dt ≤ (f − f (0))ψe dt.
e
0 0 0

Hence,
α p
(4.7) 0 It (f − f ) + f (0) ≤ f.
R
In view of f (0) = RN u0 (x)χ(x) dx > 1 and the continuity of f , we obtain
f (t) > 1 when t is small enough. Then (4.7) implies f (t) ≥ f (0) > 1 for
t ∈ [0, T ]. Taking ϕ1 (t) = (1 − t/T )m , t ∈ [0, T ] m ≥ max{1, pα/(p − 1)}, we
know there exists constant C > 0 such that
Z T Z T
(f p − f )ϕ1 dt + Cf (0)T 1−α ≤ ε f p ϕ1 dt + C(ε)T 1−pα/(p−1) .
0 0

Choosing ε small enough such that f (0) > (1 − ε)−1/(p−1) , we then have f (0) ≤
CT α−pα/(p−1) for some constant C > 0. If the solution of (1.1)–(1.2) exists
 Blow-up and Global Existence for Fractional equation 87

globally, we get f (0) = 0 by taking T → ∞, which contradicts with f (0) > 1.

Hence, Theorem 3.2 guarantees that u blows up in a finite time. 

Next, we give the main result of this paper.

Theorem 4.4. Let u0 ∈ C0 (RN ) and u0 ≥ 0, u0 6≡ 0, then

(a) If 1 < p < 1 + 2/N , then the mild solution of (1.1)–(1.2) blows up in
a finite time.
(b) If p ≥ 1+2/N and ku0 kLqc (RN ) is sufficiently small, where qc = N (p−1)/2,
then the solutions of (1.1)–(1.2) exist globally.

Proof. (a) Let Φ ∈ C0∞ (R) such that Φ(s) = 1 for |s| ≤ 1, Φ(s) = 0 for
|s| > 2 and 0 ≤ Φ(s) ≤ 1. For T > 0, we define
 m  
−α/2 2p/(p−1) t pα
ϕ1 (x) = (Φ(T |x|)) , ϕ2 (t) = 1 − , m ≥ max 1, ,
T p−1
for t ∈ [0, T ]. Assuming that u is a mild solution of (1.1), then, by Lemma 4.2,
we have
Z Z T
(4.8) (up ϕ1 ϕ2 + u0 ϕ1 C α
t DT ϕ2 ) dt dx
RN 0
Z Z T
= (u(−4ϕ1 )ϕ2 + uϕ1 C α
t DT ϕ2 ) dt dx
RN 0
Note that
−α 1/p 1/p
(4.9) |(−4ϕ1 )ϕ2 + ϕ1 C α
t DT ϕ2 | ≤ CT ϕ1 ϕ2
for some positive constant C independent of T . Then, by (4.8), (4.9) and Hölder
inequality, we have
Z Z T Z Z T
−α 1/p 1/p
(up ϕ1 ϕ2 + u0 ϕ1 C
t D α
T 2ϕ ) dt dx ≤ CT uϕ1 ϕ2 dt dx
RN 0 RN 0
Z Z T 1/p
−α+(1+αN /2)(p−1)/p p
≤ CT u ϕ1 ϕ2 dt dx .
RN 0

Hence Z
1−α
T u0 ϕ1 dx ≤ CT 1+αN /2−pα/(p−1) .
RN
It follows from p < 1 + 2/N that (N/2 + 1)α − pα/(p − 1) < 0. Therefore, if
solution of (1.1)–(1.2) exists globally, then taking T → ∞, we obtain
Z
u0 ϕ1 dx = 0
RN
and then u0 ≡ 0. Hence, by Theorem 4.3, we know u blows up in a finite time.
(b) We construct the global solution of (1.1)–(1.2) by the contraction map-
ping principle.
88 Q.G. Zhang — H.R. Sun

Since p ≥ 1 + 2/N > 1 + 2α/(αN + 2 − 2α), we know

αN (p − 1)
(4.10) > 1,
2(pα − p + 1)+
where (pα − p + 1)+ = max{0, pα − p + 1}.
√
In view of p ≥ 1 + 2/N > (4 − N + N 2 + 16)/4, we have
N (p − 1)
(4.11) > 1.
2p(2 − p)+
Hence, by (4.10), (4.11) and (p − 1)N /(2p) < (αN (p − 1))/(2(pα − p + 1)+ ), we
can choose q > p ≥ 1 + 2/N such that
α 1 αN α
(4.12) − < <
p−1 p 2q p−1
and
α αN
(4.13) −α< .
p−1 2q
Let
 
αN 1 1 α αN
(4.14) β= − = − .
2 qc q p−1 2q
Using (4.12) and (4.14), one verifies that
αN (p − 1)
(4.15) 0 < pβ < 1, α= + (p − 1)β.
2q
Assume that the initial value u0 satisfies
(4.16) sup tβ kPα (t)u0 kLq (RN ) = η < +∞.
t>0

Note that (4.13) implies 1/qc − 1/q < 2/N . If u0 ∈ Lqc (RN ), (2.8) im-
plies (4.16) holds. If u0 (x) ≤ C|x|−2/(p−1) for some constant C > 0, then
kT (t)u0 kLq (RN ) ≤ CtN/(2q)−1/(p−1) . Hence,
Z ∞
kPα (t)u0 kLq (RN ) ≤ Ctα(N/(2q)−1/(p−1)) φα (θ)θN/(2q)−1/(p−1) dθ.
0

Since N/(2q) − 1/(p − 1) > −1,

Z ∞
φα (θ)θN/(2q)−1/(p−1) dθ < ∞.
0

Therefore, we also obtain that (4.16) is satisfied in this case.

Let Y = {u ∈ L∞ ((0, ∞), Lq (RN )) | kukY < ∞}, where
kukY = sup tβ ku(t)kLq (RN ) .
t>0

For u ∈ Y , we define
Z t
Φ(u)(t) = Pα (t)u0 + (t − s)α−1 Sα (t − s)|u|p−1 u(s)ds.
0
 Blow-up and Global Existence for Fractional equation 89

Denote BM = {u ∈ Y | kukY ≤ M }. For any u, v ∈ BM , t ≥ 0,

(4.17) tβ kΦ(u)(t) − Φ(v)(t)kLq (RN )

Z t
≤ tβ (t − s)α−1 kSα (t − s)(up (s) − v p (s))kLq (RN ) ds.
0

Since q > p > N (p − 1)/4, so p/q − 1/q < 4/N . Hence, Hölder inequality,
Lemma 2.2, (4.15) and (4.17) imply that there exists constant C > 0 such that
tβ kΦ(u) − Φ(v)kLq (RN )
Z t
≤ Ctβ (t − s)α−1−αN (p/q−1/q)/2 kup − v p k q ds
0 L p (RN )
Z t
β p−1 p−1
≤ Ct (t − s)α−1−αN (p−1)/(2q) (kukLq (RN ) + kvkLq (RN ) )ku − vkLq (RN ) ds
0
Z t
β
≤ Ct M p−1
(t − s)α−1−αN (p−1)/(2q) s−pβ dsku − vkY
0
p−1 β−pβ−αN (p−1)/(2q)+α
= CM t
Z 1
· (1 − τ )−αN (p−1)/(2q)+α−1 τ −pβ dτ ku − vkY
0
Z 1
= CM p−1 (1 − τ )−αN (p−1)/(2q)+α−1 τ −pβ dτ ku − vkY
0
Γ((p − 1)β)Γ(1 − pβ)
= CM p−1 ku − vkY .
Γ(1 − β)
If we choose M small enough such that
Γ((p − 1)β)Γ(1 − pβ) 1
CM p−1 < ,
Γ(1 − β) 2
then kΦ(u) − Φ(v)kY ≤ ku − vkY /2. Since
Z t
β p β
t kΦ(u)(t)kLq (RN ) ≤ η + CM t (t − s)−αN (p/q−1/q)/2−1+α s−pβ ds
0
Γ((p − 1)β)Γ(1 − pβ)
≤ η + CM p , t ∈ [0, +∞),
Γ(1 − β)
we can choose η and M small enough such that
Γ((p − 1)β)Γ(1 − pβ)
η + CM p ≤ M.
Γ(1 − β)
Therefore, by contraction mapping principle we know Φ has a fixed point u ∈
BM . Next, we will prove u ∈ C([0, ∞), C0 (RN )).
First, we prove that for T > 0 small enough, u ∈ C([0, T ], C0 (RN )). In fact,
the above proof shows that u is the unique solution in
n o
BM,T = u ∈ L∞ ((0, T ), Lq (RN )) sup tβ ku(t)kLq (RN ) ≤ M .
0<t<T
90 Q.G. Zhang — H.R. Sun

By Theorem 3.2 and u0 ∈ C0 (RN ) ∩ Lq (RN ), we know that for T small enough,
e ∈ C([0, T ], C0 (RN ) ∩ Lq (RN )).
(1.1) has a unique solution u
Hence, we can take T small enough such that sup tβ ke u(t)kLq (RN ) ≤ M .
0<t<T
Then, by uniqueness, we know u ≡ u e for t ∈ [0, T ] and then u ∈ C([0, T ], C0 (RN ))
q N
∩ C([0, T ], L (R )).
Next, we show that u ∈ C([0, T ], C0 (RN )) by a bootstrap argument. For
t > T , we have
Z t
u − Pα (t)u0 = (t − s)α−1 Sα (t − s)up ds
0
Z T Z t
= (t − s)α−1 Sα (t − s)up ds + (t − s)α−1 Sα (t − s)up ds
0 T
= I10 + I11 .
It follows from u ∈ C([0, T ], C0 (RN )) that
I10 ∈ C([T, ∞), C0 (RN )) ∩ C([T, ∞), Lq (RN )).
For T1 > T , we know up ∈ L∞ ((T, T1 ), Lq/p (RN )). Note that q > N (p − 1)/2,
we can choose r > q such that N (p/q − 1/r)/2 < 1. Then analogous to the proof
of Lemma 2.4, we can show that I11 ∈ C([T, T1 ], Lr (RN )). By the arbitrariness
of T1 , we know I11 ∈ C([T, ∞), Lr (RN )) and so u ∈ C([T, ∞), Lr (RN )).
We take r = qχi , χ > 1 such that
 
N p 1
− i < 1, i = 1, 2, . . . ,
2 q qχ
i
then u ∈ C([T, ∞), Lqχ (RN )). By finite steps, we have p/(qχi ) < 2/N , so
u ∈ C([0, ∞), C0 (RN )). 

Remark 4.5. Theorems 4.3, 4.4 and (1.7) guarantee that Lqc (RN ) is the
only possible Lebesgue space where smallness of the initial value could imply
global existence.

Remark 4.6. The Fujita exponent can also be obtained by equaling the time
decay rate N α/2 of kpα (t)u0 kL∞ (RN ) (see (2.8)) and the blow-up rate α/(p − 1)
of fractional ordinary differential equation C α p
0 Dt u = u , p > 1, u = u0 (see [6]
for α = 1).

References
[1] D.G. Aronson and H.F. Weinberger, Multidimensional nonlinear diffusion arising in
population genetics, Adv. Math. 30 (1978), 33–76.
[2] E.G. Bazhlekova, Subordination principle for fractional evolution equations, Fract. Calc.
Appl. Anal. 3 (2000), 213–230.
[3] C.M. Carracedo and M.S. Alix, The Theory of Fractional Powers of Operators, North-
Holland Mathematics Studies 187, Elsevier, 2001.
 Blow-up and Global Existence for Fractional equation 91

[4] T. Cazenave, F. Dickstein and F.B. Weissler, An equation whose Fujita critical ex-
ponent is not given by scaling, Nonlinear Anal. 68 (2008), 862–874.
[5] T. Cazenave and A. Haraux, An Introduction to Semilinear Evolution Equations, Ox-
ford Lecture Series in Mathematics and its Applications, 1998.
[6] K. Deng and H.A. Levine, The role of critical exponents in blow-up theorems: the sequel,
J. Math. Anal. Appl. 243 (2000), 85–126.
[7] S.D. Eidelman and A.N. Kochubei, Cauchy problem for fractional diffusion equations,
J. Differential Equations 99 (2004), 211–255.
[8] A.Z. Fino and M. Kirane, Qualitative properties of solutions to a time-space fractional
evolution equation, Quart. Appl. Math. 70 (2012), 133–157.
[9] H. Fujita, On the blowing up of solutions of the Cauchy problem for ut = 4u + uα+1 ,
J. Fac. Sci. Univ. Tokyo Sect. I 13 (1966), 109–124.
[10] K. Hayakawa, On nonexistence of global solutions of some semilinear parabolic equations,
Proc. Japan Acad. 49 (1973), 503–505.
[11] A.A. Kilbas, H.M. Srivastava and J.J. Trujillo, Theory and Applications of Fractional
Differential Equations, vol 204. Elsevier Science B.V., Amsterdam, 2006.
[12] M. Kirane, Y. Laskri and N.E. Tatar, Critical exponents of Fujita type for certain evo-
lution equations and systems with spatio-temporal fractional derivatives, J. Math. Anal.
Appl. 312 (2005), 488–501.
[13] K. Kobayashi, T. Siaro and H. Tanaka, On the growing up problem for semilinear heat
equations, J. Math. Soc. Japan 29 (1977) 407–424.
[14] A.N. Kochubei, Fractional parabolic systems, Potential Anal. 37 (2012), 1–30.
[15] M. Li, C. Chen and F.B. Li, On fractional powers of generators of fractional resolvent
families, J. Funct. Anal. 259 (2010), 2702–2726.
[16] F. Mainardi, On the initial value problem for the fractional diffusion-wave equation,
Waves and Stability in continuous Media (S. Rionero and T. Ruggeri, eds.), World Scien-
tific, Singapore, 1994, 246–251.
[17] M.M. Meerschaert, E. Nane and P. Vellaisamy, Fractional Cauchy problems on
bounded domains, Ann. Probab. 37 (2009), 979–1007.
[18] R. Metzler and J. Klafter, The restaurant at the end of the random walk: recent de-
velopments in the description of anomalous transport by fractional dynamics, J. Phys. A:
Math. Gen. 37 (2004), 161–208.
[19] E. Mitidieri and S.I. Pohozaev, A priori estimates and blow-up of solutions to nonlinear
partial differential equations and inequalities, Proc. Steklov Inst. Math. 234 (2001), 1–383.
[20] G.M. Mophou and G.M. N’Guérékata, On a class of fractional differential equations
in a Sobolev space, Appl. Anal. 91 (2012), 15–34.
[21] R.R. Nigmatullin, The realization of the generalized transfer equation in a medium with
fractal geometry, Phys. Status Solidi 133 (1986), 425–430.
[22] A. Pazy, Semigroups of Linear Operators and Applications to Partial Differential Equa-
tions, Springer–Verlag, New York, 1983.
[23] I. Podlubny, Fractional Differential Equations, Academic Press, New York, 1999.
[24] K. Sakamoto and M. Yamamoto, Initial value/boundary value problems for fractional
diffusion-wave equations and applications to some inverse problems, J. Math. Anal. Appl.
382 (2011), 426–447.
[25] W.R. Schneider and W. Wyss, Fractional diffusion and wave equations, J. Math. Phys.
30 (1989), 134–144.
[26] R.N. Wang, D.H. Chen and T.J. Xiao, Abstract fractional Cauchy problems with almost
sectorial operators, J. Differential Equations 252 (2012), 202–235.
92 Q.G. Zhang — H.R. Sun

[27] F.B. Weissler, Existence and nonexistence of global solutions for a semilinear heat equa-
tion, Israel J. Math. 38 (1981), 29–40.
[28] G.M. Zaslavsky, Fractional kinetic equation for Hamiltonian chaos, Phys. D 76 (1994),
110–122.
[29] Y. Zhou and F. Jiao, Existence of mild solutions for fractional neutral evolution equa-
tions, Computers Math. Appl. 59 (2010), 1063–1077.

Manuscript received January 6, 2013

Quan-Guo Zhang
School of Mathematics and Statistics
Key Laboratory of Applied Mathematics and Complex Systems
Lanzhou University
Lanzhou, Gansu 730000, P.R. CHINA
and
Department of Mathematics
Luoyang Normal University,
Luoyang, Henan 471022, P.R. CHINA

Hong-Rui Sun
School of Mathematics and Statistics
Key Laboratory of Applied Mathematics and Complex Systems
Lanzhou University
Lanzhou, Gansu 730000, P.R. CHINA
E-mail address: hrsun@lzu.edu.cn

TMNA : Volume 46 – 2015 – No 1