DISCRETE AND CONTINUOUS doi:10.3934/dcds.2016.36.

1279
DYNAMICAL SYSTEMS
Volume 36, Number 3, March 2016 pp. 1279–1319

REACTION-DIFFUSION EQUATIONS WITH FRACTIONAL

DIFFUSION ON NON-SMOOTH DOMAINS WITH VARIOUS
BOUNDARY CONDITIONS

Ciprian G. Gal
Department of Mathematics
Florida International University
Miami FL 33199, USA

Mahamadi Warma
University of Puerto Rico
Rio Piedras Campus
Department of Mathematics
P.O. Box 70377, San Juan PR 00936-8377, USA

(Communicated by Manuel del Pino)

Abstract. We investigate the long term behavior in terms of finite dimen-

sional global attractors and (global) asymptotic stabilization to steady states,
as time goes to infinity, of solutions to a non-local semilinear reaction-diffusion
equation associated with the fractional Laplace operator on non-smooth do-
mains subject to Dirichlet, fractional Neumann and Robin boundary condi-
tions.

1. Introduction. The main concerns in the present paper are to investigate the
existence, the regularity and the long-time behavior of solutions to some non-local
reaction-diffusion equations associated with the fractional Laplace operator with
Dirichlet, fractional Neumann and fractional Robin type boundary conditions on
non-smooth subsets of RN . In order to introduce the fractional Laplacian, let
0 < s < 1, Ω ⊂ RN an arbitrary open set and set
|u(x)|
Z
1
L (Ω) := {u : Ω → R measurable, N +2s
dx < ∞}.
Ω (1 + |x|)
For u ∈ L1 (RN ), x ∈ RN and ε > 0, we write
u(x) − u(y)
Z
(−∆)sε u(x) = CN,s dy
{y∈RN ,|y−x|>ε} |x − y|N +2s
with the normalized constant CN,s given by
N +2s


s22s Γ 2
CN,s = N ,
π 2 Γ(1 − s)

2010 Mathematics Subject Classification. 35R11, 35A15, 35B41, 35K65.

Key words and phrases. The fractional Laplace operator, fractional Neumann and Robin
boundary conditions on open sets, global attractor, convergence to steady states, semi-linear
reaction-diffusion equation, asymptotic behavior.
The work of the second author is partially supported by the Air Force Office of Scientific
Research under the Award No: FA9550-15-1-0027.

1279
1280 CIPRIAN G. GAL AND MAHAMADI WARMA

where Γ denotes the usual Gamma function. The fractional Laplacian (−∆)s u of
the function u is defined by the formula
u(x) − u(y)
Z
(−∆)s u(x) = CN,s P.V. N +2s
dy = lim(−∆)sε u(x), x ∈ RN , (1.1)
RN |x − y| ε↓0

provided that the limit exists. We notice that if 0 < s < 1/2 and u is smooth
(for example, Lipschitz continuous), then the integral in (1.1) is in fact not singular
near x. We also recall that in the whole space RN , using the Fourier transform,
(−∆)s can be also defined as a pseudo-differential operator with symbol |ξ|2s . If
one wishes to consider the fractional Laplace operator (−∆)s on open subsets Ω of
RN it cannot be used on Ω automatically due to its nonlocal character. In order
to give a proper definition, we follow [28, 29, 30, 48] in the following fashion. Let
Ω ⊂ RN be an arbitrary open set. We restrict the integral kernel of the fractional
Laplacian to the open set Ω. For u ∈ L1 (Ω), x ∈ Ω and ε > 0, we let
u(x) − u(y)
Z
s
AΩ,ε u(x) = CN,s dy,
{y∈Ω,|y−x|>ε} |x − y|N +2s
and we define the operator
u(x) − u(y)
Z
AsΩ u(x) = CN,s P.V. dy = lim AsΩ,ε u(x), x ∈ Ω, (1.2)
Ω |x − y|N +2s ε↓0

provided that the limit exists. As in the case RN , if s ∈ (0, 1/2) and u is smooth
then the integral in (1.2) is not singular near x. We call the operator AsΩ the regional
fractional Laplacian (cf. [28, 29, 30]). The regional fractional p-Laplace operator
with p ∈ (1, ∞) has been also introduced in [49]. Let now u ∈ D(Ω), the space
of infinitely continuously differentiable functions with compact support in Ω. Since
u = 0 on RN \Ω, a simple calculation gives
u(x) − u(y)
Z
AsΩ u(x) := CN,s P.V. dy
Ω |x − y|N +2s
u(x) − u(y)
Z Z
u(x)
= CN,s P.V. N +2s
dy − CN,s dy
RN |x − y| N
R \Ω |x − y|N +2s
= (−∆)s u(x) − VΩ (x)u(x),
where the potential VΩ is given by
Z
VΩ (x) := CN,s |x − y|−N −2s dy, x ∈ Ω. (1.3)
RN \Ω

More precisely, we have

(−∆)s u(x) = AsΩ u(x) + VΩ (x)u(x), for all u ∈ D(Ω). (1.4)
The operator AsΩdescribes a particle jumping from one point x ∈ Ω to another y ∈ Ω
−N −2s
with intensity proportional to |x − y| . Parabolic problems associated with
s
the operator AΩ have been intensively studied in [6, 11, 29, 30] and the references
therein, employing probabilistic approaches, and in [48] by using the method of
Dirichlet forms on non-smooth domains. Based on (1.4), we then view the fractional
Laplacian (−∆)s with domain D(Ω) as a perturbation of the regional fractional
operator AsΩ with the non-negative potential VΩ . Recently, various elliptic and
parabolic equations associated with the fractional Laplace operator with Dirichlet
boundary conditions were also investigated by Caffarelli et al. [8, 9, 10].
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1281

In this paper, we shall be concerned with non-local diffusion processes associated

with the fractional Laplace operator with various boundary conditions. To be more
precise, we consider diffusion processes described by the following systems

s
∂t u + d(−∆) u + f (u) = 0
 in Ω × (0, ∞) ,
u=0 on (RN \Ω) × (0, ∞), (1.5)

u (0) = u0 in Ω,


and 
s
∂t u + dAΩ u + f (u) = 0
 in Ω × (0, ∞) ,
u=0 on ∂Ω × (0, ∞), (1.6)

u (0) = u0 in Ω,


and 
s
∂t u + dAΩ u + f (u) = 0
 in Ω × (0, ∞) ,
2−2s
dBN,s N u + γu = 0 on ∂Ω × (0, ∞), (1.7)

u (0) = u0 in Ω.


In (1.5), (1.6) and (1.7), f = f (u) plays the role of nonlinear source, not necessarily
monotone and d > 0 is a diffusion coefficient. In (1.5), the operator (−∆)s denotes
the fractional Laplace operator defined in (1.1) and in (1.6) and (1.7), AsΩ is the
regional fractional Laplacian given by (1.2). Finally in (1.7), N 2−2s u denotes the
fractional normal derivative of the function u in direction of the outer normal vector
(see Section 2 below), BN,s is a normalized constant (see (2.17) below) and γ ∈
L∞ (∂Ω) is a non-negative function.
Our motivation for considering such problems is two-fold. First, the systems
(1.5)-(1.7) and their stationary versions have been used recently to describe the
motion of nonlinear deflects in crystalline materials in the field of dislocation dy-
namics (see, e.g., [26, 36]). In the theory of phase-field and interfacial dynamics,
these equations are usually referred as the fractional Allen-Cahn equation (see, e.g.,
[31, 41]). Moreover, nonlocal reaction-diffusion equations have been also considered
in the monograph [4] but the integral operators there are generally smooth or only
mildly singular (i.e., the kernel is at least integrable over RN ). On the other hand,
s
the linear parabolic equation ∂t u + (−∆) u = 0, s ∈ (0, 1), instead of the usual
parabolic equation ∂t u − ∆u = 0, is a much studied topic of anomalous diffusion in
physics, probability and finance (see, e.g., [1, 33, 40, 42]). We also refer the reader
to an interesting tutorial in [47] which introduces the main concepts behind normal
and anomalous diffusion. Second, our work is further motivated by the need to de-
velop a complete dynamical theory for these problems where not much seems to be
known about basic issues, such as global existence and regularity, uniqueness, blow-
up phenomena and longtime behavior of solutions, as time goes to infinity. This
seems to be due to the fact that the parabolic structure of (1.5)-(1.7) has not been
exploited before, an issue which is intimately connected with an L2 -L∞ smoothing
result in (0, ∞) × Ω of solutions. This is essential to the study of the asymptotic
behavior of these systems, in terms of global attractors and ω-limit sets. We also
emphasize the generality of our results by assuming only minimal conditions on the
regularity of Ω: we shall assume Ω to be simply an open subset of RN in the case
of problems (1.5)-(1.6), while in the third case (1.7) it suffices to assume that ∂Ω
is Lipschitz continuous. Our current contribution is also motivated by our recent
work on parabolic equations with classical diffusion on rough domains and nonlocal
boundary conditions (see [24]).
1282 CIPRIAN G. GAL AND MAHAMADI WARMA

The main novelties of the present paper are the following:

(I) We show the existence of global and unique strong solutions (which are Lips-
chitz continuous in time for t ∈ (0, ∞)) for any of the preceding problems provided
that the nonlinearity f ∈ C 1 (R) obeys
f (τ )
lim inf > −λ∗ , (1.8)
|τ |→+∞ τ
for some constant λ∗ ∈ [0, Cs ), where roughly speaking Cs = C (Ω, N, s) > 0
is the best Sobolev/Poincaré constant in the embedding W s,2 (Ω) ⊂ L2 (Ω) (see
Section 3). This condition turns out to be optimal in the sense that if it is violated
by some function f then blow-up of some strong solutions occurs in finite time
(see Section 6). For the latter issue, our approach is based on the concavity and
eigenvalue methods inspired by [37] and [35], respectively. Global existence of strong
solutions of (1.5)-(1.7) under assumption (1.8) is deduced by performing a Moser-
type iteration procedure as in, e.g., [24] for the case of the Laplace operator. Here a
new inequality of Poincaré type and a comparison lemma for various energy forms
allows to control the iteration at every step (see Section 3.1). These facts together
with some classical arguments allow us also to develop a complete weak solution
theory, say when f is a polynomial of arbitrary degree satisfying the additional
0
condition f (τ ) ≥ −cf , for some cf > 0, for all τ ∈ R.
(II) We prove that every weak solution of the problems (1.5)-(1.7) “instanta-
neously” regularizes to a strong solution in both space and time. Then, taking
advantage of this smoothness, we can show that our problems have a gradient struc-
ture, and as a result establish the existence of a finite-dimensional global attractor
in the phase space L2 (Ω) . In some cases, owing to the L2 -L∞ smoothing property
of these equations we can also establish explicit both sided estimates for the fractal
dimension of the global attractor. In particular, for problem (1.5) we establish the
(sharp) two-sided estimate for the fractal dimension of the global attractor As :
N
! 2s
0 N
−f (0)  c  2s
f
|Ω| ≤ dimF As , L2 (Ω) ≤ c2


c1 |Ω| , (1.9)
d d
provided that f is a polynomial density function of arbitrary growth such that
0 0
f (τ ) ≥ −cf , for any τ ∈ R, assuming f (0) < 0 as well. Here, the constants
c1 , c2 > 0 depend on the shape of Ω and N only, but are independent of s. We
recall that the dimension of the global attractor can be used in practice to indicate
the number of degrees of freedom needed to simulate the given dynamical system
since this dimension is usually associated with the temporal and spatial complexity
of the long-time dynamics. Thus, the bound (1.9) indicates that the “permanent
regime” for problem (1.5) is indeed more structurally complex when compared to
that of the classical reaction-diffusion equation for the Laplace operator ∆. It is also
worth noting that the foregoing bounds also stabilize as s → 1− to the (classical)
dimension bounds for the well-established reaction-diffusion equation associated
with the operator ∆.
(III) The ω-limit sets of the problems (1.5)-(1.7) can exhibit a complicated struc-
ture if the function f is non-monotone. This can happen if the stationary problems
associated with (1.5)-(1.7) possess a continuum of nonconstant solutions. We show
the validity of the so-called Lojasiewicz-Simon inequality for our problems under
the assumption that a certain elliptic boundary value problem has Hölder continu-
ous (up to the boundary ∂Ω) solutions. To be more precise, we need only require
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1283

that the bounded solution of

AK w = h ∈ L∞ (Ω) , in Ω,


has the property: w ∈ C 0,ν Ω , for some ν ∈ (0, 1). We also give an example when
such a condition is met. Then, for any of these problems we prove the convergence
of a given trajectory u = u(t; u0 ), u0 ∈ L2 (Ω) , as time goes to infinity, to a single
equilibrium which solves the corresponding stationary version associated with (1.5)-
(1.7). More precisely, assuming also that f is a real analytic function over R, any
weak solution u to problems (1.5)-(1.7) satisfies
− ζ1
ku (t) − u∗ kL∞ (Ω) ∼ (1 + t) as t → ∞,

where ζ ∈ (0, 1) depends on u∗ ∈ L∞ (Ω) ∩ W s,2 (Ω), such that u∗ solves the
corresponding stationary problems.
Finally, we can also mention that the present analysis can be exploited to extend
and establish existence and existence of finite dimensional attractor results for sys-
tems of reaction-diffusion equations for a vector valued function → −
u = (u1 , ..., uk )
(k ≥ 2). For instance, our framework requires only minor modifications: the func-
tion spaces become product spaces, and the principal dissipation operators become
block operators on these product spaces, typically with block diagonal form. The
nonlinearities in these models can be treated in a similar way as in Section 3.
We also remark that
one can also allow for time-dependent external forces h (t) ,
h ∈ Cb R; L2 (Ω) , acting on the right-hand side of these systems. In this case,
one can generalize the notion of global attractor and replace it by the notion of
pullback attractor, for example. One can still study the set of all complete bounded
trajectories, that is, trajectories which are bounded for all t ∈ R+ . We leave the
details to the interested reader.
The plan of the paper goes as follows. In Section 2, we introduce the functional
analytic framework associated with (1.5)-(1.7). Then, in Section 3 (and correspond-
ing subsections) we prove well-posedness and regularity results for any of the prob-
lems (1.5)-(1.7). In Section 4 we establish the existence of a compact semiflow in
L2 (Ω) and derive optimal estimates on the global attractor, while in the remaining
Sections 5, 6 we deal with convergence of solutions and with blow-up phenomena,
respectively.

2. Preliminaries. In this section we introduce the function spaces needed to in-

vestigate our problems and we show some intermediate results. In particular we
introduce the fractional normal derivative of a function u mentioned in the intro-
duction. We also give the integration by parts formula for the regional fractional
Laplacian and we introduce the fractional Neumann and Robin type boundary con-
ditions associated with the operator AsΩ . Some generation of semigroup results, and
the regularity of weak solutions of elliptic equations associated with these operators
are also given.

2.1. The functional setup. Let Ω ⊂ RN be an arbitrary open set. For s ∈ (0, 1),
we denote by
|u(x) − u(y)|2
Z Z
W s,2 (Ω) := {u ∈ L2 (Ω) : N +2s
dxdy < ∞}
Ω Ω |x − y|
1284 CIPRIAN G. GAL AND MAHAMADI WARMA

the fractional order Sobolev space endowed with the norm

 12
|u(x) − u(y)|2
Z Z Z
CN,s
kukW s,2 (Ω) := |u|2 dx + N +2s
dxdy .
Ω 2 Ω Ω |x − y|

If Ω is a bounded open set with a Lipschitz continuous boundary, then by [16,

Theorem 6.7], there exists a constant C > 0 such that for every u ∈ W s,2 (Ω),
kukLq (Ω) ≤ CkukW s,2 (Ω) , (2.1)
for all q satisfying
2N
q ∈ [1, 2? ] with 2? := if N > 2s and q ∈ [1, ∞) if N = 2s.
N − 2s
By [16, Section 7], (2.1) implies that for every q ∈ [1, 2? ), the embedding W s,2 (Ω) ,→
Lq (Ω) is compact. Moreover, by [14, Theorem 11.1], there exists a constant C > 0
such that for every u ∈ W s,2 (Ω),
kukLr (∂Ω) ≤ CkukW s,2 (Ω) , (2.2)
for all r satisfying
2(N − 1)
r ∈ [1, 2? ] with 2? := if N > 2s and r ∈ [1, ∞) if N = 2s.
N − 2s
If N < 2s, then
N
W s,2 (Ω) ,→ C 0,α (Ω) with α := s − .
2
For an arbitrary open set Ω ⊂ RN , we let
W s,2 (Ω)
W0s,2 (Ω) = D(Ω) .

By definition, W0s,2 (Ω) is the smaller closed subspace of W s,2 (Ω) containing D(Ω).
By [16, Remark 6.6], there exists a constant C > 0 such that for every u ∈ W0s,2 (Ω),
kukLq (Ω) ≤ CkukW s,2 (Ω) , ∀ q ∈ [2, 2? ]. (2.3)
In particular, if Ω is bounded, we have that (2.3) holds for every q ∈ [1, 2? ]. In
that case, by [16, Section 7] again, (2.3) also implies that for every q ∈ [1, 2? ), the
s,2 N

embedding W0s,2 (Ω) ,→ Lq (Ω) is compact. Let W f s,2 (Ω) = D(Ω)W (R ) . By [2,
0
Theorem 10.1.1], it can be characterized as follows:
f s,2 (Ω) = {u ∈ W s,2 (RN ) : ũ = 0 on RN \Ω},
W (2.4)
0

where ũ is the quasi-continuous version (with respect to the capacity defined with
the space W s,2 (RN )) of u. Then W f s,2 (Ω) ⊂ W s,2 (Ω) and it is well known that they
0 0
coincide if s 6= 1/2 and they may be different if s = 1/2. Finally, we mention that
in the case where W f s,2 (Ω) = W s,2 (Ω), we have that for u ∈ W s,2 (Ω),
0 0 0
2
|u(x) − u(y)|
Z Z Z
CN,2
k|uk|2 = N +2s
dxdy + VΩ |u|2 dx (2.5)
2 Ω Ω |x − y| Ω
|u(x) − u(y)|2
Z Z
CN,2
= N +2s
dxdy
2 RN RN |x − y|
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1285

defines an equivalent norm on the space W0s,2 (Ω). In fact, by [16, Lemma 6.1], there
exists a constant C = C(Ω, N, s) > 0 such that VΩ (x) ≥ C for every x ∈ Ω. This
implies that for every u ∈ W0s,2 (Ω),

|u(x) − u(y)|2
Z Z Z
CN,2
kuk2W s,2 (Ω) = N +2s
dxdy + |u|2 dx
0 2 RN RN |x − y| Ω
|u(x) − u(y)|2
 Z Z Z 
CN,2 2
≤C N +2s
dxdy + V Ω |u| dx
2 RN RN |x − y| Ω
= Ck|uk|2 .

Let u ∈ W0s,2 (Ω) and let ũ be the extension of u by 0 on RN \Ω. Then ũ ∈ W s,2 (RN )
and since the extension operator is continuous from W0s,2 (Ω) into W s,2 (RN ) (see e.g.
[16, Section 6]) we have that there exists a constant C > 0 such that kũkW s,2 (RN ) =
k|uk| ≤ CkukW s,2 (Ω) for every u ∈ W0s,2 (Ω). This completes the proof of (2.5).
0

Remark 1. We notice that, if Ω has a Lipschitz continuous boundary, then by

[6] (see also [48, Theorem 4.8] for a more general version), W0s,2 (Ω) = W s,2 (Ω) for
every 0 < s ≤ 21 .

For more information on the fractional order Sobolev spaces we refer to [2, 14,
16, 27, 34, 38, 48] and their references.

2.2. The fractional Laplacian with Dirichlet boundary conditions. Let Ω ⊂

RN be an arbitrary bounded domain. Let 0 < s < 1 and let VΩ be the potential
given in (1.3). Throughout the remainder of the article, if we write u ∈ Wf s,2 (Ω),
0
we mean that u ∈ W s,2 (RN ) and u = 0 on RN \Ω.
Let EE be the bilinear symmetric closed form with domain D(EE ) = W f s,2 (Ω)
0
s,2
and defined for u, v ∈ W
f (Ω) by
0

(u(x) − u(y))(v(x) − v(y))

Z Z
CN,s
EE (u, v) = dxdy
2 |x − y|N +2s
Z Ω Ω
+ VΩ (x)u(x)v(x)dx
Ω
(u(x) − u(y))(v(x) − v(y))
Z Z
CN,s
= dxdy.
2 R N RN |x − y|N +2s

Let AE be the closed linear selfadjoint operator on L2 (Ω) associated with EE in

the sense that
(
D(AE ) := {u ∈ W f s,2 (Ω), ∃ v ∈ L2 (Ω), EE (u, ϕ) = (v, ϕ)L2 (Ω) ∀ ϕ ∈ W
f s,2 (Ω)}
0 0
AE u = v.
(2.6)
We call AE a realization of the fractional Laplace operator (−∆)s on L2 (Ω) with
the Dirichlet boundary condition. We have the following more explicit description
of the operator AE .

Proposition 2.1. Let AE be the operator defined in (2.6). Then

f s,2 (Ω), (−∆)s u ∈ L2 (Ω)}, AE u = (−∆)s u.
D(AE ) = {u ∈ W (2.7)
0
1286 CIPRIAN G. GAL AND MAHAMADI WARMA

Proof. Set D e := {u ∈ W f s,2 (Ω), (−∆)s u ∈ L2 (Ω)}. Let u ∈ D(AE ). Then by

0
f s,2 (Ω),
definition, there exists a function v ∈ L2 (Ω), such that for every ϕ ∈ W 0
(u(x) − u(y))(ϕ(x) − ϕ(y))
Z Z Z Z
CN,s
vϕdx = vϕ dx = dxdy
Ω RN 2 RN RN |x − y|N +2s
Z Z
= (−∆)s/2 u(−∆)s/2 ϕdx = ϕ(−∆)s udx.
RN RN

We have shown that u ∈ D e and AE u := v = (−∆)s u. Now, let u ∈ D e and set

s 2 s,2
v := (−∆) u ∈ L (Ω). Let ϕ ∈ W f (Ω). Then
0
Z Z Z Z
s s
vϕdx = ϕ(−∆) udx = ϕ(−∆) udx = (−∆)s/2 u(−∆)s/2 ϕdx
Ω Ω RN RN
(u(x) − u(y))(ϕ(x) − ϕ(y))
Z Z
CN,s
= dxdy = EE (u, ϕ).
2 RN R N |x − y|N +2s
We have shown that D e ⊂ D(AE ) and the proof of (2.7) is complete.

Next, let ED be the bilinear symmetric closed form with domain D(ED ) =
W0s,2 (Ω) and defined for u, v ∈ W0s,2 (Ω) by
(u(x) − u(y))(v(x) − v(y))
Z Z
CN,s
ED (u, v) = dxdy.
2 Ω Ω |x − y|N +2s
Let AD be the closed linear selfadjoint operator on L2 (Ω) associated with ED in
the sense that
(
D(AD ) := {u ∈ W0s,2 (Ω), ∃ v ∈ L2 (Ω), ED (u, ϕ) = (v, ϕ)L2 (Ω) ∀ ϕ ∈ W0s,2 (Ω)}
AD u = v.
(2.8)
We call AD a realization of the regional fractional Laplace operator AsΩ on L2 (Ω)
with the Dirichlet boundary condition. We have the following more explicit descrip-
tion of the operator AD .
Proposition 2.2. Let AD be the operator defined in (2.8). Then
D(AD ) = {u ∈ W0s,2 (Ω), AsΩ u ∈ L2 (Ω)} and AD u = AsΩ u. (2.9)
Proof. The proof Rfollows as the proof of Proposition 2.1 by using also the integration
by part formula Ω vAsΩ u dx = ED (u, v) for every u, ϕ ∈ W0s,2 (Ω) with AsΩ u ∈
L2 (Ω).
We need not make any confusion between the operator AE and AD . They are
different and coincide only if RN \Ω has capacity zero with respect to the capacity
defined with the space W s,2 (RN ) (of course, this cannot be the case since Ω is
bounded). On one hand, we have shown that AE u = AD u+VΩ u, for every u ∈ D(Ω).
On the other hand, the potential VΩ is in general very difficult to describe. For
example, if Ω has a Lipschitz continuous boundary then it has been shown in [27,
Formula (1.3.2.12), p. 19] that there exist some constants 0 < C1 ≤ C2 such that
for every x ∈ Ω,
C1 ρ−2s (x) ≤ VΩ (x) ≤ C2 ρ−2s (x),
where ρ(x) := dist(x, ∂Ω), x ∈ Ω. Instead of the fractional Laplace operator (−∆)s ,
whose definition is independent of the open set Ω, the regional fractional Laplace
operator AsΩ depends on Ω and hence on the potential VΩ . But we have the following
convergence result.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1287

Proposition 2.3. Let Ω ⊂ RN be an arbitrary bounded open set. Then for every
u ∈ D(Ω) and v ∈ W01,2 (Ω), we have that
Z Z Z
lim vAsΩ udx = lim v(−∆)s udx = − v∆udx. (2.10)
s↑1 Ω s↑1 Ω Ω

Proof. First, let u ∈ D(Ω). Then using [7], the fact that lims↑1 (1 − s)Γ(1 − s) = 1
and the classical integration by part formula for the Laplace operator, we get that
s22s−1 Γ N +2s


|u(x) − u(y)|2
Z Z Z
s 2
lim uAΩ udx = lim N (1 − s) N +2s
dxdy (2.11)
s↑1 Ω s↑1 π 2 (1 − s)Γ(1 − s) Ω Ω |x − y|
Z Z
= |∇u|2 dx = − u∆udx.
Ω Ω
Proceeding as in (2.11), we also have that
Z Z Z Z
s 2
lim u(−∆) udx = |∇u| dx = − u∆udx = − u∆udx. (2.12)
s↑1 Ω RN RN Ω
We have show (2.10) for u = v ∈ D(Ω). Replacing u by u + v in (2.11) and (2.12)
for u, v ∈ D(Ω), we get (2.10) for every u, v ∈ D(Ω). Finally we get (2.10) for every
u ∈ D(Ω) and v ∈ W01,2 (Ω) by density and using that W01,2 (Ω) is continuously
embedded into W0s,2 (Ω). The proof is finished.
2.3. The fractional normal derivative. In this subsection, we introduce the
fractional normal derivative mentioned in the introduction. This will be used to
define the fractional Neumann and Robin boundary conditions for the operator AsΩ .
Throughout the remainder of this subsection, Ω ⊂ RN denotes a bounded open set
of class C 1,1 and we will also use the following notations:
ρ(x) = dist(x, ∂Ω) = inf{|y − x| : y ∈ ∂Ω}, ∀ x ∈ Ω,
Ωδ = {x ∈ Ω : 0 < ρ(x) < δ}, δ > 0 is a real number,
~n(z) = the inner normal vector of ∂Ω at the point z ∈ ∂Ω,
ν(z) = −~n(z) the outer normal vector of ∂Ω at the point z ∈ ∂Ω.
The following definition is taken from [28, Definition 2.1] (see also [29, Definition
7.1] for the one-dimensional case).
Definition 2.1. For u ∈ C 1 (Ω), z ∈ ∂Ω and 0 ≤ α < 2, we define the boundary
operator N α by
du(z + ~n(z)t) α
N α u(z) = − lim t , (2.13)
t↓0 dt
whenever the limit exists.
Remark 2. Let 0 ≤ α < 2 and let N α be the boundary operator defined in (2.13).
(a) If α = 0, then N 0 u(z) = −∇u·~n(z) = ∂u(z) 1
∂ν for every u ∈ C (Ω) and z ∈ ∂Ω.
α 1
(b) If 0 < α < 2, then N u(z) = 0 for every u ∈ C (Ω) and z ∈ ∂Ω.
Next, let β > 0. By [28, p.294], there exist a real number δ > 0 (depending only
on Ω) and a function hβ ∈ C 2 (Ω) (depending on Ω and β) such that
(
ρ(x)β−1 , ∀ x ∈ Ωδ , when β ∈ (0, 1) ∪ (1, ∞);
hβ (x) = (2.14)
ln(ρ(x)), ∀ x ∈ Ωδ , when β = 1.
1288 CIPRIAN G. GAL AND MAHAMADI WARMA

For β > 0 we define the space

Cβ2 (Ω) = {u : u(x) = f (x)hβ (x) + g(x), ∀ x ∈ Ω, for some f, g ∈ C 2 (Ω)}.
When β > 1, we always assume that u ∈ Cβ2 (Ω) is defined on Ω by continuous
extension. The following explicit representation of the operator N α is taken from
[48, Lemma 6.3].
Lemma 2.2. Let 1 < β ≤ 2, u := f hβ + g ∈ Cβ2 (Ω) be the representation of u. Let
u0 := f hβ so that u = u0 + g. Then the following assertions hold.
(a) If β ∈ (1, 2), then for z ∈ ∂Ω,
u(x) − u(z) u0 (x)
N 2−β u(z) = (1 − β) lim = (1 − β) lim . (2.15)
Ω3x→z ρ(x)β−1 Ω3x→z ρ(x)β−1

(b) If β = 2, then for z ∈ ∂Ω,

u(x) − u(z)
N 0 u(z) = − lim . (2.16)
Ω3x→z ρ(x)
Next, let
∞
|τ − 1|1−2s − (τ ∨ 1)1−2s
Z
C1,s 1
Cs := dτ, < s < 1,
2s(2s − 1) 0 τ 2−2s 2
and let the constant BN,s be such that

C1,s C s
 if N = 1
N −1
BN,s := 2π 2 Cs π/2 (2.17)
Z
CN,s cos2s (θ) sinN −2 (θ) dθ, if N ≥ 2.
Γ( N 2−1 ) 0



We have the following fractional Green type formula for the regional fractional
Laplace operator.
Theorem 2.3. Let 12 < s < 1 and let AsΩ be the nonlocal operator defined in (1.2).
2
Then, for every u := f h2s + g = u0 + g ∈ C2s (Ω) and v ∈ W s,2 (Ω),
(u(x) − u(y))(v(x) − v(y))
Z Z Z
1
v(x)AsΩ u(x)dx = CN,s dxdy
Ω 2 Ω Ω |x − y|N +2s
Z
− BN,s vN 2−2s u dσ
∂Ω
(u(x) − u(y))(v(x) − v(y))
Z Z
1
= CN,s dxdy
2 Ω Ω |x − y|N +2s
u(x) − u(z)
Z
+ BN,s (2s − 1) v(z) lim dσz
∂Ω x→z ρ(x)2s−1
(u(x) − u(y))(v(x) − v(y))
Z Z
1
= CN,s dxdy
2 Ω Ω |x − y|N +2s
Z  
u0
+ BN,s (2s − 1) v(z) (z)dσz ,
∂Ω ρ2s−1
 
u0
where by ρ2s−1 (z) at the point z ∈ ∂Ω, we mean
 
u0 u0 (x)
2s−1
(z) = lim .
ρ Ω3x→z ρ(x)2s−1
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1289

We mention that the first identity in Theorem 2.3 has been obtained in [28,
2
Theorem 3.3] under the assumption that v also belongs to C2s (Ω). Its validity for
s,2
every v ∈ W (Ω) and the second and third identities have been proved in [48,
Theorem 5.7].
Definition 2.4. For 12 < s ≤ 1 and u ∈ C2s 2
(Ω), we call the function BN,s N 2−2s u
the fractional normal derivative of the function u in direction of the outer normal
vector.
We make some comments about the fractional normal derivative introduced
above.
Remark 3. We mention that another definition of fractional normal derivative,
called non-local normal derivative, has been introduced in [18, 32] (see also [17]) for
functions u defined on RN . More precisely, for 0 < α < 1 and u ∈ L1 (RN ), the
non-local normal derivative is defined by
u(x) − u(y)
Z
Nα u(x) = CN,s N +2α
dy, x ∈ RN \Ω. (2.18)
Ω |x − y|
The definition of Nα u in (2.18) requires that the function is defined on all RN .
This is different from the fractional Normal derivative N α u given in (2.13) where
the function u is defined only on Ω. Starting with a function defined only on Ω,
it seems impossible to deal with Nα u. For example if u ∈ W s,2 (Ω) and letting
ũ ∈ W s,2 (RN ) be an extension to all RN , then the relation (2.18) can make sense
but the definition cannot be independent of the extension, except in the case where
there is only one such possible extension. This shows that the expression Nα u
cannot be used in our context since we consider functions defined a priori only on
Ω. We recall that it has been shown in [17, Proposition 5.1] (see also [18, 32]) that
if Ω ⊂ RN is a bounded domain with Lipschitz continuous boundary ∂Ω, then for
every u, v ∈ C02 (RN ),
Z Z
∂u
lim vNα u dx = v dσ.
α↑1 RN \Ω ∂Ω ∂ν
As we have seen in Remark 2, the fractional normal derivative N α u is continuous
with respect to α, so that for every u ∈ C22 (Ω) = C 1 (Ω) we have that N 1 u = ∂u ∂ν ,
i.e., the classical normal derivative of the function u in direction of the outer normal
vector ν. Next, let BN,s be the constant given in (2.17). First, we notice that using
a change of variable, we get that
Z π/2
1 1 s+ 1 −1
Z
N −1
N −2
2
cos s(θ) sin (θ)dθ = t 2 (1 − t) 2 −1 dt (2.19)
0 2 0
 
1 2s + 1 N − 1
= B ,
2 2 2
2s+1 N −1



1Γ 2 Γ 2
= ,
Γ N +2s


2 2
where B denotes the usual Beta function. Replacing this expression (2.19) in (2.17),
we get that in fact BN,s = Cs and hence, it is independent of N . Moreover, we
have that lims↑1 Cs = 1. This shows that the integration by parts formula given
in Theorem 2.3 is consistent with the well-known integration by part formula for
the Laplace operator where there is no constant depending on the dimension in the
boundary integral.
1290 CIPRIAN G. GAL AND MAHAMADI WARMA

2.4. The fractional Neumann boundary conditions. Throughout this section,

we assume that Ω ⊂ RN is a bounded domain with Lipschitz continuous boundary
∂Ω. We consider the bilinear symmetric closed form EN with domain D(EN ) =
W s,2 (Ω) and given for u, v ∈ W s,2 (Ω) by
(u(x) − u(y))(v(x) − v(y))
Z Z
CN,s
EN (u, v) = dxdy.
2 Ω Ω |x − y|N +2s
Since W s,2 (Ω) = W0s,2 (Ω) for all 0 < s ≤ 1/2 (by Remark 1), we have that EN = ED
if 0 < s ≤ 1/2. Therefore, we assume that 1/2 < s < 1.
Let AN be the closed linear selfadjoint operator associated with EN in the sense
that
(
D(AN ) := {u ∈ W s,2 (Ω), ∃ v ∈ L2 (Ω), EN (u, ϕ) = (v, ϕ)L2 (Ω) , ∀ ϕ ∈ W s,2 (Ω)}
AN u = v.
(2.20)
We call AN a realization of the regional fractional Laplace operator AsΩ on L2 (Ω)
with the fractional Neumann type boundary conditions. In fact, we have the fol-
lowing more explicit description of the operator AN which has been proved in [48,
Proposition 6.1] exploiting Theorem 2.3.
Proposition 2.4. Let AN be the operator defined in (2.20). Assume also that Ω
is a bounded open set of class C 1,1 . Then
2 2
D(AN ) ∩ C2s (Ω) = {u ∈ C2s (Ω), N 2−2s u = 0 on ∂Ω}, AN u = AsΩ u.
The following result shows in particular, that as s ↑ 1, the operator AN converges
(in some sense) to the realization ∆N in L2 (Ω) of the Laplace operator with the
classical Neumann boundary conditions.
Proposition 2.5. For every u ∈ C 2 (Ω) and v ∈ W 1,2 (Ω) we have that
Z Z Z Z
∂u
lim vAsΩ udx = ∇u∇vdx = − v∆udx + vdσ. (2.21)
s↑1 Ω Ω Ω ∂Ω ∂ν

Proof. Since Ω has a Lipschitz continuous boundary, we have that W 1,2 (Ω) ,→
W s,2 (Ω) (see e.g. [16, Proposition 2.2]). Let u ∈ C 2 (Ω). Since u ∈ C 2 (Ω), then
by Remark 2, we have that N 2−2s u(z) = 0 for every z ∈ ∂Ω. Then, using the
definition, the integration by part formula for the operator AsΩ given in Theorem
2.3, the convergence result of fractional order Sobolev spaces contained in [7], the
fact that lims↑1 (1 − s)Γ(1 − s) = Γ(1) = 1 and the integration by part formula for
the Laplace operator, we have that
|u(x) − u(y)|2
Z Z Z
s 1
lim uAΩ udx = lim CN,s N +2s
dxdy (2.22)
s↑1 Ω 2 s↑1 Ω Ω |x − y|
s22s Γ N +2s


(1 − s) |u(x) − u(y)|2
Z Z
1 2
= lim N dxdy
2 s↑1 π 2 (1 − s)Γ(1 − s) Ω Ω |x − y|N +2s
Z Z Z
∂u
= |∇u|2 dx = − u∆udx + udσ.
Ω Ω ∂Ω ∂ν

It follows from (2.22) that for every u, v ∈ C 2 (Ω), we have that

Z Z Z
s ∂u
lim vAΩ udx = − v∆udx + vdσ.
s↑1 Ω Ω ∂Ω ∂ν

Now we obtain (2.21) for every u ∈ C 2 (Ω) and v ∈ W 1,2 (Ω) by density.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1291

As we have mentioned in Remark 3, since our functions are a priori defined only
on Ω, we have that N 2−2s u = 0 is the right fractional homogeneous Neumann type
boundary conditions for the regional fractional Laplace operator.

2.5. The fractional Robin boundary conditions. Let Ω ⊂ RN be a bounded

domain with Lipschitz continuous boundary ∂Ω and s ∈ (0, 1). Let γ ∈ L∞ (∂Ω)
satisfy
γ(x) ≥ γ0 for σ − a.e. x on ∂Ω, (2.23)
for some constant γ0 > 0. We consider the bilinear symmetric form ER with domain
D(ER ) = W s,2 (Ω) and defined for u, v ∈ D(ER ) by
(u(x) − u(y))(v(x) − v(y))
Z Z Z
CN,s
ER (u, v) = dxdy + γuvdσ.
2 Ω Ω |x − y|N +2s ∂Ω

We assume that s ∈ (0, 1) is such that W0s,2 (Ω) 6= W s,2 (Ω), that is, 1/2 < s < 1,
otherwise we are in the situation of the Dirichlet boundary condition. It follows
from [48, Theorem 6.4] that the form ER is closed on L2 (Ω).
Let AR be the closed linear selfadjoint operator associated with the form ER in
the sense that
(
D(AR ) := {u ∈ W s,2 (Ω), ∃ v ∈ L2 (Ω), ER (u, ϕ) = (v, ϕ)L2 (Ω) , ∀ ϕ ∈ W s,2 (Ω)}
AR u = v.
(2.24)
We call AR a realization of the regional fractional Laplace operator AsΩ on L2 (Ω)
with the fractional Robin type boundary conditions. The following result has been
proved in [48, Proposition 6.5] by using the integration by parts formula given in
Theorem 2.3.
Proposition 2.6. Let AR be the operator defined in (2.24). Assume also that Ω is
a bounded open set of class C 1,1 . Then
2 2
D(AR ) ∩ C2s (Ω) = {u ∈ C2s (Ω), BN,s N 2−2s u + γu = 0 on ∂Ω}, AR u = AsΩ u,
where BN,s is the constant given in (2.17).
We refer to [28, 29, 48] for more details. We notice that it also follows from
Proposition 2.5 that, as s ↑ 1, the operator AR converges (in some sense) to the
realization ∆R in L2 (Ω) of the Laplace operator with the classical Robin boundary
conditions.

2.6. Generation of semigroup. Let 0 < s < 1 and let AK , K ∈ {E, D, N , R}

be the operators introduced above. We also let WEs,2 (Ω) = W
f s,2 (Ω), W s,2 (Ω) =
0 D
s,2 s,2 s,2
W0 (Ω) and WK (Ω) = W (Ω) if K ∈ {N , R}. We introduce the following
assumption.
(H): If K = N or K = R, we assume that 21 < s < 1 and Ω ⊂ RN is a
bounded domain with Lipschitz continuous boundary. If K = E or K = D,
then Ω ⊂ RN is an arbitrary bounded open set.
Indeed recall that by the foregoing considerations, W0s,2 (Ω) = W s,2 (Ω) provided
that 0 < s ≤ 1/2. We have the following result.
Theorem 2.5. Let 0 < s < 1 and let assumption (H) be satisfied. Then the
following assertions hold.
1292 CIPRIAN G. GAL AND MAHAMADI WARMA

(a) The operator −AK generates a submarkovian semigroup (e−tAK )t≥0 on L2 (Ω)
and hence, can be extended to contraction strongly continuous semigroups on
Lp (Ω) for every p ∈ [1, ∞), and to a contraction semigroup on L∞ (Ω).
(b) The operator AK has a compact resolvent, and hence has a discrete spectrum.
The spectrum of AK is an increasing sequence of real numbers 0 ≤ λ1 < λ2 <
· · · < λn < . . . that converges to +∞. Moreover, 0 is an eigenvalue of AN and
is not an eigenvalue of AK for K ∈ {E, D, R}, and if un is an eigenfunction
associated with λn , then un ∈ D (AK ) ∩ L∞ (Ω).
(c) Denoting the generator of the semigroup on Lp (Ω) by Ap,K , so that AK =
A2,K , then the spectrum of A
p,K is independent of p for every p ∈ [1, ∞].
∞
(d) Let θ ∈ (0, 1]. Then D AθK embeds continuously  into
 L (Ω) provided that
N N 2


θ > 4s . Let p ∈ (2, ∞) and assume that θ > 4s 1 − p . Then also D AθK ⊂
Lp (Ω) continuously.
Proof. Let 0 < s < 1 and let AK , K ∈ {E, D, N , R} be the operators introduced
above. Assume the assumption (H).
(a) The proof of this part is contained in [48, Theorems 6.2 and 6.6]. We notice
that in [48] the operator AK for K ∈ {N , R} has been considered. The proof of the
corresponding result for AK , K ∈ {E, D} follows similarly.
(b) By [48, Theorems 6.2 and 6.6], the operator AK for K ∈ {N , R} has a
compact resolvent. We have shown above that the embedding W0s,2 (Ω) ,→ L2 (Ω)
is compact. Hence, the operator AK for K ∈ {E, D} also has a compact resolvent.
Since AK is a nonnegative self-adjoint operator and has a compact resolvent, then
it has a discrete spectrum which is an increasing sequence of real numbers 0 ≤
λ1 < λ2 < · · · < λn . . . , that converges to +∞. It is easy to see that 0 is an
eigenvalue of AN and is not an eigenvalue of AK for K ∈ {E, D, R}. Next, let
s,2
un ∈ WK (Ω) be an eigenfunction associated with λn . Then, AK un = λn un . Let
α > 0 be a real number. Since α ∈ ρ(−AK ), we have that αI + AK is invertible.
From AK un = λn un we have that
un = (αI + AK )−1 (λn + α)un = (λn + α)(αI + AK )−1 (un ).
By [48, Theorems 6.2 and 6.6], the semigroup (e−tAK )t≥0 , for K ∈ {N , R}, is
ultracontractive in the sense that it maps L2 (Ω) into L∞ (Ω). It also follows from
(2.3) that the semigroup (e−tAK )t≥0 , for K ∈ {E, D}, is ultracontractive. More
precisely, there is a constant C > 0 such that for every f ∈ Lp (Ω) and t > 0,
N
ke−tAK f kL∞ (Ω) ≤ Ct− 2sp kf kLp (Ω) , K ∈ {E, D, R}, (2.25)
and
N
ke−tAN f kL∞ (Ω) ≤ Ct− 2sp et kf kLp (Ω) . (2.26)
Since for every f ∈ L2 (Ω) and α > 0,
Z ∞
−1
(αI + AK ) f= e−αt e−tAK f dt,
0
N
it follows from (2.25) and (2.26) and the fact that un ∈ Lp (Ω) for some p > 2s that
there exists a constant M > 0 such that
kun kL∞ (Ω) ≤ M (λn + α)kun kLp (Ω) .
This completes the proof of part (b).
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1293

(c) Let p ∈ [1, ∞] and let Ap,K be the generator of the semigroup on Lp (Ω).
Since AK = A2,K has a compact resolvent and Ω is bounded, it follows from the
ultracontractivity that each semigroup has a compact resolvent on Lp (Ω) for p ∈
[1, ∞]. Now it follows from [15, Corollary 1.6.2] that the spectrum of Ap,K is
independent of p.
θ
(d) Since I + AK is invertible

we have that the L2 -norm of (I + AK ) defines an
θ 2
equivalent norm on D AK . Besides, for every f ∈ L (Ω),
Z ∞
−θ 1
(αI + AK ) f = tθ−1 e−αt e−tAK f dt, α > 0.
Γ (θ) 0

We shall prove the first claim in the case K ∈ {N } (the argument in the cases
−tAN
K ∈ {E, D, R} is similar).

Using (2.26) for t ∈ (0, 1) and the contractivity of e
θ
for t > 1, for u ∈ D AK , we deduce
Z 1 Z ∞
N
− 4s
kukL∞ (Ω) ≤ C kukD(Aθ ) t +θ−1
dt + C kukD(Aθ ) e−t dt.
K K
0 1

The first integral is finite if and only if θ > N/4s. For the second claim, we begin by
interpolating the inequality (2.26) with the L2 (Ω)-contractivity of e−tAN to obtain
that
≤ Ct− 4s (1− p ) et(1− p ) kf k 2
N 2 2
ke−tAN f k p
L (Ω) (2.27)
L (Ω)

2 θ
for every p ∈ (2, ∞). As above with α >

1−2/p ∈ (0, 1) , the L -norm of (αI + AK )
θ
defines an equivalent norm on D AK so
that (2.27) for t ∈ (0, 1) and the contrac-
tivity of e−tAN for t > 1, for u ∈ D AθK , allow us to deduce once again that
Z 1 Z ∞
kukLp (Ω) ≤ C kukD(Aθ ) t
N
− 4s (1− p2 )+θ−1 dt + C kukD(Aθ ) e−αt dt < ∞
K K
0 1

provided that the first integral is finite, i.e., θ > N 1 − 2p−1 / (4s). The proof in

the remaining cases K ∈ {E, D, R} is analogous and thus omitted.

We notice that the assumption 12 < s < 1 if K = N or K = R in (H) is not

a restriction, since, otherwise Dirichlet, fractional Neumann and Robin boundary
conditions coincide, that is, AN ≡ AR ≡ AD if 0 < s ≤ 21 .
We conclude the section by the following regularity result taken from [44, Propo-
sition 1.1].

Proposition 2.7. Let Ω ⊂ RN be a bounded domain with Lipschitz continuous

f s,2 (Ω) be a weak solution of the elliptic problem
boundary, f ∈ L∞ (Ω) and let u ∈ W 0

(−∆)s u = f in Ω. (2.28)

Then, u ∈ C 0,s (RN ).

Remark 4. We notice that even if (−∆)s and AsΩ are related by the relation
(1.4), due to the effect of the potential VΩ one cannot immediately deduce from
Proposition 2.7 a similar result for the elliptic problem associated with AsΩ . To
have such a result, one needs to give a complete proof.
1294 CIPRIAN G. GAL AND MAHAMADI WARMA

3. Well-posedness and regularity. Without loss of generality, we take d ≡ 1 in

this section. We consider all problems (1.5)-(1.7) in unified form

∂t u + AK u + f (u) = 0, in Ω × (0, ∞) ,
(3.1)
u (0) = u0 , in Ω,
where AK , K ∈ {D, N , R, E} , is the self-adjoint operator associated with the re-
gional fractional Laplacian, subject to Dirichlet, Neumann, Robin and the fractional
Laplace operator associated with Dirichlet (i.e., u = 0 on RN \Ω) boundary condi-
tions, respectively, as introduced in the previous section.
We recall that we have set WEs,2 (Ω) := W f s,2 (Ω), W s,2 (Ω) := W s,2 (Ω) and
0 D 0
s,2 −s,2
WK (Ω) := W s,2 (Ω) when K ∈ {N , R}. We also use the notation WK (Ω) :=
s,2 ∗ s,2
(WK (Ω)) , as the topological dual of WK (Ω) , 0 < s < 1. Furthermore, we en-
dow the domain D (AK ) of the operator AK with the graph norm kAK ukL2 (Ω) +
kukL2 (Ω) . Finally, the symbol h·, ·i stands for the duality pairing between any Ba-
nach space X and its dual X ∗ .
Our assumptions on the nonlinearity f that we will need in this section are as
follows.
1
(H1): When K ∈ {D, R, E} the function f ∈ Cloc (R) satisfies
f (τ )
lim inf > −λ∗ , (3.2)
|τ |→+∞ τ
for some constant λ∗ ∈ [0, CK,s ), where CK,s = C (K, Ω, s) > 0 is the best
Sobolev/Poincaré constant in
2 s,2
kϕkL2 (Ω) ≤ CK,s EK (ϕ, ϕ), for all ϕ ∈ WK (Ω) . (3.3)
When K = N , we assume the function fN (τ ) := f (τ ) − χτ, for some χ > 0,
also satisfies (3.2) with λ∗ ∈ [0, CN ,s ), where CN ,s = C(N, s, Ω) > 0 is such
that
 
2 2
kϕkL2 (Ω) ≤ CN ,s EN (ϕ, ϕ) + kϕkL1 , for all ϕ ∈ W s,2 (Ω) .
1
(H2): f ∈ Cloc (R) satisfies
bf |τ |p − cf ≤ f (τ ) τ ≤ C
C ef |τ |p + e
cf , for all τ ∈ R,
for some appropriate positive constants C
bf , cf , C
ef , e
cf and some p > 1.
(H3): f ∈ C 1 (R) satisfies
0
f (τ ) ≥ −Cf , for all τ ∈ R,
for some positive constant Cf .
Concerning regularity conditions for the domain Ω we assume the following.
(H4): Ω is an arbitrary bounded open set if K ∈ {D, E} and Ω is a bounded
domain with Lipschitz continuous boundary if K ∈ {N , R}.
In what follows we shall use classical (linear/nonlinear semigroup) definitions
of strong solutions to the unified problem (3.1). “Strong” solutions are defined
via nonlinear semigroup theory for bounded initial data and satisfy the differential
equations almost everywhere in t > 0. We first introduce the rigorous notion of
(global) weak solutions to the problem (3.1) as in the classical case for the semilinear
parabolic equation with ”Laplacian” diffusion. Throughout the remainder of this
article the solution of our system is a function that depends on both time and spatial
variables but in our proofs we sometime omit the dependence in x.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1295

Definition 3.1. Let u0 ∈ L2 (Ω) be given and assume (H2) holds for some p > 1.
The function u is said to be a weak solution of (3.1) if, for a.e. t ∈ (0, T ) , for any
T > 0, the following properties are valid:
• Regularity:
(
s,2
u ∈ L∞ (0, T ); L2 (Ω) ∩ Lp ((0, T ) × Ω) ∩ L2 ((0, T ); WK


(Ω)),
−s,2
0 (3.4)
∂t u ∈ L2 ((0, T ); WK (Ω)) + Lp ((0, T ) × Ω) ,
0
where p = p/ (p − 1) .
• Variational identity: for the weak solutions the following equality
h∂t u (t) , ξi + EK (u (t) , ξ) + hf (u (t)) , ξi = 0 (3.5)
s,2
holds for all ξ ∈ WK (Ω) ∩ Lp (Ω) , a.e. t ∈ (0, T ). Finally, we have, in the
2
space L (Ω) , u (0) = u0 almost everywhere.
• Energy identity: weak solutions satisfy the following identity
Z t
1 2
ku (t)kL2 (Ω) + [EK (u (τ ) , u (τ )) + hf (u (τ )) , u (τ )i] dτ
2 0
1 2
= ku (0)kL2 (Ω) . (3.6)
2


Remark 5. Note that by (3.4), u ∈ Cw [0, T ] ; L2 (Ω) , that is the space of all
L2 (Ω)-valued weakly continuous functions on the interval [0, T ]. Therefore the
initial value u (0) = u0 is meaningful when u0 ∈ L2 (Ω).
Finally, our notion of (global) strong solution is as follows.
Definition 3.2. Let u0 ∈ L∞ (Ω) be given. A weak solution u is “strong” if, in
addition, it fulfills the regularity properties:
1,∞
u ∈ Wloc ((0, T ]; L2 (Ω)) ∩ C ([0, T ] ; L∞ (Ω)) , (3.7)
such that u (t) ∈ D(AK ), a.e. t ∈ (0, T ) , for any T > 0.
This section consists of two main parts. At first we will establish the existence
and uniqueness of a (local) strong solution on a finite time interval using the theory
of monotone operators exploited and developed in [24]. Then exploiting a modified
Moser iteration argument we show that the strong solution is actually a global
solution. In the second part, we will show the existence of (globally-defined) weak
solutions which satisfy the energy identity (3.6) and the variational form (3.5). Then
combining the energy method with another refined iteration scheme we also show
that any weak solution with initial data in L2 (Ω) acquires additional smoothness
in an infinitesimal time.

3.1. Weak and strong solutions. We first prove a Poincaré-type inequality in

the space W s,2 (Ω), 0 < s < 1. Its application is crucial in the proof of strong
solutions to semilinear parabolic equations with fractional diffusion.
Lemma 3.3. Let 0 < s < 1 and let Ω ⊂ RN be as in (H4) with K ∈ {D, N , R, E}.
Then for all  ∈ (0, 1) there is ζ > 0 such that
s,2
kuk2L2 (Ω) ≤ EK (u, u) + −ζ kuk2L1 (Ω) , for all u ∈ WK (Ω).
1296 CIPRIAN G. GAL AND MAHAMADI WARMA

Proof. By a scaling argument it suffices to prove the inequality for kukL2 (Ω) = 1.
Suppose that there is no ζ > 0 such that the inequality holds for a given  ∈ (0, 1).
s,2
Then for any k ∈ N there is uk ∈ WK (Ω) such that
kuk k2L2 (Ω) = 1 > EK (uk , uk ) + −k kuk k2L1 (Ω) .
The foregoing inequality implies that the resulting sequence (uk ) is bounded in
s,2 s,2
WK (Ω). Since the identity operator is a compact map from WK (Ω) into L2 (Ω)
1
and into L (Ω), respectively, we find a subsequence, again denoted by (uk ), that
s,2
converges strongly in L2 (Ω) and in L1 (Ω) to some limit function u ∈ WK (Ω). By
assumption we have kukL2 (Ω) = 1. On the other hand, the inequality shows that
kuk k2L1 (Ω) ≤ k for all k, such that kukL1 (Ω) = 0 and thus u = 0 a.e. in Ω. This is
a contradiction which altogether completes the proof of the lemma.
s,2
We notice that it follows from Lemma 3.3 that for u ∈ WK (Ω),
1/2
(EK (u, u)) + kukL1 (Ω)
s,2
defines an equivalent norm on WK (Ω). In fact, it is clear that there exists a
1/2
constant C > 0 such that (EK (u, u)) + kukL1 (Ω) ≤ kukW s,2 (Ω) . Using Lemma 3.3
K
we get the converse inequality.
The next inequality is essential in comparing various energy forms.
Lemma 3.4. Let E be the energy given by
Z Z
E (u, v) = (u(x) − u(y))(v(x) − v(y))K (x, y) dxdy,
Ω Ω

for some positive kernel K : Ω × Ω → R+ . Then

4p p−1 p−1
p−1
2 E(|u| u, |u| u) ≤ E(u, |u| u), (3.8)
2 2

(p + 1)
for all functions u for which the terms in (3.8) make sense and all p > 1.
Proof. We prove the inequality by elementary analysis. Define the function g :
R × R → R by
  4p  p−1 p−1
2
p−1 p−1
g (z, t) = (z − t) |z| z − |t| t − 2 |z| 2
z − |t| 2
t
(p + 1)
Using the definition of E, we first notice that (3.8) is equivalent to showing that
g(z, t) ≥ 0 for all (z, t) ∈ R2 . (3.9)
Indeed, assume this were true so that for u : Ω → R there holds
1 
p−1 p−1

(u (x) − u (y)) |u (x)| u (x) − |u (y)| u (y)
p
4  p−1 p−1
2
≥ 2 |u (x)| 2
u(x) − |u (y)| 2
u(y) . (3.10)
(p + 1)
Then (3.8) is an immediate consequence of (3.10). We now prove our claim. First,
we observe that
g(z, t) = g(t, z), g(z, 0) ≥ 0, g(0, t) ≥ 0 and g(z, t) = g(−z, −t).
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1297

Therefore, without any restriction, we may assume that z ≥ t. A simple calculation

shows that
2 h p−1 p−1
i Z z p−1
|z| 2
z − |t| 2
t = |τ | 2 dτ.
p+1 t
Since the function ϕ : R → R given by ϕ(τ ) = |τ |p (p ≥ 2) is convex, then using
the well-known Jensen inequality, it follows that
4p  p−1 p−1
2 
 2 h p−1 p−1
i2
2
z − |t| 2
t = p  |z| 2
z − |t| 2
t 
  
(p + 1)2 p+1
|z|

Z z 2
 p−1 
= p  |τ | 2 dτ 
t
Z z 2
p−1 dτ 
= p(z − t)2 

|τ | 2
z − t
Z tz
dτ
≤ p(z − t)2 |τ |p−1
z −t
Z zt
= p(z − t) |τ |p−1 dτ
t
 
p−1 p−1
= (z − t) |z| z − |t| t .

We have shown the claim (3.9) and this completes the proof of lemma.

We will now state a well-known result for the non-homogeneous Cauchy problem
 0
u (t) + A (u) 3 g (t) , t ∈ [0, T ] ,
(3.11)
u|t=0 = u0 .
Theorem 3.5. ([43, Chapter IV, Theorem 4.3]) Let ϕ : H → (−∞, +∞] be a
proper, convex, and lower-semicontinuous functional on the Hilbert space H and
set A = ∂ϕ, where ∂ϕ denotes the subdifferential of the functional ϕ. Let u be
the generalized solution
√ 0 of (3.11) with g ∈ L2 (0, T ; H) and u0 ∈ D (A). Then
1 2
ϕ (u) ∈ L (0, T ) , tu (t) ∈ L ((0, T ); H) and u (t) ∈ D (A) for a.e. t ∈ [0, T ] .
The second one is a more general version of [43, Chapter IV, Proposition 3.2]
and was proved in [24, Theorem 6.3 and Corollary 6.4].
Theorem 3.6. Let the assumptions of Theorem 3.5 be satisfied. Assume that A =
∂ϕ is strongly accretive in H, that is, A − ωI is accretive for some ω > 0 and, in
addition,
g ∈ L∞ ([δ, ∞); H) ∩ W 1,2 ([δ, ∞); H) ,
for every δ > 0. Let u be the unique generalized solution of (3.11) for u0 ∈ D (A).
It follows that
u ∈ L∞ ([δ, ∞); D (A)) ∩ W 1,∞ ([δ, ∞); H) .
Now we state the first main theorem of this section.
Theorem 3.7. Assume that the nonlinearity f obeys (H1) and assume Ω satisfies
(H4). For every u0 ∈ L∞ (Ω) , there exists a unique strong solution of (3.1) in the
sense of Definition 3.2.
1298 CIPRIAN G. GAL AND MAHAMADI WARMA

Proof. Step 1. (Local existence). Fix K ∈ {D, N , R, E}. Let u0 ∈ L∞ (Ω) ⊂

L2 (Ω)
L2 (Ω) = D(AK ) . Let ϕK be the functional on L2 (Ω) with domain D(ϕK ) =
s,2 s,2
WK (Ω) and defined by ϕK (u) = 12 EK (u, u) for u ∈ WK (Ω). From Theorem 2.5
we know that −AK = −∂ϕK generates a strongly continuous (linear) semigroup
(e−tAK )t≥0 of contraction operators on L2 (Ω). Finally, e−tAK is non-expansive on
L∞ (Ω) , that is,
−tA
u0 L∞ (Ω) ≤ ku0 kL∞ (Ω) , t ≥ 0 and u0 ∈ L∞ (Ω) ,

e K
(3.12)

and AK is strongly accretive only in the case K ∈ {D, R, E} whereas AN is only

accretive on L2 (Ω) (this follows from the definition of AN and from Theorem 2.5).
Thus, the operator version of the original problem (3.1) reads
∂t u = −AK u − fK (u) , K ∈ {D, E, R} (3.13)
and
∂t u = −AN ,χ u − fN (u) , with AN ,χ = AN + χI, (3.14)
where we have set fK (τ ) = f (τ ) , K ∈ {D, E, R} and fN (τ ) = f (τ ) − χτ . Clearly,
AN ,χ is also strongly accretive on L2 (Ω) by construction and satisfies (3.12).
We adapt an argument we have developed in [24, Theorem 3.4]. We prove the
existence of a (locally-defined) strong solution to (3.13), (3.14) by a fixed point
argument. We shall focus on the case K ∈ {D, R, E} as the case K = N is similar.
To this end, fix 0 < T ∗ ≤ T , consider the space
n o
XT ∗ ,R∗ ≡ u ∈ C ([0, T ∗ ] ; L∞ (Ω)) : ku (t)kL∞ (Ω) ≤ R∗

and define the following mapping

Z t
−tAK
Σ (u) (t) = e u0 − e−(t−τ )AK f (u (τ )) dτ, t ∈ [0, T ∗ ] . (3.15)
0

Note that XT ∗ ,R∗ , when endowed with the norm of C ([0, T ∗ ] ; L∞ (Ω)) , is a closed
subset of C ([0, T ∗ ] ; L∞ (Ω)) , and since f is continuously differentiable, Σ (u) (t)
is continuous on [0, T ∗ ]. We will show that, by properly choosing T ∗ , R∗ > 0,
Σ : XT ∗ ,R∗ → XT ∗ ,R∗ is a contraction mapping with respect to the metric induced
by the norm of C ([0, T ∗ ] ; L∞ (Ω)) . The appropriate choices for T ∗ , R∗ > 0 will be
specified below. First, we show that u ∈ XT ∗ ,R∗ implies that Σ (u) ∈ XT ∗ ,R∗ , that
1
is, Σ maps XT ∗ ,R∗ to itself. From (3.12) and the fact that f ∈ Cloc (R), we observe
that the mapping Σ satisfies the following estimate
kΣ (u (t))kL∞ (Ω)
Z t
−(t−τ )AK
≤ ku0 kL∞ (Ω) + (f (0) + (f (u (τ )) − f (0))) ∞ dτ

e
0 L (Ω)
 
∗ ∗
≤ ku0 kL∞ (Ω) + t |f (0)| + Qf 0 (R ) R ,

for some positive continuous function Qf 0 which depends only on the size of the
0
nonlinearity f . Thus, provided that we set R∗ ≥ 2 ku0 kL∞ (Ω) , we can find a
sufficiently small time T ∗ > 0 such that
 
2T ∗ |f (0)| + Qf 0 (R∗ ) R∗ ≤ R∗ , (3.16)
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1299

in which case Σ (u (t)) ∈ XT ∗ ,R∗ , for any u (t) ∈ XT ∗ ,R∗ . Next, we show that by
possibly choosing T ∗ > 0 smaller, Σ : XT ∗ ,R∗ → XT ∗ ,R∗ is also a contraction.
Indeed, for any u1 , u2 ∈ XT ∗ ,R∗ , exploiting again (3.12), we estimate
kΣ (u1 (t)) − Σ (u2 (t))k∞,Ω
Z t
≤ Qf 0 (R∗ )
−(t−τ )AK
(u1 (τ ) − u2 ) (τ ) dτ

e
0 ∞,Ω
≤ tQf 0 (R∗ ) ku1 − u2 kC([0,T ∗ ];L∞ (Ω)) . (3.17)
∗
This shows that Σ is a contraction on XT ∗ ,R∗ provided that T < 1 is much smaller
than the one determined by (3.16) and T ∗ Qf 0 (R∗ ) < 1. Therefore, owing to the
contraction mapping principle, we conclude that problem (3.13) has a unique local
solution u ∈ XT ∗ ,R∗ . This solution can certainly be (uniquely) extended on a right
maximal time interval [0, Tmax ), with Tmax > 0 depending on ku0 kL∞ (Ω) , such
that, either Tmax = ∞ or Tmax < ∞, in which case limt%Tmax ku (t)kL∞ (Ω) = ∞.
Indeed, if Tmax < ∞ and the latter condition does not hold, we can find a sequence
tn % Tmax such that ku (tn )kL∞ (Ω) ≤ C. This would allow us to extend u as a
solution to Equation (3.13) to an interval [0, tn + δ), for some δ > 0 independent
of n. Hence u can be extended beyond Tmax which contradicts the construction of
Tmax > 0. To conclude that the solution u belongs to the class in Definition 3.2, let
us further set G (t) := −f (u (t)) , for u ∈ C ([0, Tmax ) ; L∞ (Ω)) and notice that u is
the “generalized” solution of
∂t u + AK u = G (t) , t ∈ [0, Tmax ), (3.18)
such that u (0) = u0 ∈ L∞ (Ω) ⊂ L2 (Ω) = D(AK ). By Theorem 3.5, the
“gener-
alized” solution u has the additional regularity ∂t u ∈ L2 (δ, Tmax ); L2 (Ω) , which
1
together with the facts that u is continuous on [0, Tmax ) and f ∈ Cloc (R), yield
1,2 2 ∞ 2



G∈W (δ, Tmax ); L (Ω) ∩ L (δ, Tmax ); L (Ω) . (3.19)
Thus, we can apply Theorem 3.6 to deduce that
u ∈ L∞ ([δ, Tmax ); D (AK )) ∩ W 1,∞ [δ, Tmax ); L2 (Ω) ,


(3.20)
such that the solution u is Lipschitz continuous on [δ, Tmax ), for every δ > 0. Thus,
we have obtained a locally-defined strong solution in the sense of Definition 3.2. As
to the variational equality in Definition 3.1, we note that this equality is satisfied
even pointwise (in time t ∈ (0, Tmax )) by the strong solutions. Our final point is
to show that Tmax = ∞, because of condition (H1). This ensures that the strong
solution constructed above is also global.
Step 2. (Energy estimate) Let m ≥ 1 and consider the function Em : (0, ∞) →
[0, ∞) defined by Em (t) := ku(t)km+1Lm+1 (Ω) . First, notice that Em is well-defined on
(0, Tmax ) because u is bounded in Ω × (0, Tmax ) and because Ω has finite measure.
Since u is a strong solution on (0, Tmax ) , see Definition 3.2 (or (3.20)), recall that
u is continuous from [0, Tmax ) → L∞ (Ω) and Lipschitz continuous on [δ, Tmax ) for
every δ > 0. Thus, u (as function of t) is differentiable a.e., whence, the function
Em (t) is also differentiable for a.e. t ∈ (0, Tmax ) .
For strong solutions and t ∈ (0, Tmax ), integration by parts procedure yields the
following standard energy identity:
Z
1 d
E1 (t) + EK (u (t) , u (t)) + f (u (t)) u (t) dx = 0. (3.21)
2 dt Ω
1300 CIPRIAN G. GAL AND MAHAMADI WARMA

Assumption (H1) in the case K ∈ {D, R, E} implies that

f (τ ) τ ≥ −λ∗ τ 2 − Cf , (3.22)
for some Cf > 0 and for all τ ∈ R. This inequality allows us to estimate the
s,2
nonlinear term in (3.21). We have (by using an equivalent norm in WK (Ω)) that
1 d 2
E1 (t) + CK,s ku (t)kW s,2 (Ω) ≤ Cf |Ω| + λ∗ E1 (t) , (3.23)
2 dt K

where |Ω| denotes the N -dimensional Lebesgue measure of Ω. In view of (3.3) and
Gronwall’s inequality, (3.23) gives the following estimate for t ∈ (0, Tmax ) ,
Z t
2 2
ku (t)kL2 (Ω) + 2 (CK,s − λ∗ ) ku (τ )kW s,2 (Ω) dτ (3.24)
0
2
≤ ku0 kL2 (Ω) e−ρt + C (f, |Ω|) ,
for some constants ρ = ρ (N, Ω) > 0, C (f, |Ω|) > 0. The proof of the energy
inequality in the case K = N is analogous (in this case, fN obeys (3.22)).
Step 3. (The iteration argument). In this step, c > 0 will denote a constant that
is independent of t, Tmax , m, k and initial data, which only depends on the other
structural parameters of the problem. Such a constant may vary even from line to
line. Moreover, we shall denote by Qτ (m) a monotone nondecreasing function in
m of order τ, for some nonnegative constant τ, independent of m. More precisely,
Qτ (m) ∼ cmτ as m → +∞. We begin by showing that Em (t) satisfies a local
recursive relation which can be used to perform an iterative argument. Testing the
m−1
variational equation (3.5) for the strong solution with |u| u, m ≥ 1, gives on
account of (3.22) and Lemma 3.4 the following inequality:
d 4m m−1 m−1
Em (t) + EK (|u (t)| 2 u, |u (t)| 2 u) (3.25)
dt m+1
 m

≤Q1 (m + 1) Em (t) + (Em (t)) m+1 ,

in all cases K ∈ {D, N , R, E}. Next, set mk + 1 = 2k , k ∈ N, and define

Z
2k
Mk := sup |u (t, x)| dx = sup Emk (t) . (3.26)
t∈(0,Tmax ) Ω t∈(0,Tmax )

Our goal is to derive a recursive inequality for Mk using (3.25). In order to do so,
for q > 1 fixed that we will choose below, we define
mk − mk−1 1 q−1
pk := = < 1, q k := 1 − pk = 2 .
q (1 + mk ) − (1 + mk−1 ) 2q − 1 2q − 1
We aim to estimate the terms on the right-hand side of (3.25) in terms of the
L1+mk−1 (Ω)-norm of u. First, the Hölder inequality and the Sobolev inequality
s,2
(i.e., WK (Ω) ⊂ L2q (Ω), with q = q (N, s) ∈ (1, N/ (N − 2s)], if N > 2s and
q ∈ (1, ∞) if N = 2s, see (2.3) and (2.1)) yield
Z Z pk Z q k
1+mk (1+mk )q 1+mk−1
|u| dx ≤ |u| dx |u| dx (3.27)
Ω Ω Ω
isk Z q k
h mk −1 mk −1
1+mk−1
≤ c EK (|u| 2
u, |u| 2
u) |u| dx ,
Ω
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1301

with sk = pk q ≡ q/ (2q − 1) ∈ (0, 1). Applying Young’s inequality on the right-hand

side of (3.27), we get for every  > 0,
Z mk −1 mk −1
1+mk
Q1 (mk + 1) |u| dx ≤EK (|u| 2 u, |u| 2 u)
Ω
Z 2
1+mk−1
+ Qα (mk + 1) |u| dx , (3.28)
Ω

for some α > 0 independent of k, since zk := q k / (1 − sk ) ≡ 2. In order to esti-

mate the last term on the right-hand side of (3.25), we define two decreasing and
increasing sequences (lk )k∈N and (wk )k∈N , respectively, such that
mk + 1 q k mk
lk := and wk := ,
sk mk mk (1 − sk ) + 1
and observe that they satisfy
 
1 2 (q − 1)
1 < lk ≤ 2 2 − , ≤ wk ≤ 2
q 3q − 2
for all k ∈ N (in particular, wk → 2 as k → ∞). The application of Young’s
inequality in (3.28) yields again
Z  mmk+1
k mk −1 mk −1
1+mk
Q1 (mk + 1) |u| dx ≤EK (|u| 2 u, |u| 2 u) (3.29)
Ω
Z wk
1+mk−1
+ Qβk (mk + 1) |u| dx ,
Ω
for every  > 0, where now
c βk
Qβk (mk + 1) ∼ (1 + mk )
1/(lk −1)
with
mk + 1 2q − 1
βk := → , as k → ∞.
mk (1 − sk ) + 1 q−1
Hence, inserting (3.28), (3.29) into inequality (3.25), choosing a sufficiently small
0 <  ≤ 0 := 41 , and simplifying, we obtain for t ∈ (0, Tmax ) ,
Z
d 1+mk 0 mk −1 mk −1
|u (t, x)| dx + EK (|u (t)| 2 u, |u (t)| 2 u) (3.30)
dt Ω 2
Z 2
1+mk−1
≤ Qδ (mk + 1) |u| dx ,
Ω
for some positive constant δ > 0 independent of k.
s,2
Next, since u (t) ∈ WK (Ω) ∩ L∞ (Ω) , for a.e. t ∈ (0, Tmax ), it follows from [48,
1+mk
s,2
Remark 2.5] that |u (t)| 2
∈ WK (Ω). Thus, we can apply Lemma 3.3 to infer
that
Z Z 2
mk −1 mk −1
dx − −ζ
mk +1 1+mk−1
0 EK (|u| 2
u, |u| 2
u) ≥ |u| 0 |u| dx , (3.31)
Ω Ω

for some ζ > 0 independent of u, k. We can now combine (3.31) with (3.30) to
deduce Z Z
d 2k 1 2k
|u (t, x)| dx ≤ Qδ 2k Mk−1

2
|u (t, x)| dx + , (3.32)
dt Ω 2 Ω
1302 CIPRIAN G. GAL AND MAHAMADI WARMA

for t ∈ (0, Tmax ) . Integrating (3.32) over (0, t), we infer from Gronwall-Bernoulli’s
inequality [13, Lemma 1.2.4] that there exists yet another constant c > 0, indepen-
dent of k, such that
Z 
2k kδ 2
Mk ≤ max |u0 | dx, c2 Mk−1 , for all k ≥ 2. (3.33)
Ω
On the other hand, let us observe that there exists a positive constant C∞ =
C∞ (ku0 kL∞ (Ω) ) ≥ 1, independent of k, such that ku0 kL2k (Ω) ≤ C∞ . Taking the
2k -th root on both sides of (3.33), and defining Xk := supt∈(0,Tmax ) ku (t)kL2k (Ω) ,
we easily arrive at
n
1 o
Xk ≤ max C∞ , c2δk 2k Xk−1 , for all k ≥ 2. (3.34)

By straightforward induction in (3.34) (see [3, Lemma 3.2]; cf. also [13, Lemma
9.3.1]), we finally obtain the estimate
( )
sup ku (t)kL∞ (Ω) ≤ lim Xk ≤ c max C∞ , sup ku (t)kL2 (Ω) . (3.35)
t∈(0,Tmax ) k→+∞ t∈(0,Tmax )

It remains to notice that (3.35) together with the bound (3.24) shows that the norm
ku (t)kL∞ (Ω) is uniformly bounded for all times t > 0 with a bound, independent
of Tmax , depending only on ku0 kL∞ (Ω) , the “size” of the domain and the non-linear
function f. This gives Tmax = +∞ so that strong solutions are in fact global. This
completes the proof of the theorem.
Remark 6. Strong solutions to the system (3.1) exhibit an improved regularity in
time, we have
s,2
u ∈ C ([0, T ] ; L∞ (Ω)) ∩ C((δ, T ]; WK (Ω)), (3.36)
for any T > δ > 0. This follows from the fact that the nonlinear function f is con-
tinuously differentiable. Note that the second regularity in (3.36) is a consequence
of the first one, the time regularity in (3.7) (see Definition 3.2) and the variational
identity (3.5).
The following result is immediate.
Corollary 3.1. Suppose that the assumptions of Theorem 3.7 are satisfied and
let K ∈ {D, N , R, E}. The reaction-diffusion system (3.1) defines a (nonlinear)
continuous semigroup
TK (t) : L∞ (Ω) → L∞ (Ω) ,
given by
TK (t) u0 = u (t) , (3.37)
where u is the (unique) strong solution in the sense of Definition 3.2.
In the final part of this section, we aim to prove the existence of weak solutions
in the sense of Definition 3.1.
Theorem 3.8. Assume that the nonlinearity f obeys (H2), (H3) and Ω satisfies
(H4). Then, for any initial data u0 ∈ L2 (Ω) , there exists a unique (globally-
defined) weak solution
u ∈ C [0, T ] ; L2 (Ω)

in the sense of Definition 3.1.

 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1303

Proof. We divide the proof into three steps. For practical purposes, C will denote
a positive constant that is independent of time, T ,  > 0 and initial data, but which
only depends on the other structural parameters. Such a constant may vary even
from line to line.
Step 1. (Approximation scheme). First, we consider a sequence u0 ∈ L∞ (Ω)∩
s,2
WK (Ω) such that u0 → u0 = u (0) in L2 (Ω) . Next, for each K ∈ {D, N , R, E}
let u (t) be a strong solution, in the sense of Definition 3.2, of the system

∂t u + AK u + f (u ) = 0, in Ω × (0, ∞) ,
(3.38)
u (0) = u0 , in Ω.
Note that such a smooth solution exists since every function that satisfies (H2)
also obeys (3.22). Testing the weak formulation associated with problem (3.38), cf.
(3.5), with ξ = u (t) we find
Z
1 d 2
ku (t)kL2 (Ω) + EK (u (t) , u (t)) + f (u (t)) u (t) dx = 0,
2 dt Ω
for all t ∈ (0, T ). Invoking assumption (H2), we infer
d 2 bf ku (t)kp p
ku (t)kL2 (Ω) + 2EK (u (t) , u (t)) + 2C L (Ω) ≤ C |Ω| . (3.39)
dt
Integrate the foregoing inequality over (0, T ) we deduce
Z t 
2 bf ku (τ )kp p
ku (t)kL2 (Ω) + 2EK (u (τ ) , u (τ )) + 2C L (Ω) dτ
0
2
≤ ku (0)kL2 (Ω) e−ρt + C (3.40)
for all t ∈ (0, T ) , for some ρ > 0 independent of  > 0. As usual, on account of
(3.40), we deduce the following uniform (in  > 0) bounds
u ∈ L∞ (0, T ); L2 (Ω) ∩ Lp ((0, T ); Lp (Ω)) ,


(3.41)
s,2
u ∈ L2 ((0, T ); WK (Ω)),
for any T > 0. Hence, by (3.41) we also get
0 0
−s,2
AK u ∈ L2 ((0, T ); WK (Ω)), f (u ) ∈ Lp ((0, T ); Lp (Ω)), (3.42)
uniformly in  > 0. Here, recall that AK is the nonnegative self-adjoint operator
associated with the bilinear form EK . Comparison in (3.5) then gives
0 0
−s,2
∂t u ∈ L2 ((0, T ); WK (Ω)) + Lp ((0, T ); Lp (Ω)) (3.43)
uniformly in  > 0.
Step 2. (Passage to limit). From the above properties (3.41)-(3.43), we see that
as  → 0+ ,
u → u weakly star in L∞ (0, T ); L2 (Ω) ,

u → u weakly in Lp ((0, T ); Lp (Ω)) , (3.44)

0 0
−s,2
∂t u → ∂t u weakly in L2 ((0, T ); WK (Ω)) + Lp ((0, T ); Lp (Ω)),
s,2
along a subsequence. Since the continuous embedding WK (Ω) ⊂ L2 (Ω) is com-
pact, then we can exploit standard embedding results for vector valued functions
(see, e.g., [12]), to deduce
u → u strongly in L2 (0, T ); L2 (Ω) .


(3.45)
1304 CIPRIAN G. GAL AND MAHAMADI WARMA

By refining in (3.45), u converges to u a.e. in Ω × (0, T ). Then, by means of

known results in measure theory [12], the continuity of f and the convergence of
0
(3.45) imply that f (u ) converges weakly to f (u) in Lp ((0, T ) × Ω), while from
(3.41)-(3.42) and the linearity of AK , we further see that
−s,2
AK u → AK u weakly in L2 ((0, T ); WK (Ω)). (3.46)
We can now pass to the limit as  → 0 in the weak form (3.5) for u to deduce the
desired weak solution u, satisfying the variational identity (3.5) and the regularity
properties (3.4). The energy identity (3.6) is an immediate consequence of [12,
Theorem II.1.8] and the simple observation by which the distributional derivative
0 0
−s,2
∂t u (t) from D ([0, T ] ; WK (Ω) + Lp (Ω)) can be represented as ∂t u (t) = Z1 (t) +
Z2 (t) , with
−s,2
Z1 (t) := − AK u (t) ∈ L2 ((0, T ); WK (Ω)),
0 0
Z2 (t) := − f (u (t)) ∈ Lp ((0, T ); Lp (Ω)).
s,2
These spaces are precisely the dual of the space L2 ((0, T ); WK (Ω)), and the space
p p
L ((0, T ); L (Ω)), respectively. In particular, we obtain that every weak solution

2
u ∈ C [0, T ] ; L2 (Ω) , and that the map t 7→ ku (t)k2,Ω is absolutely continuous on
[0, T ], such that u satisfies the energy identity (3.6).
Step 2. (Uniqueness and continuous dependence). As usual, consider any
two weak solutions u1 , u2 , and set u (t) = u1 (t) − u2 (t). According to the energy
identity (3.6) and assumption (H3), we obtain
d 2 2
ku (t)kL2 (Ω) + 2EK (u (t) , u (t)) ≤ 2Cf ku (t)kL2 (Ω) . (3.47)
dt
Upon integration over (0, t), we infer
2 2
ku1 (t) − u2 (t)kL2 (Ω) ≤ CeCt ku1 (0) − u2 (0)kL2 (Ω) . (3.48)
This yields the desired continuous dependance result with respect to the initial data.
The proof of the theorem is finished.
Consequently, problem (3.1) defines a dynamical system in the classical sense.
Corollary 3.2. Let the assumptions of Theorem 3.8 be satisfied. The semilin-
ear parabolic equation (3.1) defines a (nonlinear) continuous semigroup SK (t) :
L2 (Ω) → L2 (Ω), K ∈ {D, N , R, E}, given by
SK (t) u0 = u (t) , (3.49)
where u is the (unique) weak solution in the sense of Definition 3.1.
3.2. Regularity of weak solutions. The main result of this section is concerned
with proving that any weak solution with initial condition in L2 (Ω) acquires ad-
ditional smoothness in an infinitesimal time; more precisely, it becomes a strong
solution in the sense of Definition 3.2. Moreover, the same result also establishes the
s,2
existence of an absorbing ball for the semigroup SK in the space WK (Ω) ∩ L∞ (Ω).
The latter is an essential property in the theory of attractors (see the next section).
Fix now K ∈ {D, N , R, E} .
Theorem 3.9. Let the assumptions of Theorem 3.8 be satisfied. Then, for u0 ∈
L2 (Ω) any orbit u (t) = SK (t) u0 of (3.1) satisfies
s,2
u ∈ L∞ ((ρ, ∞); WK (Ω) ∩ L∞ (Ω)) ∩ W 1,2 ((0, ∞); L2 (Ω)),
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1305

for every ρ > 0, and the following estimate holds:

 Z t 
2 2
sup ku (t)kW s,2 (Ω)+L∞ (Ω) + k∂t u (τ )kL2 (Ω) dτ ≤ Cδ , (3.50)
K
t≥ρ 0

for some constant Cρ = C (ρ) > 0, independent of t and initial data. Moreover,
u (t) ∈ D (AK ) for a.e. t > ρ. The constant Cρ in (3.50) is uniformly bounded in ρ
if ρ ≥ 1.

Proof. Step 1. (The bound in L∞ (Ω)). In this case, as in the proof of Theorem
3.8, we can use strong solutions in order to provide sufficient regularity to justify
all the calculations performed in the proof below. At the very end one can pass to
the limit and obtain the estimate even for the weak solutions.
0 0
Let now τ > τ > 0 and fix µ := τ − τ . We claim that there exists a positive
constant C = C (µ) ∼ µ−η (for some η > 0), independent of t and the initial data,
such that
sup ku (t)kL∞ (Ω) ≤ C sup ku (σ)kL2 (Ω) . (3.51)
t≥τ 0 σ≥τ

The argument leading to (3.51) follows exactly as in [21, Theorem 2.3] (cf. also
[22, 24]). It is based on the following recursive inequality for Emk (t), which is a
consequence of (3.32) and (3.27)-(3.30):
 2
k l


sup Emk (t) ≤ C 2 sup Emk−1 (σ) , for all k ≥ 1, (3.52)
t≥tk−1 σ≥tk
0
where the sequence {tk }k∈N is defined recursively tk = tk−1 − µ/2k , k ≥ 1, t0 = τ .
Here we recall that C = C (µ) > 0, l > 0 are independent of k and that C (µ) is
uniformly bounded in µ if µ ≥ 1 (see [21, Theorem 2.3]). We can iterate in (3.52)
with respect to k ≥ 1 and obtain that

l  
l  2  2k k
l
sup Emk (t) ≤ C 2k C 2k−1 · ... · C (2) (sup ku (σ)kL2 (Ω) )2
t≥tk−1 σ≥τ
k Pk k Pk
≤ C (2 ) 2(l2 ) (sup ku (σ)k
1 i
2k
L2 (Ω) ) . (3.53)
i=1 2i i=1 2i

σ≥τ

Therefore, we can take the P 2k -th root on both sides

P∞ ofi (3.53) and let k → +∞.
∞ 1
Using the facts that ζ := i=1 2i < ∞, ζ := i=1 2i < ∞, we easily deduce
(3.51). From the inequality (3.51) together with the energy estimate (3.40), which
is also satisfied by the weak solution u (t) with initial datum u0 ∈ L2 (Ω), we deduce
sup ku (t)kL∞ (Ω) ≤ Cδ , (3.54)
t≥ρ

with Cρ ∼ ρ−η , for some η > 0, for each ρ > 0. This yields the first part in (3.50).
s,2
Step 2. (The bound in WK (Ω)). The argument relies on using the test function
ξ = ∂t u (t) into the variational equation (3.5). However, in order to further justify
this choice in (3.5) we actually need to require more regularity of the strong solution,
in particular we need to have
1,q s,2
u ∈ Wloc ((0, ∞); WK (Ω)), for some q > 1. (3.55)
Due to the non-smooth nature of the domain Ω and its boundary ∂Ω, one gener-
ally lacks any further information on both weak and strong solutions than the one
provided by Definitions 3.1 and 3.2. In order to overcome this difficulty, we need
to further truncate the strong solutions resulting in approximate solutions which
1306 CIPRIAN G. GAL AND MAHAMADI WARMA

will now have the desired regularity (3.55). The latter is provided by a proper basis
associated with the nonnegative self-adjoint operator AK on L2 (Ω). Indeed,  as a
basis for L2 (Ω) we can choose the complete system of eigenfunctions ξiK i∈N for
AK in L2 (Ω) with
ξiK ∈ D (AK ) ∩ L∞ (Ω) ,
which is a key regularity provided by the statement of Theorem 2.5. According
to the general spectral theory, the eigenvalues λi can be increasingly ordered and
counted according to their multiplicities in order to form a real divergent sequence.
Moreover, the respective eigenvectors ξiK turn out to form an orthogonal basis
s,2
in WK (Ω) and L2 (Ω) , respectively. The eigenvectors ξiK may be assumed to
s,2
be normalized in L2 (Ω). Let Pn : WK (Ω) → span {ξ1 , ξ2 , ..., ξn } be the usual
orthogonal projector and consider a Galerkin truncation of u in the form
Xn
u,n (t) = ψi (t) ξiK , ψi ∈ C 1 (0, T ) , (3.56)
i=1

solving the problem ∂t u,n = −Pn (AK uε,n + f (u,n )) such that u,n (0) = Pn u (0).
We recall that the latter problem is uniquely solvable by the Cauchy-Lipschitz
theorem since f ∈ C 1 , and that the solution u,n has the desired regularity (3.55).
Thus, they key choice of the test function ξ = ∂t un, into the variational formulation
(3.5) is now allowed for these truncated solutions u,n . We infer
 
d 1 2
EK (u,n (t) , u,n (t)) + (F (u,n (t)) , 1)L2 (Ω) + k∂t u,n (t)kL2 (Ω) = 0, (3.57)
dt 2
Rσ
for all t ≥ 0. As usual, F denotes the primitive of f , i.e., F (σ) = 0 f (y) dy.
Multiply the foregoing equation by t ≥ ρ > 0 and integrate over (0, t) to get
  Z t
1 2
t EK (u,n (t) , u,n (t)) + (F (u,n (t)) , 1)L2 (Ω) + τ k∂t u,n (τ )kL2 (Ω) dτ
2 0
Z t 
1
= EK (u,n (t) , u,n (t)) + (F (u,n (τ )) , 1)L2 (Ω) dτ,
0 2
for all t ≥ ρ. Recalling that, due to (H2)-(H3), F is bounded from below, indepen-
p
dently of n, and |F (σ)| ≤ C (1 + |σ| ), we infer from (3.40) (which is also satisfied
by u,n ) and (3.54),
Z t  
2 1
EK (u,n (t) , u,n (t)) + k∂t u,n (τ )kL2 (Ω) dτ ≤ c 1 + , (3.58)
0 t
for some constant c > 0 independent of t, n, . On the basis of standard lower-
semicontinuity arguments, we can now pass to the limit, first with respect to n → ∞
and then as  → 0+ , to obtain the desired inequality (3.50), owing once more to
estimates (3.40)-(3.54) and uniqueness (cf. Theorem 3.8). The proof is finished.

4. Finite dimensional attractors. The first main result of this section is the
following. As before, we fix K ∈ {D, N , R, E} .
Theorem 4.1. Let the assumptions of Theorems 3.8 be satisfied for some f ∈
C 2 (R). There exists a compact attractor AK b L2 (Ω) for the parabolic system
(3.1) which attracts the bounded sets of L2 (Ω). Moreover, AK is the maximal
bounded attractor
in D (AK ) ∩ L∞ (Ω) and has finite fractal dimension, that is,
dimF AK , L2 (Ω) < ∞.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1307

Proof. Step 1. (Global attractor). By the proof of Theorem 3.8, (3.40), there
is a ball BK in L2 (Ω) which is absorbing in L2 (Ω) , meaning that for any bounded
set U ⊂ L2 (Ω) there exists t0 = t0 (kU kL2 (Ω) ) > 0 such that SK (t)U ⊂ BK for all
t ≥ t0 . Moreover, by Theorem 3.9, (3.50) and (3.54), we infer the existence of a
new time t1 ≥ 1 such that
 
sup ku (t)kW s,2 (Ω) + ku (t)kL∞ (Ω) ≤ C, (4.1)
K
t≥t1

for some positive constant C independent of time and the initial data. We also
observe that the Galerkin truncated solutions u,n satisfy (in the weak sense of
Definition 3.1) the following ”time-differentiated” version of the original problem
0
∂t u,n + AK u,n + f (u,n ) u,n = 0 in Ω × (0, ∞) ,
where we have set u,n = ∂t u,n . In particular, testing the aforementioned equation
with 2tu,n we deduce upon integrating over (0, t) that
Z t
2 2
t ku,n (t)kL2 (Ω) + 2 τ ku,n (τ )kW s,2 (Ω) dτ
K
0
Z t  
0 2
= − f (u,n (τ )) u,n (τ ) , u,n (τ ) + ku,n (τ )kL2 (Ω) dτ
0
Z t
2
≤ (Cf + 1) ku,n (τ )kL2 (Ω) dτ,
0

where in the last line we have used assumption (H3). Exploiting (3.58) we obtain
in the limit as (, n) → (0, ∞) that
2
sup k∂t u (t)kL2 (Ω) ≤ C.
t≥t1

The usual comparison argument in equation (3.38) together with the uniform bound
(4.1) yields u ∈ L∞ ((t1 , ∞); D (AK )) uniformly with respect to time. Thus, for any
bounded set U ⊂ L2 (Ω) , we have that ∪t≥t1 SK (t)U is relatively compact in L2 (Ω),
when endowed with the metric topology of L2 (Ω). Finally, applying [46, Theorem
I.1.1] we have that the set
AK = ∩τ ≥0 ∪t≥τ SK (t)BK
is a compact attractor for SK , and the rest of the result is immediate.
Step 2. (Uniform differentiability on AK ). We show that the bound obtained
in (4.1) is sufficient to show the uniform differentiability of SK on the attractor AK
0
with Υ (t; u0 ) ξ := SK (t) ξ as a solution of
0
∂t U + AK U + f (u (t)) U = 0, U (0) = ξ. (4.2)
To this end, consider two solutions u1 , u2 of problem (3.1) with initial conditions
ui (0) ∈ AK , i = 1, 2 and let U be the solution of (4.2) with ξ = u1 (0) − u2 (0).
Then the function ω (t) = u1 (t) − u2 (t) − U (t) satisfies the equation
0
∂t ω + AK ω + f (u) ω + g = 0, ω (0) = 0, (4.3)
0
with g := f (u1 ) − f (u2 ) − f (u) (u1 − u2 ). Next, by Taylor’s theorem and the fact
that ui ∈ L∞ (Ω) (as both u1 , u2 lie on the attractor AK ⊂ L∞ (Ω)), we infer that
2
|g (x)| ≤ C |u1 (x) − u2 (x)| , for some C > 0. Let r be the conjugate exponent
1308 CIPRIAN G. GAL AND MAHAMADI WARMA

s,2 ∗
to 2∗ from the Sobolev embedding inequality (2.1) such that WK (Ω) ⊂ L2 (Ω).
Therefore, if we write g (t) = g (u (x, t)) it follows that
Z
r 2r−2+δ 2−δ
kg (t)kLr (Ω) ≤ C |u1 (t) − u2 (t)| |u1 (t) − u2 (t)| dx
Ω
2−δ
≤ C ku1 (t) − u2 (t)kL2 (Ω) ,
owing to Hölder’s inequality and the L∞ (Ω) bound on u1 , u2 . Choosing now δ =
2 − r (1 + ε) , for some ε ∈ (0, (2 − r) /r), we easily deduce from (3.48) that
1+ε
kg (t)kLr (Ω) ≤ CeC(1+ε)t ku1 (0) − u2 (0)kL2 (Ω) .
Testing now equation (4.3) by ω (t) in L2 (Ω) yields
d 2
kωkL2 (Ω) + 2EK (ω, ω)
dt
2 2(1+ε)
≤ 2Cf kωkL2 + CeC(1+ε)t ku1 (0) − u2 (0)kL2 (Ω)
which upon integrating over (0, t) gives
2 2(1+ε)
kω (t)kL2 ≤ C (t) ku1 (0) − u2 (0)kL2 (Ω) ,
for some function C (t) > 0. Thus, the flow SK (t) is indeed differentiable on AK
0
and the derivative SK (t) is given by the solution of the linearized equation (4.2).
Finally, we also observe that for ξ ∈ L2 (Ω) the set Υ (t; u0 ) ξ is relatively bounded
s,2
in WK (Ω) ∩ L∞ (Ω), whence the mapping Υ (t; u0 ) is also compact in L2 (Ω) for
each t > 0. The desired finite dimensionality of the global attractor AK follows
from standard results in the theory of infinite dimensional dynamical systems (see,
e.g., [12, 46]). The proof of the theorem is now complete.
The following lemma states other basic properties of the dynamical
system associ-
ated with problem (3.1). In particular, it shows that SK (t) , L2 (Ω) is a “gradient”
system, namely, we have the following.
Lemma 4.2. Let the assumptions of Theorem 3.9 be satisfied. Then the functional
s,2
L K : WK (Ω) ∩ L∞ (Ω) → R, given by
1
LK (u(t)) := EK (u (t) , u (t)) + (F (u (t)) , 1)L2 (Ω) ,
2
has along the strong solutions of (3.1), the derivative
d 2
LK (u (t)) = − k∂t u (t)kL2 (Ω) , a.e. t > 0.
dt
In other words, the functional LK is decreasing, and becomes stationary exactly on
equilibria u∗ , which are solutions of the system:
AK u + f (u) = 0 in Ω. (4.4)
Proof. The proof is a consequence of the calculation (3.57) and the fact that strong
solutions are smooth enough, see Definition 3.2 and Remark 6.
The foregoing Lemma 4.2 can now be used to study the asymptotic behavior of
the solutions of (3.1) by means of the LaSalle’s invariance principle. To this end,
to any (weak) trajectory of (3.1) we associate the respective ω-limit set ωL2 :
ωL2 := y ∈ L2 (Ω) : ∃tn → ∞, zn ∈ L2 (Ω) such that


SK (tn ) zn → y in L2 -topology .

 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1309

The following lemma states some basic
properties of the ω-limit sets associated
with the dynamical system SK (t) , L2 (Ω) .
Lemma 4.3. (i) Any ω-limit set ωL2 is nonempty, compact and connected.
(ii) The trajectory approaches its own limit set in the norm of L2 (Ω), i.e.,
lim distL2 (Ω) (SK (t) u0 , ωL2 ) = 0.
t→∞

(iii) Any ω-limit set is invariant: new trajectories which start at some point in ωL2
remain in ωL2 for all times t > 0.
Proof. The proof is immediate owing to the continuity properties of the strong
s,2
solution and the compactness of the embedding WK (Ω) ⊂ L2 (Ω) .
The second main result is concerned with the parabolic problem

∂t u + dAE u + f (u) = 0, in Ω × (0, ∞) ,
(4.5)
u (0) = u0 , in Ω,
in the case K = E (recall that u = 0 in RN \Ω, N ≥ 1), with d > 0 playing the role
of a diffusion coefficient. Recall that 0 < s < 1.
Theorem 4.4. Let the assumptions of Theorem 3.8 be satisfied. The fractal di-
mension of AE admits the estimate
 N/2s
2

c∗ Cf
dimF AE , L (Ω) ≤ N/2s |Ω| , (4.6)
CE d
as either Cf → ∞ or d → 0+ , where c∗ depends on the shape of Ω and N only.
s 2s/N
Here we have set CE = (4π) Γ (1 + N/2) and Cf > 0 is such that (H3) is
satisfied.
Proof. In order to deduce (4.6), it is sufficient (see, e.g., [12, Chapter III, Definition
0
4.1]) to estimate the j-trace of the operator L (t, U (t)) := −AE U − f (u (t)) U, for
u ∈ AE . We have
m
X
Trace (L (t, U (t)) Qm ) = (L (t, U (t)) ϕj , ϕj )L2 (Ω)
j=1
m D E m
X 0 X
=− f (u (t)) ϕj , ϕj − d λj (ϕj , ϕj )L2 (Ω) ,
2
j=1 j=1

where the set of real-valued functions ϕj ∈ L2 (Ω) ∩ WEs,2 (Ω) is an orthonormal

basis in Qm (L2 (Ω)). Since the family ϕj is orthonormal in Qm L2 (Ω) , using
0
assumption (H3) on f (i.e., f (σ) ≥ −Cf , for all σ ∈ R), we find
m
X
Trace (L (t, U ) Qm ) ≤ −d λj + Cf m.
j=1
s
We now consider the eigenvalue problem (−∆) ϕ = λϕ in Ω and ϕ = 0 in RN \Ω,
which is equivalent to the eigenvalue problem AE ϕ = λϕ, ϕ ∈ D (AE ). By [5], the
eigenvalues λj obey the following Weyl asymptotic formula:
  2s
s jΓ (1 + N/2) N  
λj = (4π) + o j 2s/N as j → ∞. (4.7)
|Ω|
1310 CIPRIAN G. GAL AND MAHAMADI WARMA

− 2s 2s s 2s/N
From (4.7), since λj ≥ CE |Ω| N
j N with CE = (4π) Γ (1 + N/2) we obtain
m
− 2s 2s
X
Trace (L (t, U ) Qm ) ≤ −dCE |Ω| N
j N + Cf m
j=1
− 2s 2s
≤ −dc0 CE |Ω| N
m N +1 + Cf m,
for some c0 > 0 which only depends on the shape of Ω and N . Let us define the
function on the right-hand side as ρ (m). The function ρ is concave and the non-zero
root of the equation ρ (m) = 0 is
 N/2s
Cf
m∗ = |Ω| .
dc0 CE
Thus, we can apply [12, Corollary 4.2 and Remark 4.1] to deduce that
dimF AE , L2 (Ω) ≤ max {m∗ , 1} ,

from which (4.6) follows.

Remark 7. It is worth emphasizing that when s = 1, in (4.7) we obtain the
classical Weyl’s formula for the Dirichlet Laplacian eigenvalue problem −∆ϕ = λϕ
in Ω, ϕ = 0 on ∂Ω. Moreover, the upper bound in (4.6) also stabilizes as s → 1 to
the corresponding upper bound for the fractal dimension of the parabolic equation
∂t u − ∆u + f (u) = 0 in Ω × (0, ∞) , and u = 0 on ∂Ω × (0, ∞), see, for instance, [46,
Chapter VI], [12]. We conjecture that a similar bound on the dimension of AK also
holds in the remaining cases K ∈ {D, N , R} where at the moment Weyl asymptotic
formulas are not yet available.
Remark 8. One can also provide a lower bound on the dimension of AE :
0
!N/2s
2

−f (0)
dimF AE , L (Ω) ≥ c |Ω| , (4.8)
dCE
0
for some c > 0 which depends only on the shape of Ω and N , if f (0) < 0 is
sufficiently large or d > 0 is sufficiently small. This estimate is obtained in the
same spirit of [46, Chapter VII] (see also [22]) and relies on the fact that, owing
to the boundedness of u ∈ L∞ (Ω), the semigroup SE (t) is uniformly differentiable
with derivative of Hölder class C α , α ∈ (0, 1) (in fact, in our case α = 1). More
precisely, there exists a smooth manifold W loc (u∗ ) (of class C 1,α ) localized in an
open neighborhood of the hyperbolic equlibrium u∗ = 0, with finite instability
dimension dim X∗ (u∗ ) < ∞. Here, X∗ (u∗ ) is the unstable subspace of −AK −
0
f (u∗ ) which is tangent to W loc (u∗ ) at the point u∗ and we recall that the global
attractor always contains localized unstable manifolds.

5. Asymptotic stabilization to single equilibria. Let u be a weak solution of

(3.1) according to Theorem 3.8. We show that any such weak solution converges
(in a certain sense) to a single steady state as time tends to infinity. Recall that
any weak solution of (3.1) regularizes in finite time to a strong solution by Theorem
3.9. Moreover, observe that, by virtue of (3.50) and the proof of Theorem 4.1, all
stationary solutions u∗ ∈ ωL2 (u) of problem AK u∗ + f (u∗ ) = 0 in Ω, are bounded
in L∞ (Ω) ∩ D (AK ). Setting now, for each K ∈ {D, E, N , R} ,
Z Z σ
1
LK (u) := EK (u, u) + F (u) dx, F (σ) := f (t) dt,
2 Ω 0
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1311

it is easy to see that LK ∈ C 1 L2 (Ω) , R but LK ∈ / C 2 L2 (Ω) , R no matter how
smooth F is due to the nature of the nonlocal term EK (cf. also [19, 23, 39], where
the same issue occurs for other nonlocal problems).
Consequently, we shall employ a generalized version of the Lojasiewicz-Simon
theorem which is well-suited for our nonlocal problem (3.1). As usual, we fix K ∈
−N −2s
{D, E, N , R} and set J (r) := |r| . Recall that the Lojasiewicz-Simon result
applies in principle to functionals which can be written as a maximal monotone
operator plus a linear compact perturbation. The version that applies to our cases
K ∈ {D, E, N , R} is formulated in the subsequent lemma requiring the following
condition:
(H-er): Let w be a bounded solution of the elliptic boundary value problem
AK w = h in Ω, for some h ∈ L∞ (Ω). Then, w ∈ C 0,ν Ω for some ν ∈ (0, 1).

Lemma 5.1. Let F ∈ C 2 be a real analytic function satisfying (H1), (H3) and
let Ω obey condition (H4). Assume condition (H-er). Then, there exist constants
θ ∈ (0, 12 ], C > 0, ε > 0 such that the following inequality holds:
1−θ
|LK (u) − LK (u∗ )| ≤ C||AK u + f (u) ||L2 (Ω) (5.1)
s,2
for all u ∈ WK (Ω) ∩ L∞ (Ω) provided that ku − u∗ kL2 (Ω) ≤ ε.

We employ an abstract version of the Lojasiewicz-Simon inequality provided by

[20, Theorem 6] and which has been specifically tailored to our needs. We need the
following assumptions:
A1: Let (V, k·kV ) and (W, k·kW ) be Banach spaces densely and continuously
embedded into the Hilbert space (H, k·kH ) and its dual (H ∗ , k·kH ∗ ), respec-
tively. We assume that the restriction j|V of the duality map j ∈ L (H, H ∗ )
to V is an isomorphism from V onto W = j (V ).
A2: Let T ∈ L (H, H ∗ ) be a self-adjoint and completely continuous operator
such that its restriction T|V to V is a completely continuous operator in
L (V, W ). For fixed π ∈ W and d ∈ R consider the quadratic functional
Ψ : H → R given by
Ψ (u) = hT u, ui + hπ, ui + d, u ∈ H.
A3: Let U be an open subset of V and Φ : U → R be a Fréchet differentiable
function. Additionally, assume that the Fréchet derivative DΦ : U → R is a
real analytic operator which satisfies
2
hDΦ (u) − DΦ (v) , u − vi ≥ c1 ku − vkH ,
kDΦ (u) − DΦ (v)kH ∗ ≤ c2 ku − vkH
for all u, v ∈ U, for some constants c1 , c2 > 0. Moreover, assume that the
second Fréchet derivative D2 Φ (u) ∈ L (V, W ) is an isomorphism for all u ∈ U.
Theorem 5.2. ([20, Theorem 6]) Let the assumptions (A1)-(A3) hold for the energy
F defined by F = Φ + Ψ. Let u∗ ∈ U be a critical point of F, i.e., u∗ is a solution
of the Euler-Lagrange equation DF (u∗ ) = 0. Then there exist constants θ ∈ (0, 12 ],
C > 0, ε > 0 such that for each u ∈ U which satisfies ku − u∗ kH ≤ ε we have the
following inequality:
1−θ
|F (u) − F (u∗ )| ≤ C inf {kDF (z)kH ∗ : z ∈ H} .
1312 CIPRIAN G. GAL AND MAHAMADI WARMA

Proof of Lemma 5.1. We will apply the abstract result of Theorem 5.2 to the energy
functional LK with
V = W = L∞ (Ω) , H = L2 (Ω) .
According to its definition we can split LK into the sum of a convex (entropy)
functional Φ : L2 (Ω) → R ∪ {∞}, with a suitable effective domain, and a non-
local interaction functional Ψ : dom(Ψ) → R. To this end, we define the lower-
semicontinuous and strongly convex functional
 Z 
µ 
F (u) + u2 dx, if u ∈ L∞ (Ω)

Φ (u) := Ω 2
 +∞, otherwise,
where µ > Cf (with Cf > 0 as in assumption (H3)), with closed effective domain
dom(Φ) = L∞ (Ω), and the quadratic functional Ψ : L2 (Ω) → R, given by
s,2
Ψ (u) := ((AK − µ/2) u, u)L2 (Ω) , for all u ∈ WK (Ω) = dom (Ψ) .
We have that Φ is Fréchet differentiable on any open subset U of L∞ (Ω) with Fréchet
derivative DΦ : U → L∞ (Ω) having the form
Z
hDΦ (u) , ξi = (F 0 (u) + µu) · ξdx,
Ω
for all u ∈ U and ξ ∈ L∞ (Ω). The analyticity of DΦ as a mapping on L∞ (Ω) is
standard and can be proved exactly as in, e.g., [20, Remark 3]. Moreover, due to
assumption (H3), there holds
2
(DΦ (u1 ) − DΦ (u2 ) , u1 − u2 )L2 (Ω) ≥ ς ku1 − u2 kL2 (Ω) ,
for some ς ∈ (0, µ − Cf ) (which exists since µ > Cf ), for all u1 , u2 ∈ U, and
kDΦ (u1 ) − DΦ (u2 )kL2 (Ω) ≤ C ku1 − u2 kL2 (Ω) ,
for some positive constant C. Moreover, computing the second Fréchet derivative
D2 Φ of Φ, Z
(F 00 (u) + µ) ξ1 · ξ2 dx

2 
D Φ (u) ξ1 , ξ2 =
Ω
yields that D2 Φ ∈ L (L∞ (Ω) , L∞ (Ω)) is an isomorphism for every ϕ ∈ U. Next,
defining the linear operator TJ : L2 (Ω) → L2 (Ω), we have
Ψ (u) = hTJ u, ui + (π, u)L2 (Ω) = (AK u, u)L2 (Ω) + (π, u)L2 (Ω) ,
with π := − (µ/2) u ∈ L∞ (Ω) (indeed, every weak solution of the nonlinear ellip-
tic problem AK u + f (u) = 0, belongs to L∞ (Ω) ∩ D (AK ), owing to assumption
(H1)). By the regularity results provided in the previous section the operator
s,2
TJ ∈ L(L2 (Ω) , WK (Ω)) and hence it is compact from L2 (Ω) to

itself. Moreover,
by assumption (H-er), we have that TJ ∈ L(L∞ (Ω) , C 0,ν Ω ) is also compact
from L∞ (Ω) to C(Ω). Hence, the hypotheses of Theorem 5.2 are satisfied and the
sum
LK = Φ + Ψ : L2 (Ω) → R ∪ {∞}
is a well defined, bounded from below functional with nonempty, closed, and convex
effective domain dom(LK ) . Noting that
∗ the Fréchet derivative DLK (u) = AK u +
0
F (u) (here DLK : L2 (Ω) → L2 (Ω) , with L2 (Ω) being identified with its dual
by means of the Riesz isometry), inequality (5.1) follows from Theorem 5.2. The
proof is finished.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1313

We make some comments on the assumption (H-er).

Remark 9. We notice that by Proposition 2.7, if Ω is a bounded domain with
Lipschitz continuous boundary, then the assumption (H-er) is satisfied for the
operator AE . It is also satisfied by the operator AK , K ∈ {D, N , R} for bounded
domains in dimension N = 1 and if s > 1/2 (cf. Section 2). We think that
assumption (H-er) is also satisfied for AK , K ∈ {D, N , R} for bounded domains
with Lipschitz boundary, in any space dimension and for any s ∈ (0, 1). Such a result
is not yet available in the literature, and is interesting on its own independently of
the application given in the present paper, and also necessitates a careful study.
Since this is not the main objective of the paper, we will not go into details.
We can now prove the following convergence result by the application of Lemma
5.1.
Theorem 5.3. Let the assumptions of Theorem 3.8 hold and let F be real ana-
lytic. Then, for any initial datum u0 ∈ L2 (Ω) the corresponding weak solution u to
problem (3.1) satisfies
lim ku (t) − u∗ kL2 (Ω) = 0, (5.2)
t→∞
where u∗ ∈ D (AK ) ∩ L∞ (Ω) is a solution to (4.4), i.e., AK u∗ + f (u∗ ) = 0 in Ω.
Moreover, the following convergence rate holds:
θ
− 1−2θ
ku (t) − u∗ kL2 (Ω) ∼ (1 + t) as t → ∞, (5.3)
where θ ∈ (0, 12 ] is the constant given in (5.1).
Proof. By the energy identity in Lemma 4.2, we have
LK (u (t)) → L∗ = LK (u∗ ) , as t → ∞,
and the limit energy L∗ is the same for every steady-state solution u∗ ∈ ωL2 (u).
Moreover, we can integrate the energy equality in Lemma 4.2,
d 2
LK (u (t)) = − k∂t u (t)kL2 (Ω)
dt
over (t, ∞) to get
Z ∞
2
k∂t u (τ )kL2 (Ω) dτ = LK (u (t)) − L∗ = LK (u (t)) − LK (u∗ ) . (5.4)
t
By virtue of Lemma 5.1 and equation (3.1), we have
1−θ
|LK (u (t)) − LK (u∗ )| ≤ C kAK u (t) + f (u (t))kL2 (Ω)
≤ C k∂t u (t)kL2 (Ω)
provided that
ku − u∗ kL2 (Ω) ≤ ε. (5.5)
This, combined with the previous identity, yields
Z ∞ 1
2
k∂t u (τ )kL2 (Ω) dτ ≤ C k∂t u (t)kL(1−θ)
2 (Ω) , (5.6)
t
for all t > 0, as long as (5.5) holds. Note that, in general, the quantities θ, C and
ε above may depend on u∗ . Finally, let us set
M = ∪ {I : I is an open interval on which (5.5) holds} .
1314 CIPRIAN G. GAL AND MAHAMADI WARMA

Clearly, M is nonempty since u∗ ∈ ωL2 (u). We can now use (5.6), the fact that
∂t u (t) ∈ L2 (0, ∞) (cf. Lemma 4.2), and exploit [19, Lemma 6.1] with α = 2 (1 − θ)
to deduce that k∂t u (·)kL2 (Ω) ∈ L1 (M) and
Z
k∂t u (τ )kL2 (Ω) dτ ≤ C (u∗ ) < ∞. (5.7)
M
We now claim that we can find a sufficiently large time τ > 0 such that (τ, ∞) ⊂
M. To this end, recalling (5.4) and the above bounds, we also have that ∂t u ∈
L2 ((0, ∞); L2 (Ω)) and, furthermore, for any δ > 0 there exists a time t∗ = t∗ (δ) > 0
such that
k∂t ukL1 (M∩(t∗ ,∞);L2 (Ω)) ≤ δ, k∂t ukL2 ((t∗ ,∞);L2 (Ω)) ≤ δ. (5.8)
Next, observe that by Theorem 3.9 (see also (4.1)), there is a time t1 > 0 such that
sup ku (t)kW s,2 (Ω)+L∞ (Ω) ≤ C. (5.9)
K
t≥t1

Now, let (t0 , t2 ) ⊂ M, for some t2 > t0 ≥ t∗ (δ) , |t0 − t2 | ≥ 1 such that (5.9) holds
(w.l.o.g., we shall assume that t∗ ≥ t1 ). Using (5.8) and (5.9), we obtain
Z t2
2
ku (t0 ) − u (t2 )kL2 (Ω) =2 h∂t u (τ ) , u (τ )i dτ
t0
Zt2
≤2 k∂t u (τ )kL2 (Ω) ku (τ )kL2 (Ω) dτ
t0
≤C k∂t ukL1 ((t0 ,t2 );L2 (Ω)) kukL∞ ((t∗ ,∞);L∞ (Ω)) ≤ Cδ. (5.10)
Therefore we can choose a time t∗ (δ) = τ < t0 < t2 , such that
ε
ku (t0 ) − u (t2 )kL2 (Ω) < (5.11)
3
provided that (5.5) holds for all t ∈ (t0 , t2 ). Since u∗ ∈ ωL2 (u), a large (refined) τ
can be chosen such that
ku (τ ) − u∗ kL2 (Ω) < ε/3; (5.12)
hence, (5.11) yields (τ, ∞) ⊂ M. Indeed, taking
n o
t = inf t > τ : ku (t) − u∗ kL2 (Ω) ≥ ε ,


we have t > τ and u t − u∗ L2 (Ω) ≥ ε if t is finite. On the other hand, in view
of (5.11) and (5.12), we have
2
ku (t) − u∗ kL2 (Ω) ≤ ku (t) − u (τ )kL2 (Ω) + ku (τ ) − u∗ kL2 (Ω) <
ε,
3
for all t > t ≥ τ , and this leads to a contradiction. Therefore, t = ∞ and by
(5.8) the integrability of ∂t ϕ in L1 ((τ, ∞); L2 (Ω)) follows. Hence, ωL2 (u) = {u∗ }
and (5.2) holds. The convergence rate (5.3) is an immediate consequence of the
definition of LK and (5.7). We leave the details to the interested reader. The proof
is finished.
Remark 10. Exploiting the L2 (Ω) → (L∞ (Ω) ∩ D (AK )) smoothing property of
the weak and stationary solutions together with a similar argument from Theorem
3.9, and the convergence rate (5.3) it is possible to show the convergence rate:
1
−κ
ku (t) − u∗ kL∞ (Ω) ∼ (1 + t) , as t → ∞,
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1315

for some positive constant κ = κ (θ, u∗ ) ∈ (0, 1) . Indeed, we can also prove (3.51)
for the difference u − u∗ owing to the boundedness of u, u∗ ∈ L∞ (Ω) .

6. Some blow-up results. Our goal in this section is to show that assumption
(H1) is in fact quite optimal for global well-posedness of strong solutions of the
problem (3.1). We recall that this condition implies in particular that if f is a
source with a bad sign at infinity then it can only be of at most linear growth at
infinity. Indeed, we will show below by the concavity method of Levine–Payne [37]
that as soon as f has superlinear growth and a bad sign at infinity as |σ| → ∞,
p−1
as provided by the example f (σ) = − |σ| σ, σ ∈ R, with p > 1, then blowup in
finite time of some strong solutions occurs.
1
Theorem 6.1. Let Ω satisfy condition (H4) and suppose that f ∈ Cloc (R) obeys
1
F (σ) − f (σ) σ ≥ 0, for all σ ∈ R (6.1)
2
s,2
and u0 ∈ L∞ (Ω) ∩ WK (Ω) is an initial datum such that
Z
1
EK (u0 , u0 ) + F (u0 ) dx < 0, K ∈ {D, N , R, E}, (6.2)
2 Ω
then the strong solution of (3.1) must blow up in finite time.
Proof. For any strong solution of problem (3.1), which exists locally on some interval
t ∈ (0, Tmax ) by the proof of Theorem 3.7 (Step1), we have
Z
1 d 2
ku (t)kL2 (Ω) + EK (u (t) , u (t)) = − f (u (t)) u (t) dx (6.3)
2 dt Ω
and Z
0 2
QK (t) = (∂t u (t)) dx,
Ω
where we have set
Z
1
QK (t) := − EK (u (t) , u (t)) − F (u (t)) dx.
2 Ω
In particular, one has
Z t
2
QK (t) = QK (0) + k∂t u(τ )kL2 (Ω) dτ. (6.4)
0
Defining as usual the function
Z tZ
VK (t) := u2 (τ )dxdτ + A,
0 Ω
for some constant A > 0 to be determined later on, we see that
 Z 
0 2 00
VK (t) = ku (t)kL2 (Ω) , VK (t) = −2 EK (u (t) , u (t)) + f (u (t)) u (t) dx ,
Ω
(6.5)
owing to (6.3). Furthermore, by assumption (6.1) and (6.4), it holds
 Z t 
00 2
VK (t) ≥ 4QK (t) = 4 QK (0) + k∂t u(τ )kL2 (Ω) dτ . (6.6)
0
Clearly, since
Z tZ Z
0
VK (t) = 2 u(τ )∂t u(τ )dxdτ + u20 dx
0 Ω Ω
1316 CIPRIAN G. GAL AND MAHAMADI WARMA

it follows for any ε > 0, that

Z t Z  Z t Z 
0 2 2
VK (t) ≤4 (1 + ε) u2 (τ )dxdτ (∂t u) (τ )dxdτ
0 Ω 0 Ω
  Z 2
1 2
+ +1 u0 dx .
ε Ω

For α > 0, combining these estimates together yields

00 0 2
VK (t) VK (t) − (1 + α) VK (t) > 0

provided that ε and α are small enough such that (1 + α) (1 + ε) ≤ 1 and A > 0 is
large enough (since QK (0) > 0, by assumption). The foregoing inequality implies
as usual that
0 0
VK (t) VK (0)
1+α > 1+α for t > 0,
VK (t) VK (0)
which yields that the quantity VK (t) cannot remain finite for all time t > 0. The
proof is finished.

p−1
Remark 11. Let f (σ) = − |σ| σ with p > 1 and note that it satisfies (6.1)
but it fails to verify condition (H1). Therefore, for any initial condition u0 which
satisfies (6.2), blow-up must occur.

Remark 12. The same blow-up result also holds for some initial datum for which
EK (u0 , u0 ) > 0, see [25].

We conclude this section with another blow-up result by exploiting the well-
known eigenvalue method of Kaplan [35]. Let K ∈ {E, D} .

Theorem 6.2. Assume that h ∈ C 1 (R) is a concave decreasing function and

 0 
lim sup h (σ) < −λ1 ,
σ→∞

where λ1 > 0 is the first eigenvalue of the self-adjoint operator AE and such that
Z ∞
1
dσ < ∞ and f (σ) − κVΩ (x) σ ≤ h (σ) < 0 for all σ > σ0 , a.e. in Ω,
σ0 |h (σ)|

for some σ0 ≥ 0. Here κ = 0 if K = E and κ = 1 if K = D. There exist strong

solutions of problem (3.1) that blow up in finite time.

Proof. Let φ be the positive solution (cf. [45, Proposition 9]) of the eigenvalue
problem
AE φ = λ1 φ in Ω, φ = 0 in RN \Ω,
R
with Ω φ (x) dx = 1, and where λ1 > 0 is the first eigenvalue of AE . Next recall that
−2s/N
R potential VΩ (x) ≥ C |Ω|
the a.e. in Ω, see [48, Lemma 5.10]. Let yK (t) :=
Ω
u (x, t) φ (x) dx. Then we get, using (3.1) and Jensen’s inequality for the function
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1317

−h that
Z Z
0
yK (t) = ∂t uφdx = (−AK u − f (u)) φdx
Ω Ω
Z Z
=− u (AK φ + κVΩ φ) dx − (f (u) − κVΩ u) φdx
Ω
Z ZΩ 
≥ −λ1 u (x, t) φ (x) dx − h u (x, t) φ (x) dx
Ω Ω
= −λ1 yK (t) − h (yK (t)) ,
for as long as it exists, owing to the fact that
(u(x) − u(y))(φ(x) − φ(y))
Z Z
CN,s
(AK u + κVΩ u, φ)L2 = dxdy
2 R N R N |x − y|N +2s
= (u, AE φ)L2 .
If u (t) remains finite for all time then yK (t) is well-defined for all time t > 0.
0
However, from the ODE theory of equation yK = −h (yK ), yK will blow up in finite
time provided that yK (0) is sufficiently large. Therefore, the solution of (3.1) must
blow up in finite time under the given assumptions.
Remark 13. In particular, any function f obeying
f (σ)
lim sup p < 0,
σ→∞ σ (ln (σ))
for some p > 1, satisfies the above conditions.
Remark 14. By symmetry we can get the analogue of Theorem 6.2 in the case
when solutions of (3.1) blow up in finite time toward −∞ provided that h is a
decreasing, convex function such that
 0 
lim sup h (σ) < −λ1
σ→−∞

and
Z σ0
1
dσ < ∞ and f (σ) − κVΩ (x) σ ≥ h (σ) > 0 for all σ < σ0 , a.e. in Ω,
−∞ |h (σ)|
for some σ0 ≤ 0. We only need to apply the foregoing theorem on the equation
satisfied by −u.

REFERENCES

[1] S. Abe and S. Thurner, Anomalous diffusion in view of Einsteins 1905 theory of Brownian
motion, Physica, A 356 (2005), 403–407.
[2] D. R. Adams and L. I. Hedberg, Function Spaces and Potential Theory, Grundlehren der
Mathematischen Wissenschaften, 314, Springer-Verlag, Berlin, 1996.
[3] N. D. Alikakos, Lp -bounds of solutions to reaction-diffusion equations, Comm. Partial Dif-
ferential Equations, 4 (1979), 827–868.
[4] F. Andreu-Vaillo, J. M. Mazón, J. D. Rossi and J. J. Toledo-Melero, Nonlocal Diffusion
Problems, Mathematical Surveys and Monographs, 165, American Mathematical Society,
Providence, RI; Real Sociedad Matemática Española, Madrid, 2010.
[5] R. M. Blumenthal and R. K. Getoor, The asymptotic distribution of the eigenvalues for a
class of Markov operators, Pacific J. Math., 9 (1959), 399–408.
[6] K. Bogdan, K. Burdzy and Z. Q. Chen, Censored stable processes, Probab. Theory Related
Fields, 127 (2003), 89–152.
[7] J. Bourgain, H. Brezis and P. Mironescu, Limiting embedding theorems for W s,p when s ↑ 1
and applications, J. Anal. Math., 87 (2002), 77–101.
1318 CIPRIAN G. GAL AND MAHAMADI WARMA

[8] L. Caffarelli, J.-M. Roquejoffre and Y. Sire, Variational problems for free boundaries for the
fractional Laplacian, J. Eur. Math. Soc., 12 (2010), 1151–1179.
[9] L. Caffarelli, S. Salsa and L. Silvestre, Regularity estimates for the solution and the free
boundary of the obstacle problem for the fractional Laplacian, Invent. Math., 171 (2008),
425–461.
[10] L. Caffarelli and L. Silvestre, An extension problem related to the fractional Laplacian, Comm.
Partial Differential Equations, 32 (2007), 1245–1260.
[11] Z.-Q. Chen and T. Kumagai, Heat kernel estimates for stable-like processes on d-sets, Sto-
chastic Process. Appl., 108 (2003), 27–62.
[12] V. V. Chepyzhov and M. I. Vishik, Attractors for Equations of Mathematical Physics, Amer.
Math. Soc., Providence, RI, 2002.
[13] J. W. Cholewa and T. Dlotko, Global Attractors in Abstract Parabolic Problems, Cambridge
University Press, 2000.
[14] D. Danielli, N. Garofalo and D.-M. Nhieu, Non-doubling Ahlfors measures, perimeter mea-
sures, and the characterization of the trace spaces of Sobolev functions in Carnot-Carath
éodory spaces, Mem. Amer. Math. Soc., 182 (2006), x+119 pp.
[15] E. B. Davies, Heat Kernels and Spectral Theory, Cambridge University Press, Cambridge,
1989.
[16] E. Di Nezza, G. Palatucci and E. Valdinoci, Hitchhiker’s guide to the fractional Sobolev
spaces, Bull. Sci. Math., 136 (2012), 521–573.
[17] S. Dipierro, X. Ros-Oton and E. Valdinoci, Nonlocal problems with Neumann boundary
conditions, arXiv:1407.3313v1.
[18] Q. Du, M. Gunzburger, R. B. Lehoucq and K. Zhou, A nonlocal vector calculus, nonlocal
volume-constrained problems, and nonlocal balance laws, Math. Models Methods Appl. Sci.,
23 (2013), 493–540.
[19] E. Feireisl, F. Issard-Roch and H. Petzeltová, A non-smooth version of the Lojasiewicz-Simon
theorem with applications to non-local phase-field systems, J. Differential Equations, 199
(2004), 1–21.
[20] H. Gajewski and J. Griepentrog, A descent method for the free energy of multicomponent
systems, Discrete Contin. Dyn. Syst., 15 (2006), 505–528.
[21] C. G. Gal, On a class of degenerate parabolic equations with dynamic boundary conditions,
J. Differential Equations, 253 (2012), 126–166.
[22] C. G. Gal, Sharp estimates for the global attractor of scalar reaction-diffusion equations with
a Wentzell boundary condition, J. Nonlinear Science, 22 (2012), 85–106.
[23] C. G. Gal and M. Grasselli, Longtime behavior of nonlocal Cahn-Hilliard equations, Discrete
Contin. Dyn. Syst., 34 (2014), 145–179.
[24] C. G. Gal and M. Warma, Long-term behavior of reaction-diffusion equations with nonlocal
boundary conditions on rough domains, submitted.
[25] C. G. Gal and M. Warma, On some degenerate non-local parabolic equation associated with
the fractionalp-Laplacian, submitted.
[26] A. Garroni and S. Müller, A variational model for dislocations in the line tension limit, Arch.
Ration. Mech. Anal., 181 (2006), 535–578.
[27] P. Grisvard, Elliptic Problems in Nonsmooth Domains, Monographs and Studies in Mathe-
matics, 24, Pitman, Boston, MA, 1985.
[28] Q. Y. Guan, Integration by parts formula for regional fractional Laplacian, Comm. Math.
Phys., 266 (2006), 289–329.
[29] Q. Y. Guan and Z. M. Ma, Reflected symmetric α-stable processes and regional fractional
Laplacian, Probab. Theory Related Fields, 134 (2006), 649–694.
[30] Q. Y. Guan and Z. M. Ma, Boundary problems for fractional Laplacians, Stoch. Dyn., 5
(2005), 385–424.
[31] C. Gui and M. Zhao, Traveling wave solutions of Allen–Cahn equation with a fractional
Laplacian, Annales de l’Institut Henri Poincaré Non Linear Analysis,, in press, (2014).
[32] M. Gunzburger and R. B. Lehoucq, A nonlocal vector calculus with application to nonlocal
boundary value problems, Multiscale Model. Simul., 8 (2010), 1581–1598.
[33] M. Jara, Nonequilibrium scaling limit for a tagged particle in the simple exclusion process
with long jumps, Comm. Pure Appl. Math., 62 (2009), 198–214.
[34] A. Jonsson and H. Wallin, Function Spaces on Subsets of RN , Math. Rep., 2 (1984), xiv+221
pp.
 FRACTIONAL REACTION-DIFFUSION EQUATIONS 1319

[35] S. Kaplan, On the growth of solutions of quasilinear parabolic equations, Comm. Pure Appl.
Math., 16 (1963), 305–330.
[36] M. Koslowski, A. M. Cuitino and M. Ortiz, A phasefield theory of dislocation dynamics,
strain hardening and hysteresis in ductile single crystal, J. Mech. Phys. Solids, 50 (2002),
2597–2635.
[37] H. A. Levine and L. E. Payne, Nonexistence theorems for the heat equation with nonlinear
boundary conditions and for the porous medium equation backward in time, J. Differential
Equations, 16 (1974), 319–334.
[38] J.-L. Lions and E. Magenes, Non-homogeneous Boundary Value Problems and Applications,
Vol. I, Springer-Verlag, New York-Heidelberg, 1972.
[39] S.-O. Londen and H. Petzeltová, Convergence of solutions of a non-local phase-field system,
Discrete Contin. Dyn. Syst. Ser. S, 4 (2011), 653–670.
[40] A. Mellet, S. Mischler and C. Mouhot, Fractional diffusion limit for collisional kinetic equa-
tions, Arch. Ration. Mech. Anal., 199 (2011), 493–525.
[41] Y. Nec, A. A. Nepomnyashchy and A. A. Golovin, Front-type solutions of fractional Allen-
Cahn equation, Phys. D, 237 (2008), 3237–3251.
[42] D. Schertzer, M. Larcheveque, J. Duan, V. V. Yanovsky and S. Lovejoy, Fractional Fokker-
Planck equation for nonlinear stochastic differential equations driven by non-Gaussian Lé vy
stable noises, J. Math. Phys., 42 (2001), 200–212.
[43] R. E. Showalter, Monotone Operators in Banach Space and Nonlinear Partial Differential
Equations, Amer. Math. Soc., Providence, RI, 1997.
[44] X. Ros-Oton and J. Serra, The Dirichlet problem for the fractional Laplacian: Regularity up
to the boundary, J. Math. Pures Appl., 101 (2014), 275–302.
[45] R. Servadei and E. Valdinoci, Variational methods for non-local operators of elliptic type,
Discrete Contin. Dyn. Syst., 33 (2013), 2105–2137.
[46] R. Temam, Infinite-Dimensional Dynamical Systems in Mechanics and Physics, Springer-
Verlag, New York, 1997.
[47] L. Vlahos, H. Isliker, Y. Kominis and K. Hizonidis, Normal and anomalous diffusion: A
tutorial, in Order and Chaos (ed. T. Bountis), Vol. 10, Patras University Press, 2008.
[48] M. Warma, The fractional relative capacity and the fractional Laplacian with Neumann and
Robin boundary conditions on open sets, Potential Anal., 42 (2015), 499–547.
[49] M. Warma, Integration by parts formula and regularity of weak solutions of non-local equa-
tions involving the fractional p-Laplacian with Neumann and Robin boundary conditions on
open sets, preprint.

Received November 2014; revised March 2015.

E-mail address: [email protected]
E-mail address: [email protected], [email protected]