Algerian Journal of Sciences AJS / Volume 1 – Issue 1 – PP.

1-12 – 2024

On a Singular Degenerate
Reaction-Diffusion Model Applied to
Quenching and Biology
Samiha Djemai1 , Salim Mesbahi2∗

1,2LMFN Laboratory, Department of Mathematics,

Faculty of Sciences, Setif 1 University Ferhat Abbas, Setif, Algeria
1 Email: [email protected]
2 Email: [email protected]

Abstract
In this paper, we are interested in studying a singular nonlinear de-
generate reaction-diffusion model having a natural growth with respect
to the gradient. Our approach uses Schauder’s fixed point theorem.
This type of problem has numerous important applications across mul-
tiple disciplines, such as biology, ecology and medicine. By employing
rigorous mathematical techniques, we aim to advance the theoretical
understanding of this type of nonlinear degenerate reaction-diffusion
problems and lay the groundwork for further developments and real-
world implementations.

Classification: 35K57, 35K67, 35K65, 35D30

Keywords: Reaction-diffusion equations, degenerate parabolic equations,
singular parabolic equations, natural growth term.

1 Introduction
Singular degenerated reaction-diffusion systems are a class of partial dif-
ferential equations that exhibit a unique mathematical structure. These
systems are characterized by the presence of a singular or degenerate term
in the equation, which can lead to complex behavior and interesting appli-
cations. In the context of quenching, singular degenerated reaction-diffusion
∗
Corresponding author
 S. Djemai and S. Mesbahi

systems can be used to model phenomena such as the quenching of flames or

the extinction of biological populations. The singular term in the equation
can represent a critical threshold or a phase transition, where the system’s
behavior changes dramatically. For example, consider a model for the spread
of a fire in a forest. The reaction-diffusion equation can describe the dynam-
ics of the fire, with the singular term representing the critical temperature
at which the fire is extinguished. By analyzing the properties of the sin-
gular degenerated reaction-diffusion system, researchers can gain insights
into the conditions under which the fire will be quenched and the factors
that influence the quenching process. Similarly, in the context of biology,
singular degenerated reaction-diffusion systems can be used to model the
dynamics of biological populations, such as the spread of infectious diseases
or the growth of tumors. The singular term in the equation can represent
a critical population density or a phase transition in the system’s behavior,
such as the onset of extinction or the transition to a state of uncontrolled
growth. The analysis of singular degenerated reaction-diffusion systems of-
ten involves techniques from nonlinear analysis, such as the study of free
boundaries, the analysis of steady-state solutions, and the investigation of
the asymptotic behavior of the solutions. These systems can exhibit rich
and complex dynamics, including the formation of patterns, the occurrence
of bifurcations, and the existence of multiple stable states. The applica-
tions of singular degenerated reaction-diffusion systems span various fields,
including ecology, epidemiology, materials science, and even social sciences.
By understanding the mathematical properties of these systems and their
connection to real-world phenomena, researchers can develop more accurate
models and gain insights that inform practical applications. We find nu-
merous real applications in biology, medicine and ecology in the works of
Mesbahi et al. Mesbahi et al. [10, 11, 18, 19, 20], and also in DiBendetto
[9], Murray [21, 22] and corresponding references therein.
The problem we will study here fits into this context, and we find that it
has many important applications in biology, medicine, the environment, and
many other interesting scientific fields. The analysis of these systems often
involves advanced mathematical techniques. By understanding the rich and
complex dynamics exhibited by these systems, researchers can develop more
accurate models that provide valuable insights in various scientific domains.
We are interested in the following nonlinear singular degenerated reaction-

2
 S. Djemai and S. Mesbahi

diffusion system having natural growth with respect to the gradient

∂u |∇u|p

 − div (a (t, x) ∇u) + d (t, x) = f (t, x) in QT
uγ

∂t (1)
 u (t, x) = 0 on Γ

u (0, x) = u0 (x) in Ω

where Ω is a bounded open subset of RN , N > 2, and Q is the cylinder

(0, T ) × Ω, T > 0, Γ = (0, T ) × ∂Ω, 2 < p < N , 0 < γ < 1, a (t, x) and
d (t, x) are two bounded measurable functions satisfying

0 < α1 ≤ a (t, x) ≤ α2 (2)

0 < β1 ≤ d (t, x) ≤ β2 (3)

where α1 , α2 , β1 and β2 are fixed real numbers such that α1 < α2 and
β1 < β2 . On the function f , we assume that it is non-negative and not
identically zero, and that it belongs to the Lebesgue space Lm (QT ) with m >
1. Moreover, the initial data u0 ∈ L∞ (Ω) satisfies the following condition
of strict positivity
∃Dω > 0, ∀ω ⊂⊂ Ω : u0 ≥ Dω
We find a detailed history of this problem and its applications in nu-
merous areas in Benkirane et al. [2], Boccardo et al. [3], Dall’Aglio et al.
[5, 6, 7], De Bonis and Giachetti [8], El Hadfi et al. [12], El Ouardy and
El Hadfi [13], Keller and Choen [15], Magliocca [16], Martı́nez-Aparicio and
Petitta [17], Nachman and Callegari [23], Youssfi et al. [25] and references
therein.
The problem described in (1) presents several key difficulties that must
be addressed. Firstly, the presence of the lower-order term introduces com-
plications. The natural growth term in the equation depends on the gradi-
ent, which adds mathematical complexity. Additionally, the singularity in
the equation depends on the variable u, further complicating the analysis.
Perhaps most challenging is the need to prove the strict positivity of the
solution within the interior of the parabolic cylinder. Establishing this posi-
tivity property is a non-trivial task. To overcome these difficulties, we must
approximate the singular problem (1) by another non-singular one, and we
show that this problem admits a non-negative solution (the proof is based
on the application of Schauder’s fixed point theorem) and that this solution
is strictly positive in the interior of the parabolic cylinder (the proof is based
on the use of the intrinsic Harnack inequality).
Below, we will use |E| to denote the Lebesgue measure of a subset E ⊂
q
R . The Hölder conjugate exponent of q > 1 is q 0 = q−1
N , while the Sobolev

3
 S. Djemai and S. Mesbahi

Np
conjugate exponent of p for 1 ≤ p < N is N −p . For a fixed k > 0, we define
the truncation function Tk as follows:
Tk (s) = max {−k, min {s, k}}
To streamline notation, we will use C to represent values that may vary
from step to step or even within the same step, depending on some param-
eters. However, C will remain constant with respect to the indices of any
sequences introduced.

2 Statement of the main result

First, we have to clarify in which sense we want to solve our problem.
 
Definition 1 A weak solution to problem (1) is a function u ∈ L1 0, T ; W01,1 (Ω)
such that for every ω ⊂⊂ Ω there exists cω such that u ≥ cω > 0 in ω×(0, T ),
  |∇u|p
a (t, x) |∇u|p−1 ∈ L1 (QT ), ∈ L1 0, T ; L1loc (Ω) . Furthermore, we


uγ
have that Z Z
∂φ
− u dtdx + a (t, x) ∇u.∇φdtdx
QT ∂t QT
|∇u|p
Z Z Z
+ d (t, x) φdtdx = f φdtdx + u0 (x) φ (0, x)
QT uγ QT Ω

for every φ ∈ Cc1 ([0, T ) × Ω).

Now, we can state the main result of this paper, it is the following
theorem.
p (N + 1 + δ)
Theorem 2 Let 0 < γ < 1, λ = and δ = min {γ, 1 − q}.
p (N + 1 + δ) − N δ
N
Assume that a satisfy (2), d satisfy (3) and f ∈ Lm (QT ) with 1 < m < p+1 .
Then there exists a solution u of problem (1) in the sense of Definition 1
verify the following regularity:
N  
(i) If λ ≤ m < , then u ∈ Lp 0, T ; W01,p (Ω) ∩ Lσ (QT ) with
p+1
N (p − δ) + p
σ=m
N − pm + p
 
(ii) If 1 < m < λ, then u ∈ Ls 0, T ; W01,s (Ω) ∩ Lσ (QT ) with

N (p − δ) + p
s=m
N + 1 − δ (m − 1)

4
 S. Djemai and S. Mesbahi

3 Approximating Scheme
Let 0 < ε < 1. We approximate problem (1) by the following nonlinear and
non-singular problem
uε |∇uε |p

∂uε

 − div (a (t, x) ∇uε ) + d (t, x) = fε (t, x) in QT
∂t (ε + |uε |)γ+1

u (t, x) = 0 on Γ
 ε


uε (0, x) = uε0 (x) in Ω
(4)
where
f
fε = and fε ∈ L∞ (QT )
1 + εf
such that
kfε kLm (QT ) ≤ kf kLm (QT ) and fε → f strongly in Lm (QT ) , m > 1 (5)
and
u0 (x)
uε0 (x) = ∈ L∞ (Ω)
1 + εu0 (x)
such that
kuε0 kL∞ (Ω) ≤ kuε kL∞ (Ω) and uε0 → u0 strongly in L1 (Ω)
 
Problem (4) admits weak solutions uε ∈ Lp 0, T ; W01,p (Ω) ∩ L∞ (QT ),
as shown in references Dall’Aglio and Orsina [4] and Lions [14]. Addi-
tionally, the solution of problem (4) is continuous in time, meaning uε ∈
C [0, T ] ; L1loc (Ω) . Since the right side of (4) is non-negative, uε is also

non-negative.
Lemma 3 Let uε be solutions to problem (4). Then
uε |∇uε |p
Z
1
1− m
d (t, x) ≤ |QT | kf kLm (QT ) + ku0 kL1 (Ω)
QT (ε + |uε |)γ+1
Proof. See DiBendetto [9].
Remark 4 According to Lemma 3, and since
uε |∇uε |p
Z
1
d (t, x) γ+1 ≥ 0 for f ∈ L (QT )
QT (ε + |uε |)
one has that
uε |∇uε |p
Z
1
1− m
d (t, x) γ+1 − f ≤ 2 |QT | kf kLm (QT ) + ku0 kL1 (Ω) < C
QT (ε + |uε |)

5
 S. Djemai and S. Mesbahi

Lemma 5 Let the assumptions of Theorem 2 be in force. Then the solution

uε of (4) satisfy the following estimate:
N
(i) If λ ≤ m < , then uε is uniformly bounded in the space
p+1
 
Lp 0, T ; W01,p (Ω) ∩ Lσ (QT )

(ii) If 1 < m < λ, then uε is uniformly bounded in the space

 
Ls 0, T ; W01,s (Ω) ∩ Lσ (QT )

where s and σ are defined in Theorem 2.

4 Proof of the main result

Now we can prove Theorem 2.
Proof of Theorem 2. In view of Lemma 5,
 
uε * u weakly in Lδ 0, T ; W01,δ (Ω) , ∀δ < s < p and a.e. in QT
p
By Remark 4, fε − d (t, x) (uuε |∇uε|
+ε)γ+1
∈ L1 (QT ) and from Lemma 5, we have
ε
s
a (t, x) ∇uε is bounded in Lρ (QT ) for all 1 ≤ ρ < p−1 < p. Then div
0
 
(a (t, x) ∇uε ) is bounded in the space L (QT ) ⊂ L (QT ) ⊂ Lp 0, T ; W0−1,p (Ω) ,
ρ0 p0 0

0
 
and then ∂u ε
is bounded in the space Lp0 0, T ; W −1,p (Ω) +L1 (Q ) . Using
∂t 0 T
the compactness results in Simon [24], we obtain
uε → u strongly in L1 (QT ) and a.e. in QT . (6)
We can use the same proof as in Abdellaoui and Redwane [1], we obtain
 
Tk (uε ) → Tk (u) strongly in Lp 0, T ; W01,p (Ω)

and also we have

∇uε → ∇u in a.e. in QT (7)
On the other hand, recalling (2), (6), (7), Lemma 5 and the dominated
convergence theorem implies that the sequence a (t, x) ∇uε converges weakly
s
to a (t, x) ∇u in Lρ (QT ) for every 1 ≤ ρ < p−1 . Therefore, for every ϕ ∈
1
Cc (Ω × [0, T )) ,
Z Z
lim a (t, x) ∇uε ∇ϕ = a (t, x) ∇u∇ϕ (8)
ε→0 Q Q

6
 S. Djemai and S. Mesbahi

Now we prove that

uε |∇uε |p |∇u|p
d (t, x) → d (t, x) , strongly locally in L1 (QT )
(uε + ε)γ+1 uγ

For any measurable compact subset E of QT , we have

uε |∇uε |p uε |∇uε |p
Z Z
d (t, x) = d (t, x)
E (uε + ε)γ+1 E∩{uε ≤k} (uε + ε)γ+1
uε |∇uε |p
Z
+ d (t, x)
E∩{uε >k} (uε + ε)γ+1
|∇uε |p
Z
≤ d (t, x)
E∩{uε ≤k} uγε
uε |∇uε |p
Z
+ d (t, x)
E∩{uε >k} (uε + ε)γ+1

We get

uε |∇uε |p uε |∇uε |p
Z Z Z
1 p
d (t, x) ≤ γ d (t, x) |∇Tk (uε )| + d (t, x)
E (uε + ε)γ+1 cε E E∩{uε >k} (uε + ε)γ+1

Let ν > 0 be fixed. For k > 1 , we use T1 (uε − Tk−1 (uε )) as a test function
in (4), yielding
Z T Z Z
∂uε
T1 (uε − Tk−1 (uε )) + a (t, x) ∇un T1 (uε − Tk−1 (uε ))
0 Ω ∂t QT
uε |∇uε |p
Z
+ d (t, x) T1 (uε − Tk−1 (uε ))
QT (uε + ε)γ+1
Z
= fε T1 (uε − Tk−1 (uε ))
QT

Recalling (2) and the fact uε ≥ 0, we can write

Z Z
S1 (uε (T )) + α1 |∇uε |p
Ω {k−1≤uε ≤k}
uε |∇uε |p
Z
+ d (t, x) T1 (uε − Tk−1 (uε ))
QT (uε + ε)γ+1
Z Z
≤ fε T1 (uε − Tk−1 (uε )) + S1 (u0 ) ,
QT QT

7
 S. Djemai and S. Mesbahi

where Z uε (T )
S1 (uε (T )) = T1 (s − Tk−1 (s)) ds.
0
It is easy to see that S1 (uε (T )) ≥ 0 a.e. in Ω. After the first and second
non-negative terms of the previous inequality are removed, we arrive at

uε |∇uε |p
Z
d (t, x) T1 (uε − Tk−1 (uε )) (9)
QT (uε + ε)γ+1
Z Z
≤ fε T1 (uε − Tk−1 (uε )) + S1 (u0 )
QT QT
Z Z Z u0
= fε T1 (uε − Tk−1 (uε )) T1 (s − Tk−1 (s)) ds.
QT Ω 0

Since T1 (uε − Tk−1 (uε )) ≥ 0,


0 if uε ≤ k − 1
T1 (uε − Tk−1 (uε )) =
1 if uε > k

recalling the condition (3) and the fact that uε > 0, we have

uε |∇uε |p
Z
d (t, x) T1 (uε − Tk−1 (uε ))
QT (uε + ε)γ+1
uε |∇uε |p
Z
= d (t, x) T1 (uε − Tk−1 (uε ))
QT ∩{uε >k} (uε + ε)γ+1
uε |∇uε |p
Z
+ d (t, x) T1 (uε − Tk−1 (uε ))
QT ∩{uε ≤k} (uε + ε)γ+1
uε |∇uε |p
Z
= d (t, x)
QT ∩{uε >k} (uε + ε)γ+1
uε |∇uε |p
Z
+ d (t, x) T1 (uε − Tk−1 (uε ))
Q∩{uε ≤k} (uε + ε)γ+1
uε |∇uε |p
Z
≥ d (t, x)
E∩{uε >k} (uε + ε)γ+1

8
 S. Djemai and S. Mesbahi

and
Z
fε T1 (uε − Tk−1 (uε ))
QT
Z Z
= fε T1 (uε − Tk−1 (uε )) + fε T1 (uε − Tk−1 (uε ))
QT ∩{uε ≤k−1} QT ∩{k−1<uε ≤k}
Z
+ fε T1 (uε − Tk−1 (uε ))
Q∩{uε >k}
Z Z
= fε T1 (uε − Tk−1 (uε )) + fε
QT ∩{k−1<uε ≤k} QT ∩{uε >k}
Z Z Z
= f+ f+ f
QT ∩{uε ≤k−1} QT ∩{k−1<uε ≤k} QT ∩{uε >k}

also we have
Z Z Z u0 Z Z
S1 (u0 ) = T1 (s − Tk−1 (s)) ds = T1 (s − Tk−1 (s)) ds
Ω Ω 0 Ω [0,u0 ]∩{s>k−1}

Therefore, from (9) combined with the two later inequalities and the above
equality, we obtain

uε |∇uε |p
Z
d (t, x)
E∩{uε >k} (u + ε)γ+1
Z Z εZ
≤ f+ T1 (s − Tk−1 (s)) ds
Q∩{uε ≥k−1} E [0,u0 ]∩{s>k−1}

It follows from f ∈ Lm (Q) and T1 (s − Tk−1 (s)) ∈ L1 (Ω) that

uε |∇uε |p
Z
d (t, x) → 0 as k → ∞
E∩{uε >k} (uε + ε)γ+1

Then, there exists k0 > 1 such that

uε |∇uε |p
Z
ν
d (t, x) γ+1 ≤ 2 , ∀k > k0 , ∀ε ∈ (0, T ) (10)
E∩{uε >k} (uε + ε)
 
Since from (??) (Tk (uε ) → Tk (u)) strongly in Lp 0, T ; W01,p (Ω) , then
there exists εν , θν such that |E| ≤ θν , and we have
Z
1 ν
γ d (t, x) |∇Tk (uε )|p ≤ , ∀ε ≤ εν (11)
cε E 2

9
 S. Djemai and S. Mesbahi

p
The estimates (10) and (11) imply that d (t, x) (uuε |∇uε|
+ε)γ+1
is equi-integrable.
ε
p
This fact, together with the a.e. convergence of this term to d (t, x) |∇u|
uγ ,
implies by the Vitali Theorem that

uε |∇uε |p |∇u|p
d (t, x) γ+1 → d (t, x) γ
, strongly locally in L1 (QT ) (12)
(uε + ε) u

Let ϕ ∈ Cc1 (Ω × [0, T )), taking ϕ test function in problem (4), by (5), (6),
(8) and (12), we can let ε → 0 yielding

|∇u|p
Z Z Z
∂ϕ
− u + a (t, x) ∇u∇ϕ + d (t, x) ϕ
QT ∂t QT QT uγ
Z Z
= fϕ + u0 (x) ϕ (x, 0)
QT Ω

Thus, Theorem 2 is proved.

Acknowledgements
We extend our sincere thanks to Dr. Messaoud Aounallah for his invalu-
able support and guidance in completing this scientific paper. His expertise,
constructive feedback, and tireless assistance throughout the research and
writing process were instrumental in helping us produce a high-quality, im-
pactful publication. This paper would not have reached its full potential
without his generous mentorship and collaborative spirit.

References
[1] M. Abdellaoui, H. Redwane, On some regularity results of parabolic
problems with nonlinear perturbed terms and general data. Partial Dif-
fer. Equ. Appl., 3 (2022), Article no. 2.

[2] A. Benkirane, B. El Haji, M. El Moumni, On the existence of solu-

tion for degenerate parabolic equations with singular terms. Pure Appl.
Math. Q., 14 (3–4) (2018), 591–606.

[3] L. Boccardo, L. Orsina, M. M. Porzio, Existence result for quasilin-

ear elliptic and parabolic problems with quadratic gradients terms and
sources. Adv. Calc. Var., 4 (2011), 397–419.

10
 S. Djemai and S. Mesbahi

[4] A. Dall’Aglio, L. Orsina, Nonlinear parabolic equations with natural

growth conditions and L1 data. Nonlinear Anal., 27 (1996), 59–73.
[5] A. Dall’Aglio, D. Giachetti, J. P. Puel, Nonlinear parabolic equations
with natural growth in general domains. Boll. Unione Mat. Ital., 3
(2005), 653–684.
[6] A. Dall’Aglio, D. Giachetti, S. De León, Nonlinear parabolic problem
wit a very general quadratic gradient term. Differential Integral Equa-
tions, 20 (4) (2007), 361–396.
[7] A. Dall’Aglio, L. Orsina, F. Petitta, Existence of solutions for degen-
erate parabolic equations with singular terms. Nonlinear Anal., 131
(2016), 273–288.
[8] I. De Bonis, D. Giachetti, Singular parabolic problems with possibly
changing sign data. Discrete Contin. Dyn. Syst. Ser. B, 19 (7) (2014),
2047–2064.
[9] E. DiBendetto, Degenerate parabolic equations. Springer-Verlag, New
York, 1993.
[10] S. Djemai, S. Mesbahi, Singular Reaction-Diffusion System Arising
From Quenching. Nonlinear Dyn. Syst. Theory, 23 (5) (2023), 499–506.
[11] S. Djemai, S. Mesbahi, K. I. Saffidine, Quenching Reaction-Diffusion
Systems in Bioengineering and Life Sciences. Engineering and Technol-
ogy Journal, 9 (1) (2024), 3366–3371.
[12] Y. El Hadfi, A. Benkirane, A. Youssfi, Existence and regularity results
for parabolic equations with degenerate coercivity. Complex Var. Ellip-
tic Equ., 63 (5) (2018), 517–529.
[13] M. El Ouardy, Y. El Hadfi, Some nonlinear parabolic problems with
singular natural growth term. Results Math., 77 (2022), Article no. 95.
[14] J. L. Lions, Quelques méthodes de résolution des problémes aux limites
non linéaires. Dunod, Paris, 1969.
[15] H. B. Keller, D. S. Choen, Some positone problems suggested by non-
linear heat generation. J. Math. Mech., 16 (1967), 1361–1376.
[16] M. Magliocca, Existence results for a Cauchy-Dirichlet parabolic prob-
lem with a repulsive gradient term. Nonlinear Anal., 166 (2018), 102–
143.

11
 S. Djemai and S. Mesbahi

[17] P. J. Martı́nez-Aparicio, F. Petitta, Parabolic equations with nonlinear

singularities. Nonlinear Anal., 74 (2011), 114–131.

[18] S. Mesbahi, On the existence of weak solutions for a class of singular

reaction-diffusion systems. Hacet. J. Math. Stat., 51 (3) (2022) 757–774.

[19] A. Mesbahi, S. Mesbahi, On the Existence of Periodic Solutions of a

Degenerate Parabolic Reaction-Diffusion Model. Nonlinear Dyn. Syst.
Theory, 22 (2) (2022) 197–205.

[20] S. Mesbahi, Analyse mathématique de systèmes de réaction-diffusion

quasi-linéaires. Editions Universitaires Européennes, 2019. ISBN-13 :
978-6138492917.

[21] J. D. Murray, Mathematical Biology I : An Introduction. Springer-

Verlag, New York Inc.; 3rd edition, 2002.

[22] J. D. Murray, Mathematical Biology II : Spatial Models and Biochem-

ical Applications. Springer-Verlag, New York Inc.; 3rd edition, 2002.

[23] A. Nachman, A. Callegari, A nonlinear singular boundary value prob-

lem in the theory of pseudoplastic fluids. SIAM J. Appl. Math., 38
(1980), 275–281.

[24] J. Simon, Compact sets in the space Lp (0, T ; B). Ann. Mat. Pura Appl.,
(4) 146 (1987), 65–96.

[25] A. Youssfi, A. Benkirane, Y. El Hadfi, On bounded solutions for nonlin-

ear parabolic equations with degenerate coercivity. Mediterr. J. Math.,
13 (2016), 3029–3040.

12