We establish local boundedness, together with quantitative estimates, for non-negative weak sub-solutions to a class of doubly nonlinear parabolic equations whose prototype is (uq)t−div(∣Du∣p(x,t)−2Du)=0,p(x,t)>1,q>0. The results are obtained under a logarithmic continuity condition on the variable exponent p(x,t) and suitable relations between q and the local bounds of p(x,t). In certain ranges of the exponents, an additional local integrability assumption is required.
Introduction and Main Results
Nonlinear parabolic equations with nonstandard growth conditions arise naturally in various models of continuum mechanics and have attracted considerable attention in recent decades. An important class of such equations is associated with the mathematical modeling of electrorheological fluids, whose mechanical properties depend on the strength of an applied electric field; see [17, 19, 21]. The corresponding constitutive laws lead to differential operators with variable exponent growth, reflecting the nonhomogeneous rheological behaviour of these materials. A typical model is the p(x)-Laplace equation
−div(∣Du∣p(x)−2Du)=0.
whose mathematical analysis is closely related to the theory of Lebesgue and Sobolev spaces with variable exponents; see [10]. Such operators generalize the classical p-Laplacian and give rise to a rich mathematical structure due to the loss of homogeneity and uniform ellipticity. In evolutionary problems, the exponent may depend on both spatial and temporal variables, leading to parabolic equations of the form
ut−div(∣Du∣p(x,t)−2Du)=0.
The existence and regularity properties of weak solutions to such equations have been extensively investigated. Antontsev and Shmarev [2, 3] established existence results and boundedness estimates for weak solutions to more general anisotropic parabolic equations with variable exponent growth. Xu and Chen [23] proved the local boundedness of weak solutions and, under the additional assumption that the lower bound of p(x,t) exceeds 2, established their interior Hölder continuity. Subsequently, Alkhutov and Zhikov [1] obtained Hölder continuity results for solutions to parabolic equations with variable exponents of nonlinearity. Wang [22] established intrinsic Harnack inequalities for parabolic equations with variable exponents. Bögelein and Duzaar [6] proved the local Hölder continuity of the spatial gradient of weak solutions to parabolic p(x,t)-Laplacian systems under suitable Hölder continuity assumptions on the exponent. Yao [24] extended the gradient regularity theory by establishing local Hölder continuity of the spatial gradient of weak solutions to general nonhomogeneous parabolic equations with p(x,t)-growth. Furthermore, Baroni and Bögelein [5] obtained local Calderón–Zygmund estimates for parabolic p(x,t)-Laplacian systems under a logarithmic continuity assumption on p(x,t), showing that the spatial gradient inherits higher integrability properties from the prescribed data.
In this work, we investigate the local boundedness of weak solutions to a class of doubly nonlinear parabolic equations with variable exponent growth, whose prototype is
(uq)t−div(∣Du∣p(x,t)−2Du)=0,p(x,t)>1,q>0.
For the constant exponent case p(x,t)≡p>1, equations of this type are well known, have numerous applications, and have been studied extensively for several decades (see, for instance, [13, 14, 20] and the references therein). We assume that the exponent p(x,t) is bounded and satisfies the logarithmic continuity condition (see (3)), which is weaker than Hölder continuity. For related models of electrorheological fluids with nonstandard continuity assumptions on the exponent, we refer to [15]. Under suitable relations between q and the local bounds of p(x,t), we establish quantitative local boundedness estimates, extending the regularity theory for parabolic equations with variable exponent growth to the doubly nonlinear setting.
Notations
We provide some notations that will be used in the sequel, more precisely, for fixed (x0,t0)∈ΩT, and r,η>0 we define the following cylinders
and the respective spaces of functions with zero boundary data
W0(Ω):=W01,1(Ω)∩W(Ω),
L(0,T;W0(Ω)):=L1(0,T;W01,1(Ω))∩L(0,T;W(Ω)).
For any a,b∈R+ we also use the notation a∧b:=min(a,b), a∨b:=max(a,b).
Main Results
Let Ω be a bounded domain in RN, T>0, ΩT:=Ω×(0,T), we study non-negative (sub) super-solutions to the equation
(uq)t−divA(x,t,u,Du)=0,(x,t)∈ΩT.(1)
Throughout the paper we suppose that the functions A:ΩT×R+×RN→RN are Carathéodory functions1. We also assume that the following structure conditions are satisfied
where K1, K2 are positive constants and the function p(x,t):ΩT→R+ satisfies the so-called logarithmic condition: there exists a constant K3>0 such
∣p(x,t)−p(y,τ)∣≤log∣x−y∣+∣t−τ∣1K3,(3)
for any (x,t), (y,τ)∈ΩT such that ∣x−y∣+∣t−τ∣≤1/2. In what follows, the numbers {K1,K2,K3,q,p1,p2,N} are further referred to as the data and we will write γ as a generic positive constant that can be quantitatively determined a priori only in terms of the data if it depends on {K1,K2,K3,q,p1,p2,N}, and that can change from line to line. Postponing for the moment the formal definitions of weak solution, we will proceed to present the main results.
Theorem 1.1. Let u be a non-negative, local, weak sub-solution to (1.1)-(1.3) in ΩT. Fix (x0,t0)∈ΩT such that Qρ,θ(x0,t0)⊂Qρ,ρ(x0,t0)⊂Q8ρ,8ρ(x0,t0)⊂ΩT and let
q≤pρ,o−−1+(q+1)Npρ,o−,(4)
then u is bounded in Qρ,θ(x0,t0), provided that ρ is small enough. Particularly, if
q<pρ,o−−1+(q+1)Npρ,o−,(5)
then there exists C>0 depending only on the data such that if pρ,o+<q+1, then
We now deal with the singular case. In order to formulate our next result let us define the following number
κs:=spρ,o−+N(pρ,o−−q−1).(9)
Theorem 1.2. Let u be a non-negative, local, weak sub-solution to (1.1)-(1.3) in ΩT. Fix (x0,t0)∈ΩT such that Qρ,θ(x0,t0)⊂Qρ,ρ(x0,t0)⊂Q8ρ,8ρ(x0,t0)⊂ΩT, and assume that q+1>pρ,o+. Suppose that
q>pρ,o−−1+(q+1)Npρ,o−,(10)
and let s>q+1 be any number such that
κs>0,i.e. s>pρ,o−N(q+1−pρ,o−).(11)
If we assume additionally that
∬Qρ,θ(x0,t0)usdxdt<∞,(12)
then u is bounded in Qρ,θ(x0,t0). Moreover, the following quantitative estimate holds:
where C1>0 is a constant depending only on the data. The same conclusion holds if
q=pρ,o−−1+(q+1)Npρ,o−,(14)
provided that (1.11) and (1.12) are satisfied.
Preliminaries
In this section we give the definition of local weak solution to (1) in both variational and mollified weak formulations, so that their equivalence can be easily inferred. Then we prove, thanks to the mollified weak formulation, the energy estimates which are required for the De Giorgi iteration lying at the core of Theorems 1.1-1.2.
Notion of Solution
A function u is a non-negative, local, weak sub (super)-solution to Eq. (1) if
for any testing functions φ∈Wloc1,1+q(0,T;L1+q(E))∩Llocp2(0,T;W01,p2(E)), φ≥0.
Remark 2.1. In the case if p(x,t) is independent of t, i.e. p(x,t)≡p(x) we give more precise definition of the solution. We say that a function u is a non-negative, local, weak (sub) super-solution to Eq. (1) if
u∈Cloc(0,T;Lloc1+q(Ω))∩Lloc(0,T;Wloc(Ω)),(17)
and the integral inequalities (16) hold for any testing functions φ∈Wloc1,1+q(0,T;L1+q(E))∩Lloc(0,T;W0(E)), ζ≥0.
A function u that is both a non-negative, local weak sub and super-solution to (1) is a non-negative local weak solution.
Remark 2.2. The structure conditions (2) and the function spaces (15) and (17) guarantee the convergence of integrals in (16). For what concerns the existence of the solutions we refer the reader to [4].
Mollification in Time
The time derivative of a weak solution exists in the sense of distribution only. In order to overcome the lack of regularity in the time variable, we define the following mollification in time:
[v]h(x,t):=h1∫0tehτ−tv(x,τ)dτ,v∈L1(ΩT).
We refer the reader to [16] for various properties of this mollification.
Lemma 2.1. For any s≥1 we have
(i) If v∈Ls(ΩT), then [v]h∈Ls(ΩT). Moreover, ∥[v]h∥Ls(ΩT)≤∥v∥Ls(ΩT), and [v]h→v in Ls(ΩT) and almost everywhere on ΩT as h→0.
(ii) [v]h∈C(0,T;Ls(Ω)).
(iii) Almost everywhere on ΩT there hold
∂t∂[v]h=h1(v−[v]h).
(iv) If Dv∈Ls(ΩT), then D[v]h=[Dv]h→Dv in Ls(ΩT) and almost everywhere on ΩT as h→0.
(v) If v∈C(0,T;Ls(Ω)), then [v]h(⋅,t)→v(⋅,t) in Ls(Ω) and almost everywhere on Ω for every t∈(0,T) as h→0.
From the differential inequalities (16) we deduce the mollified version (cf. [16])
for any non-negative φ∈Lp2(0,T;W01,p2(Ω)) (φ∈L(0,T;W0(Ω)) if p(x,t)≡p(x)).
Remark 2.3. In the case q=1, spaces (17) can be also used for p(x,t) satisfying general conditions (3). In this case the Steklov averages in t and Friedrichs’ mollifiers in x can be used, see [1] for the details.
We briefly show how to infer the convergence of the mollified energy term. Under conditions (15), by the Young and Hölder inequalities we have
Integrating this inequality with respect to t over [0,T] and reversing the order of integration in the double integral and then integrating this inequality with respect to x, we arrive at
∬ΩT∣D[u]h∣p(x,t)dxdt≤∥Du∥Lp2(ΩT)+∣ΩT∣.
From this, by Lebesgue theorem and Lemma 2.1, we obtain
Lemma 2.2. Let u be a non-negative, local, weak sub (super)-solution to (1.1)-(1.3) in ΩT. Then there exists γ>0 depending only on the data, such that for every cylinder Qr,η(y,τ)⊂ΩT, any k>0 and any smooth ζ(x,t) which is zero for (x,t)∈∂Br(y)×(τ−η,τ) one has
Proof. We only consider the case of a local weak sub-solution, the case of super-solution can be considered completely similarly. For fixed τ−η<t1<t2<τ and sufficiently small ϵ∈(0,1) define
To estimate the diffusion term we pass to the limit h→0, use (2.5), (1.2) and the Young inequality. Passing to the limit ϵ→0 and combining the proceeding inequalities, we arrive at
To complete the proof of (2.7), we first pass in the last inequality to the limit t1→τ−η. Here, we discard the second integral on the left-hand side and take then the essential supremum with respect to t2∈(τ−η,τ). This leads to an estimate of the essential supremum of the first integral. On the other hand, discarding the first integral and passing to the limit t2→τ we deduce a similar estimate for the second integral of the left-hand side. Together, this leads to (2.7), which completes the proof of the lemma.
Auxiliary Lemmas
In this subsection, we collect several auxiliary results that will be used in the sequel.
The following lemma provides a useful algebraic estimate and can be found in [7 Lemma 3.2].
Lemma 2.3. Let q>0 and g± be defined in Lemma 2.2. There exists a constant γ=γ(q)>0 such that, for all a,b∈R, the following inequality holds true:
To estimate the second term on the right-hand side of the previous inequality, we note that the inequality (ρ(u−kˉj)+)p(x,t)≤(ρ(u−kˉj)+)pρ,o+ fails in the range 0<ρ(u−kˉj)+<1. Therefore, using (3.3), we proceed as follows:
By the log-continuity condition (1.3) on the exponent p(x,t), we can always assume that
pρ,o+<pρ,o−+(q+1)Npρ,o−.(30)
provided that ρ>0 is sufficiently small. Indeed, since pρ,o− depends on ρ, we use the fixed lower bound p1, which is independent of ρ, to choose ρ such that
We now verify that this smallness condition can be achieved by a suitable choice of k. Indeed, observing that
∬Qρ,θ(u−2k)+p∗dxdt≤∬Qρ,θ∩{u≥21k}up∗dxdt.
Since u∈Lp∗(Qρ,θ), by the absolute continuity of the Lebesgue integral we obtain
k→∞lim∬Qρ,θ∩{u≥21k}up∗dxdt=0,
we choose k large enough such that (3.13) holds, and hence
Q21ρ,21θsupu≤k,
this completes the proof of Theorem 1.1.
Proof of Theorem 1.2
Case q>pρ,o−−1+(q+1)Npˉρ,o
Since q+1>p∗=pρ,o−+(q+1)Npˉρ,o, we cannot proceed as in the proof of Theorem 1.1. Instead, we estimate the integral on the right-hand side of (3.7), taking into account that P=(q+1)∨pρ,o+=q+1. For s>q+1, set
ϵ:=s−p∗s−q−1∈(0,1)
and
d:=(∬Qρ,θ(x0,t0)usdxdt)s1.
By Hölder’s inequality, taking into account that 1−ϵq+1−ϵp∗=s, we obtain
Consequently, choosing k sufficiently large to satisfy both (27) and (41), we obtain
Q21ρ,21θ(x0,t0)supu≤k.
Since q+1>pρ,o+, we have H(θ,ρ)=θ−1, and hence the preceding estimate implies (13).
Case q=pρ,o−−1+(q+1)Npρ,o−
Set
yj:=∬Qj(u−kj)+p∗dxdt.
By (34) and Lemma 2.5, we have limj→∞yj=0, provided that
y0≤γ1[H(θ,ρ)]−1−pρ,o−N.(42)
where H(θ,ρ) is defined in (28). On the other hand, by (40), we have
y0≤ks−p∗2s−p∗ds.
Therefore, condition (42) is satisfied if k is chosen such that
ks−p∗≥γ[H(θ,ρ)]1+pρ,oN∬Qρ,θ(x0,t0)usdxdt.
Since p∗=q+1 in the present case, a direct computation gives
s−p∗=pρ,oχs.
Consequently, taking into account (27) and (28), we obtain the required estimate (13). This completes the proof of Theorem 1.2.
Acknowledgements
Simone Ciani acknowledges the partial funding of GNAMPA (INdAM), the Italian funding PNR (MUR) and the department of Mathematics of the University of Bolonga. Igor Skrypnik and Mariia Savchenko were supported by the National Research Foundation of Ukraine, project No. 2025.07/0369 ”Qualitative methods of nonlinear analysis of heterogeneous structures”.
Contact Information
Simone Ciani
University of Bologna Alma Mater, Department of Mathematics, Piazza Porta San Donato 5, 40126, Bologna. E-mail: [email protected]
Igor I. Skrypnik
Institute of Applied Mathematics and Mechanics of the National Academy of Sciences of Ukraine, Batiuka Str. 19, 84116, Sloviansk, Ukraine. E-mail: [email protected]
Mariia Savchenko
Institute of Applied Mathematics and Mechanics of the National Academy of Sciences of Ukraine, Batiuka Str. 19, 84116, Sloviansk, Ukraine. E-mail: [email protected]
Yevhen Zozulia
National University of Water and Environmental Engineering, Soborna Str. 11, 33028, Rivne, Ukraine. E-mail: [email protected]
References
[1]Yu. A. Alkhutov, V. V. Zhikov, Hölder continuity of solutions of parabolic equations with variable nonlinearity exponent, J. of Math. Sciences 179 (3), (2011), doi:10.1007/s10958-011-0599-9.DOI
[2]S. Antontsev, S. Shmarev, Parabolic equations with anisotropic nonstandard growth conditions, In: Free Boundary Problems: Theory and Applications, International Series of Numerical Mathematics, Vol. 154, Birkhäuser, Basel, (2007), 33–44.DOI
[3]S. Antontsev, S. Shmarev, Anisotropic parabolic equations with variable nonlinearity. Publ. Mat. 53(2), 355–399 (2009)DOI
[4]S. Antontsev, S. Shmarev, Existence and uniqueness for doubly nonlinear parabolic equations with nonstandard growth conditions, J. Diff. Equations and Appl., 4 (1), (2012), 67-94.DOI
[5]P. Baroni, V. Bögelein, Calderón–Zygmund estimates for parabolic p(x,t)-Laplacian systems, Rev. Mat. Iberoam. 30 (4), (2014), 1355–1386. doi:10.4171/RMI/817.DOI
[6]V. Bögelein, F. Duzaar, Hölder estimates for parabolic p(x,t)-Laplacian systems. Math. Ann. 354, 907–938 (2012).DOI
[7]V. Bögelein, F. Duzaar, U. Gianazza, N. Liao, C. Scheven, Hölder Continuity of the Gradient of Solutions to Doubly Non-Linear Parabolic Equations, Preprint, https://doi.org/10.48550/arXiv.2305.08539.
[8]E. DiBenedetto, (1993) Degenerate Parabolic Equations. Universitext, Springer-Verlag, New York, 1993.DOI
[10]L. Diening, P. Harjulehto, P. Hästö, M. Ruzicka, Lebesgue and Sobolev spaces with variable exponents. Lecture Notes in Mathematics, Springer Berlin, Heidelberg, 2011. (https://doi.org/10.1007/978-3-642-18363-8)
[13]A. S. Kalashnikov, On a nonlinear equation appearing in the theory of nonstationary filtration. Trud. Sem. I. G. Petrovski, 4, (1978), 137-146.
[14]S. Kalashnikov, Some problems of the qualitative theory of nonlinear degenerate second order parabolic equations. Russian Math. Surveys, 42 (1987) 169-222.DOI
[15]A. Kaltenbach, M. Ruzicka , Existence of steady solutions for a model for micropolar electrorheological fluid flows with not globally log-Hölder continuous shear exponent. Journal of Mathematical Fluid Mechanics, (2023), 25-40. DOI:10.1007/s00021-023-00782-yDOI
[16]J. Kinnunen, P. Lindqvist, Pointwise behaviour of semicontinuous supersolutions to a quasilinear parabolic equation, Ann. Mat. Pura Appl. (4) 185 (3) (2006), 411-435.DOI
[17]T.C. Halsey, W. Toor, Structure of Electrorheological Fluids, Phys. Rev. Lett. 65 (1990), 2820-2823.pubmed.ncbi.nlm.nih.gov/10042703
[22]Y. Wang, Intrinsic Harnack inequalities for parabolic equations with variable exponents, Nonl. Anal., 83, (2013), 12-30.DOI
[23]M. Xu, Y. Z. Chen, Hölder continuity of weak solutions for parabolic equations with non-standard growth conditions, Acta Math. Sin. Engl. Ser. 22 (3), (2006), 793-806.DOI
[24]F. Yao, Hölder regularity for the general parabolic p(x,t)-Laplacian equations, Nonlinear Differ. Equ. Appl. 22 (1), (2015), 105–119. doi:10.1007/s00030-014-0277-y.DOI