• 検索結果がありません。

FOR PERIODIC PARABOLIC SUBLINEAR PROBLEMS

N/A
N/A
Protected

Academic year: 2022

シェア "FOR PERIODIC PARABOLIC SUBLINEAR PROBLEMS"

Copied!
10
0
0

読み込み中.... (全文を見る)

全文

(1)

FOR PERIODIC PARABOLIC SUBLINEAR PROBLEMS

T. GODOY AND U. KAUFMANN Received 26 May 2003

We give necessary and sufficient conditions for the existence of positive solu- tions for sublinear Dirichlet periodic parabolic problemsLu=g(x, t, u) in×R (whereΩRNis a smooth bounded domain) for a wide class of Carath´eodory functionsg:Ω×R×[0,)Rsatisfying some integrability and positivity con- ditions.

1. Introduction

LetΩbe a smooth bounded domain inRN,N2. ForT >0, 1p≤ ∞, and 1q≤ ∞, letLp(Lq) be the Banach space ofT-periodic functions f onΩ×R (i.e., satisfying f(x, t)=f(x, t+T) a.e. (x, t)×R) such that

fLp(Lq):=f(·, t)Lq()

Lp(0,T)<. (1.1)

Similarly, letLTpbe the Banach space ofT-periodic functionsf such that f|×(0,T)

Lp(Ω×(0, T)), equipped with the norm fLpT:= f|×(0,T)Lp(Ω×(0,T)). Fi- nally, letCTbe the space of continuous andT-periodic functions on×Rpro- vided with theL-norm.

For the whole paper, we fixv, s(1,] such thatN/2v+ 1/s <1,s >2. Let {ai j} and {bj}, 1i, jN, be two families of functions satisfying ai j, bj LT and ai j=aj,i. Assume thatai j(x, t)ξiξjα0|ξ|2 for someα0>0 and all (x, t)×R,ξRN. LetA be theN×N matrix whosei, j entry is ai, j, let b=(b1, . . . , bN), let 0c0Ls(Lv), and letLbe the parabolic operator given by

Lu=utdiv(Au) + b,u+c0u. (1.2) LetW={uL2((0, T), H01(Ω)) :utL2((0, T), H1(Ω))}. Given fL1T,loc(Ω

×R), we say thatuis a (weak) solution of the Dirichlet periodic problemLu= f

Copyright©2003 Hindawi Publishing Corporation Abstract and Applied Analysis 2003:17 (2003) 975–984

2000 Mathematics Subject Classification: 35K20, 35P05, 35B10, 35B50 URL:http://dx.doi.org/10.1155/S1085337503309029

(2)

inΩ×R,u=0 on×R, ifuisT-periodic,u|×(0,T)W, and

×(0,T)

u∂h

∂t + Au,h+ b,uh+c0uh

=

×(0,T)f h (1.3) for allhCc (Ω×R) (and so for all hLT such thath|×(0,T)V0, where V0:=L2((0, T), H01(Ω))). For uW, the inequality Lu f (resp., ) will be understood in the same sense.

LetW= {uL2((0, T), H1(Ω)) :utL2((0, T), H1(Ω))}. Following [6], we say thatvis a supersolution of the above problem ifv|×(0,T)W, vtL2((0, T), H1(Ω)) +L1+η(Ω×(0, T)) forη >0 small enough,v|∂Ω×(0,T)0,v(·,0)v(·, T) a.e. inΩ, and

×(0,T)

v∂h

∂t + Av,h+ b,vh+c0vh

×(0,T)f h (1.4) for all 0hCc (Ω×(0, T)) (and so for all hLT such thath|×(0,T)V0

withV0as above). A subsolution is similarly defined by reversing the above in- equalities.

LetmLs(Lv) and let P(m) :=

T

0 ess supxm(x, t)dt (1.5) (with the value “+” allowed). For suchm(cf. [8, Theorem 3.6]),P(m)>0 is necessary and sufficient for the existence of a positive principal eigenvalue for the periodic parabolic Dirichlet problem with weight functionm(i.e., an eigenvalue with a positiveT-periodic eigenfunction associated to the problemLu=λmuin Ω×R,u=0 on×R). Moreover, this positive principal eigenvalue denoted byλ1(L, m) (orλ1(m)), if exists, is unique.

We are interested in the existence of positive solutions for the semilinear pe- riodic parabolic problem

Lu=g(x, t, u) inΩ×R, u=0 on×R,

uT-periodic,

(1.6)

wheregis a given function onΩ×R×[0,).

In [9, Theorem 3.7], it is proved that λ1

sup

ξ>0

g(·, ξ) ξ

<1< λ1

infξ>0

g(·, ξ) ξ

(1.7) is a necessary and sufficient condition for the existence of positive solutions in CT for (1.6) provided that g satisfies ξg(x, t, ξ)C1[0,), ξg(x, t, ξ)/ξ

(3)

nonincreasing in (0,), and some integrability and positivity conditions. In [10, Theorem 3.1], with the same monotonicity and regularity assumptions, and as- suming also some integrability conditions, it is proved that if either infξ>0(g(·, ξ)/ξ)Ls(Lv) andP(infξ>0(g(·, ξ)/ξ))0 or infξ>0(g(·, ξ)/ξ)0, then

λ1

sup

ξ>0

g(·, ξ) ξ

<1 (1.8)

is necessary and sufficient for the existence of a positive solutionuCTof (1.6).

Our aim in this paper is to prove, following a different approach, similar results without monotonicity and C1-regularity assumptions on g (see The- orems3.1,3.2, 3.3, and 3.4). Moreover, we will also cover some cases where limξ0+(g(·, ξ)/ξ)= ∞. These theorems will be obtained using the well-known sub- and supersolutions method combined with some facts concerning linear problems with weight.

In order to relate our results to others in the literature, we mention that, for the caseξg(·, ξ)/ξnonincreasing, similar results toTheorem 3.1for elliptic problems have been obtained, for example, in [4,5,13], assuming more regu- larity in the functiong. In the periodic parabolic case, there are also well-known results if ξg(·, ξ)/ξis concave and H¨older-continuous, and g(·,0)=0 (see [2,3,12] and the references therein).

On the other side, necessary and sufficient conditions for the existence of pos- itive solutions for equations of typeLu=a(x)ub(x)up, p >1,b0 (logistic equation), are also known (see, e.g., [11,12]). More general equations of the formLu=a(x)ub(x)f(x, u), withb0 and f superlinear, were studied, for example, in [7] for f Cµ,1+µ(Ω×[0,)), f strictly increasing, andb >0, and, for the Laplacian, the case f = f(u) is treated in [1] assuming f C([0,)).

Theorem 3.2 generalizes the aforementioned results, while Theorems3.3 and 3.4also extend some well-known results, see, for example, [2,3,11,12].

Some examples are also given at the end of the paper.

2. Preliminaries and auxiliary results

As usual, for ξ[0,) and u:Ω×R[0,), we writeg(ξ) and g(u) for the functions (x, t)g(x, t, ξ) and (x, t)g(x, t, u(x, t)), (x, t)×R. We as- sume, from now on, thatg:Ω×R×[0,)Ris a Carath´eodory function (i.e., (x, t)g(x, t, ξ) is measurable for allξ[0,), andξg(x, t, ξ) is continuous in [0,) a.e. (x, t)×R) such that supσξ(g(σ)/σ) and inf0<σξ(g(σ)/σ) are measurable functions for allξ >0, and infξ>0(g(ξ)/ξ)=supξ>0(g(ξ)/ξ), that is, (1.6) is not a linear problem.

We start recalling some facts about periodic parabolic problems with weight.

Remark 2.1. (a) LetD= {mLs(Lv) :P(m)>0}. ThenDis open inLs(Lv) and the mapmλ1(m) is continuous fromDintoR(cf. [8, Theorem 3.9]). Also,

(4)

the following comparison principle holds: ifm1, m2Ls(Lv) andm1m2 in Ω×R, thenλ1(m1)λ1(m2); and if, in addition,m1< m2 in a set of positive measure, thenλ1(m1)> λ1(m2) (cf. [8, Remark 3.7]).

(b) ForλRandmLs(Lv), letµm(λ) be defined as the uniqueµRsuch that the Dirichlet periodic problemLu=λmu+µm(λ)u in Ω×Rhas a posi- tive solutionu. We recall thatµm(λ) is well defined and that the map (λ, m) µm(λ) is continuous fromR×Ls(Lv) intoR(cf. [9, Proposition 2.7]). Moreover, µm(0)>0,µmis concave and continuous, and a givenλRis a principal eigen- value associated to the weightmif and only ifµm(λ)=0 (cf. [8, Lemma 3.2]).

Also, ifλ1(m) exists, then forλ >0,µm(λ)>0 if and only ifλ < λ1(m), and if λ1(m) does not exist,µm(λ)>0 for allλ >0.

(c) LetmLs(Lv) such thatP(m)>0 and letmjbe a sequence such thatmj

converges tominLs(Lv). Then it follows from [9, Remark 2.5] thatP(mj)>0 forjlarge enough.

Remark 2.2. IfuLT is a positive solution of (1.6) and

0<ξinfM

g(ξ) ξ

LsLv, sup

0<ξM

g(ξ) ξ

LsLv,

(2.1)

for allM >0, thenuCT andu(x, t)>0 for all (x, t)×R. Indeed, this fol- lows from [9, Remark 2.2 and Corollary 2.12].

We introduce some additional notation. For (x, t, ξ)×R×(0,), let

g(x, t, ξ)=ξ sup

0<ξσ

g(x, t, σ) σ

,

g(x, t, ξ)=ξ inf

0<σξ

g(x, t, σ) σ

(2.2)

(with the values “±∞” allowed). It is easy to check that ifg(ξ) is finite forξξ0, thenξg(ξ) is continuous in (0, ξ0) a.e. inΩ×R, and that ifg(ξ) is finite for ξ0ξ, thenξg(ξ) is continuous in (ξ0,) a.e. inΩ×R. We also set

m(x, t)=inf

ξ>0

g(x, t, ξ) ξ

, m0(x, t)=sup

ξ>0

g(x, t, ξ) ξ

,

m0(x, t)=lim inf

ξ0+

g(x, t, ξ) ξ

, m(x, t)=lim sup

ξ→∞

g(x, t, ξ) ξ

.

(2.3)

(5)

Note that

m=lim

ξ→∞

g(ξ) ξ

, m0=lim

ξ0+

g(ξ) ξ

, m0=lim

ξ0+

g(ξ) ξ

, m=lim

ξ→∞

g(ξ) ξ

.

(2.4)

Lemma2.3. Let ξ0>0. Assume thatg(ξ)Ls(Lv)for allξξ0 and that either mLs(Lv)withλ1(m)>1(ifλ1(m)exists) orm0. Then, for allc >0, there exists a supersolutionwCT of (1.6) such thatwc.

Proof. We first study the casemLs(Lv). Letc >0. We claim that there exists ξcsuch thatµg(ξ)/ξ(1)>0. Indeed, forξξ0, we havemg(ξ)/ξg0)/ξ0

and also limξ→∞(g(ξ)/ξ)=mwith convergence a.e. Thus, by dominated con- vergence, limξ→∞(g(ξ)/ξ)=m in Ls(Lv) and then Remark 2.1(b) implies limξ→∞µg(ξ)/ξ(λ)=µm(λ) for allλ. Moreover, either ifP(m)>0 andλ1(m)>

1 or if P(m)0, the last statement in Remark 2.1(b) also givesµm(1)>0.

Thus, it follows thatµg(ξ)/ξ(1)>0 forξlarge enough.

We fixξmax(ξ0, c) such thatµg(ξ)/ξ(1)>0. Letkbe a function defined byk(x, t)=supξξ|g(ξ)/ξ|. Sincemkg)/ξ, we getkLs(Lv). For ξ[0,), letg(x, t, ξ)=g(x, t, ξ) +k(x, t)ξ. Theng(x, t, ξ)0 andg(ξ)/ξ Ls(Lv) forξξ. Also,µL+λk,g)/ξ(λ)=µL,g(ξ)/ξ(λ) for allλ. In particular, µL+k,g)/ξ(1)=µL,g(ξ)/ξ(1)>0. Thus, Lemma 2.9 in [9] says that the Dirichlet periodic problem (L+kg)/ξ=g) inΩ×Rhas a solutionΦ CTsatisfyingΦ(x, t)>0 a.e. (x, t)×R. Now,

gξgξ

gξ ξ

ξ

gξ++gξ ξ Φ

=gξ+gξ ξ Φ

=Lξ,

(2.5)

and thereforeξ+Φis a supersolution for (1.6).

Consider now the casem0. In this case, we have limξ→∞(g+(ξ)/ξ)=0 a.e.

inΩ×R, where, as usual, we write f =f+f. Also, 0g+(ξ)/ξg+0)/ξ0

for allξξ0, and thus limξ→∞(g+(ξ)/ξ)=0 inLs(Lv). So, limξ→∞µg+(ξ)/ξ(λ)=λ1

for allλ, whereλ1is the (positive) principal eigenvalue forLassociated to the weight 1 (because for m1, µmλ1). Thus, we can choose ξmax(ξ0, c) such thatµg+)/ξ>0, and then, as above, the Dirichlet periodic problem (L g+)/ξ=g+) inΩ×Rhas a solutionΦCT satisfyingΦ(x, t)>0 a.e.

(6)

(x, t) inΩ×R. Also,

gξg+ξ

g+ξ ξ

ξ

=g+ξ+g+ξ ξ Φ

=LΦ+ξ,

(2.6)

and this concludes the proof.

Lemma2.4. Letξ0>0. Assume thatg0)Ls(Lv),P(g0)/ξ0)>0, andλ1(g(ξ0)/

ξ0)1. Then there exists a subsolutionvCT of (1.6) such thatv(x, t)>0for all (x, t)×R.

Proof. LetΦbe the positive eigenfunction of

L+gξ0

ξ0

Φ=λ1

g+ξ0

ξ0

g+ξ0

ξ0

Φ inΩ×R, Φ=0 on∂Ω×R,

ΦT-periodic.

(2.7)

ThenΦCT andΦ(x, t)>0 for all (x, t)×R. Now,λ1(L, g(ξ0)/ξ0)<1 im- plies µL,g(ξ0)/ξ0(1)0. Thus, since µL,g(ξ0)/ξ0(1)=µL+g0)/ξ0,g+0)/ξ0(1), we get λ1(g+0)/ξ0)1.

Letε >0 be such thatε < ξ0/Φ. Taking into account the above-mentioned facts and thatξg(ξ)/ξis nonincreasing, we have

L(εΦ) +g(εΦ) L+gεΦ εΦ

εΦ

L+gξ0

ξ0

εΦ

g+ξ0

ξ0

εΦ

g+εΦ εΦ

εΦ

g+(εΦ),

(2.8)

and the lemma follows.

(7)

3. The main results

Theorem3.1. (a)Assume that

(1)m0, mLs(Lv),P(m0)>0, andP(m)>0,

(2)g(ξ0)Ls(Lv)for someξ0>0andg(ξ1)Ls(Lv)for someξ1>0.

Then, ifλ1(m0)<1< λ1(m), there exists a solutionuLT of (1.6) satisfying u(x, t)>0for all(x, t)×R.

(b)Assume (1),m0=m0,m=m, and that for allξ >0, m0=g(ξ)

ξ , (3.1)

m=g(ξ)

ξ . (3.2)

Then there exists a positive solutionuLT of (1.6) if and only ifλ1(m0)<1<

λ1(m).

Proof. Suppose thatλ1(m0)<1< λ1(m). Since, for 0< ξξ1, we haveg1)/ξ1

g(ξ)/ξm0and limξ0+g(ξ)/ξ=m0 a.e. inΩ×R, taking into account (1) and (2), we getg(ξ)/ξLs(Lv) for suchξand so limξ0+g(ξ)/ξ=m0with con- vergence inLs(Lv). Then, byRemark 2.1(c), we have limξ0+P(g(ξ)/ξ)=P(m0)

>0, and thus there existsλ1(g(ξ)/ξ) forξ >0 small enough. Moreover,Remark 2.1(a) says that limξ0+λ1(g(ξ)/ξ)=λ1(m0)<1 and soλ1(g(ξ)/ξ)<1 for such ξ. Hence,Lemma 2.4can be applied to give a subsolutionvCT of (1.6) with v(x, t)>0 for all (x, t)×R.

On the other hand, for allξξ0, we havemg(ξ)/ξg(ξ0)/ξ0, and so g(ξ)/ξLs(Lv). Therefore, takingc= vinLemma 2.3, we obtain a super- solutionwCT of (1.6) withwcv. Now, [6, Theorem 1] gives a solution uLT such thatvuwand thenu(x, t)>0 for all (x, t)×R. Thus (a) is proved.

To prove (b), suppose thatuLT is a positive solution of (1.6). ByRemark 2.2, we have u(x, t)>0 for all (x, t). Letmu:Ω×RRbe defined bymu= g(u)/u. Sincemuis measurable andmmum0, it follows thatmuLs(Lv).

Moreover, we haveLu=muuand so 1=λ1(mu). Now, the comparison principle in Remark 2.1(a) gives 1=λ1(mu)λ1(m0)=λ1(m0) and also 1λ1(m)= λ1(m). Supposeλ1(m0)=1. Since λ1(mu)=1 and mum0, we must have mu(x, t)=m0(x, t) a.e. (x, t)×R(seeRemark 2.1(a)), but sup0<ξu(g(ξ)/

ξ)g(u)/u=m0inΩ×Rcontradicting (3.1). Thenλ1(m0)<1. Suppose now thatλ1(m)=1. Reasoning as above, we get 1=λ1(mu)λ1(m)=1 and so mu=m. Thus, inf0<ξu(g(ξ)/ξ)g(u)/u=infξ>0(g(ξ)/ξ) a.e., which is

again a contradiction. Thenλ1(m)>1.

Theorem3.2. (a)Assume that (3)m0Ls(Lv),P(m0)>0,

(8)

(4)g(ξ0)Ls(Lv)for someξ0>0andg(ξ)Ls(Lv)for allξ >0, (5)eithermLs(Lv)andP(m)0orm0.

Then, ifλ1(m0)<1, there exists a solutionuLT of (1.6) satisfyingu(x, t)>0 for all(x, t)×R.

(b)Assume, in addition, (3.1) andm0=m0. Then there exists a positive solution uLT of (1.6) if and only ifλ1(m0)<1.

Proof. As in the above theorem, we haveg(ξ)/ξLs(Lv) andλ1(g(ξ)/ξ)<1 for ξ >0 small enough, and so Lemma 2.4 gives a subsolution vCT satisfying v(x, t)>0 for all (x, t). On the other side, since g(ξ)/ξg(ξ)/ξg(ξ0)/ξ0 for ξξ0, from (4), we haveg(ξ)/ξLs(Lv) for suchξ. Therefore, (a) follows as in Theorem 3.1takingc= vinLemma 2.3, and the proof of (b) follows simi-

larly to part (b) ofTheorem 3.1.

Theorem3.3. (a)Assume (2) and that (6)mLs(Lv)andP(m)>0,

(7)P(g(ξ)/ξ)>0forξ >0small andlimξ0+λ1(g(ξ)/ξ)=0.

Then, ifλ1(m)>1, there exists a solutionuLT of (1.6) satisfyingu(x, t)>0 for all(x, t)×R.

(b)Assume, in addition, (3.2) andm=m. Then there exists a positive solu- tionuLT of (1.6) if and only ifλ1(m)>1.

Proof. Reasoning as above, (a) follows from Lemmas2.3,2.4, and [6, Theorem 1]. Suppose now thatuLT is a positive solution of (1.6). Letε >0 such that ε <u. Letgεbe defined bygε(ξ)=g(ξ) ifξεandgε(ξ)=g(ε) ifξ < ε. We haveLu=g(u)g(u)gε(u) and alsogε(u)/uLs(Lv). Thus, 1λ1(gε(u)/u).

Moreover, sinceg

ε(u)/um, the comparison principle inRemark 2.1(a) gives 1λ1(m). Suppose 1=λ1(m). Theng

ε(u)/u=m. Butg

ε(u)/ug

ε(u)/

u =g(u)/u, and thereforem=g(u)/uin contradiction with (3.2).

Theorem3.4. Assume (4), (5), and (7). Then (1.6) has a positive solutionuLT satisfyingu(x, t)>0for all(x, t)×R.

Proof. The theorem follows again from Lemmas2.3,2.4, and [6, Theorem 1].

3.1. Examples. (a) Suppose there exist limξ0+(g(ξ)/ξ) and limξ→∞(g(ξ)/ξ) and assume infξ>0(g(ξ)/ξ),supξ>0(g(ξ)/ξ)Ls(Lv), with P(infξ>0(g(ξ)/ξ))>0. If limξ0+(g(ξ)/ξ)=supξ>0(g(ξ)/ξ) and limξ→∞(g(ξ)/ξ)=infξ>0(g(ξ)/ξ), from Theorem 3.1, we conclude that (1.6) has a positive solutionuLT if and only if λ1(limξ0+(g(ξ)/ξ))<1< λ1(limξ→∞(g(ξ)/ξ)).

(b) Consider the Dirichlet periodic problemLu=sinuinΩ×R.Theorem 3.2 says that this problem has a positiveT-periodic solution if and only ifλ1<1, whereλ1is the positive principal eigenvalue corresponding to the weight 1.

(9)

(c1) Consider the problem

Lu=a(x, t)uγf(x, t, u)u inΩ×R, u=0 on∂Ω×R,

uT-periodic,

(3.3)

where 0< γ1 and f is a Carath´eodory function such thatf(ξ)Ls(Lv) for all ξ >0 and f(0)=0. Assume thatγ=1,aLs(Lv),P(a)>0,alimξ→∞f(ξ)

, infξ0ξ f(ξ)Ls(Lv) for someξ0>0, and inf0<ξξ0f(ξ)Ls(Lv) for allξ0>0.

FromTheorem 3.2, it follows that (3.3) has a positive solutionuLT if and only ifλ1(a)<1.

(c2) Consider now the case 0< γ <1 anda(x, t)0 a.e. (x, t)×R. If f(ξ)= −bwithbLs(Lv) andP(b)>0, thenTheorem 3.3says that (3.3) has a positive solutionuLT if and only if 1< λ1(b). On the other hand, suppose limξ→∞f(ξ)= ∞, infξ0ξ f(ξ)Ls(Lv) for some ξ0>0, and sup0<ξξ0f(ξ) Ls(Lv) for allξ0>0. ThenTheorem 3.4gives a positive solutionuLT for (3.3).

We note that in all the cases, the positive solutionusatisfiesu(x, t)>0 for all (x, t). Moreover, recallingRemark 2.2, we also have that in (a), (b), and (c1) uCT.

Remark 3.5. An inspection of the proofs shows that all the above results remain true for the corresponding elliptic problem, replacingLs(Lv) byLr(Ω) withr >

N/2, andP(m) by ess supxm(x).

Acknowledgment

This work was partially supported by CONICET, Secyt-UNC, and Agencia C ´ordoba Ciencia.

References

[1] S. Alama and G. Tarantello,On the solvability of a semilinear elliptic equation via an associated eigenvalue problem, Math. Z.221(1996), no. 3, 467–493.

[2] H. Amann,Existence and multiplicity theorems for semi-linear elliptic boundary value problems, Math. Z.150(1976), no. 3, 281–295.

[3] ,Periodic solutions of semilinear parabolic equations, Nonlinear Analysis (Col- lection of Papers in Honor of Erich H. Rothe) (L. Cesari, R. Kannan, and H. F.

Weinberger, eds.), Academic Press, New York, 1978, pp. 1–29.

[4] H. Brezis and L. Oswald,Remarks on sublinear elliptic equations, Nonlinear Anal.10 (1986), no. 1, 55–64.

[5] D. G. De Figueiredo,Positive solutions of semilinear elliptic problems, Differential Equations (Sao Paulo, 1981), Lecture Notes in Math., vol. 957, Springer, Berlin, 1982, pp. 34–87.

[6] J. Deuel and P. Hess,Nonlinear parabolic boundary value problems with upper and lower solutions, Israel J. Math.29(1978), no. 1, 92–104.

(10)

[7] J. M. Fraile, P. Koch Medina, J. L ´opez-G ´omez, and S. Merino,Elliptic eigenvalue prob- lems and unbounded continua of positive solutions of a semilinear elliptic equation, J. Differential Equations127(1996), no. 1, 295–319.

[8] T. Godoy and U. Kaufmann,On principal eigenvalues for periodic parabolic problems with optimal condition on the weight function, J. Math. Anal. Appl.262(2001), no. 1, 208–220.

[9] ,On positive solutions for some semilinear periodic parabolic eigenvalue prob- lems, J. Math. Anal. Appl.277(2003), no. 1, 164–179.

[10] T. Godoy, U. Kaufmann, and S. Paczka,Positive solutions for sublinear periodic para- bolic problems, Nonlinear Anal.55(2003), no. 1-2, 73–82.

[11] J. Hern´andez,Positive solutions for the logistic equation with unbounded weights, Re- action Diffusion Systems (Trieste, 1995), Lecture Notes in Pure and Appl. Math., vol. 194, Dekker, New York, 1998, pp. 183–197.

[12] P. Hess,Periodic-Parabolic Boundary Value Problems and Positivity, Pitman Research Notes in Mathematics Series, vol. 247, Longman Scientific & Technical, Harlow, 1991, copublished in the United States with John Wiley & Sons, New York.

[13] K. Taira and K. Umezu,Positive solutions of sublinear elliptic boundary value problems, Nonlinear Anal.29(1997), no. 7, 761–771.

T. Godoy: Facultad de Matem´atica, Astronom´ıa y F´ısica, Universidad Nacional de C ´ordoba, Ciudad Universitaria, 5000 C ´ordoba, Argentina

E-mail address:[email protected]

U. Kaufmann: Facultad de Matem´atica, Astronom´ıa y F´ısica, Universidad Nacional de C ´ordoba, Ciudad Universitaria, 5000 C ´ordoba, Argentina

E-mail address:[email protected]

参照

関連したドキュメント

Later, the problems with non- local integral conditions for parabolic equations were investigated by Kamynin [10], Ionkin [9], Yurchuk [18], Bouziani [2]; problems for

Since Liouville theorems are closely related to the scaling method for elliptic and parabolic problems, our counter=example indicates that one has to impose additional assumptions