You are on page 1of 20
el al ht a RR AR A BOP ES SEES Advanced Nonlinear Studies 7 (2007), 300-200 Soliton Solutions to a Class of Quasilinear Elliptic Equations on R M.J. Alves, P.C. Carriao * Departamento de Matemética Universidade Federal de Minas Gerais 1270-901 Belo Horizonte-MC, Brasit ‘email: mariajosetimat.ufng.br e-mail: carrionfimat.ufng.br O.H. Miyagaki f Departamento de Motemétion Universidade Federal de Vigosa a .96571-000 Vicosa-MG, Brazil olimpio®ufi.br Received 18 April 2007 Communicated by Antonio Ambrosetti Abstract "This paper is concerned with the existence of positive solutions for a class of ‘quasilinear elliptic equations on R. The results are proved by combining the ‘concentration-compactness principle due to Lions with a minimization approach. 1991 Mathematics Subject Classification. 36110, 35320, 35160, 58137, S30. Key words. Soliton slution, Scrcinger equation, p-Laplacas, variational wethod ‘Supported in part by CNP Brasil. ‘Supported in part by CNPa Brasil sax! AGIMB-—Mileniuen Intitute MCT/Brasi 2 MJ. Alves, P.C. Carrio, O.FL. Miyagaki 1 Introduction We study the following class of quasilinear eliptic problems on R: { La -V (2) 20 jai? Pu, iv Ry ue WI9(R),u> Oi R, aay where the operator Lis defined by = Kol{(lul?)'P2((uP)'¥ ala, us [wp] Ko>0,8>1, p>1,q2>pP, (a> p), andV :R > Ris a given potential function verifying the basic condition (Wo) inf V(z) > 0. Such problems are motivated in mathematical physics; for instance, when we search certain kinds of solitary waves in nonlinear equations of the Klein-Gordon oo Schrddinger types. These questions can be reduce to problens of existence of solutions for problem (1.1). (see e. {12, 18). When Ko = 0 ad p= 2, probless (U1) in R" was studied by voveral authors For example, Rabinowitz in [19] (se also [7,3]) treated the case when V satisfies some coercivity condition such x w“ im, V(2) = +20. Another situation, namely, when Vis bounded and satisfies the periodicity condition “a V(e-+p) =V(eh 2€R, pez, wns studied, ¢. g., by Coti-Zelati and Rabinowitz [8), Kryszewski and S; and Monteechiari (16). _ ‘On the other hand, when V is asymptotic to a constant V = sup V(z), that is, zh akin [11] (Vy) V, and V n. In this paper, by employing some techniques developed by Ambrosetti and Wang in [2], we generalize problem (1.2), considering @ = 1, to the p-Laplacian operator in R and we also consider more general nonlinearities, given by exponents 8 and 4. Note that this ease is not covered, for instance, in [20] and is the main dificulty of our work, becanse the ideas of (20) do not apply in a straighforword way. We remark that the Brezis and Lieb identity appearing in [5] does not hold in our ease, even in the case p — 2, as it was observed in [18] and {2}. Also, for psf 2 we need to prove the convergence of the sequence of the derivatives, that is, we have to prove that ul ou! ae. in R, asco, for bounded sequence tty, in Sobolev spneesi. ‘This convergence was obtained by Boceardo and Murat in [4] for p > n. Our main result is the following: ‘Theorem 1.1 Let (Vo) hold, Suppove that V sutificw one of the conditions (Vi), (Vs) or (V3). If B > 1, p> Landq> pA (q>p) then problem (1.1) has at least ine positive solution. 2 Preliminary Results Set {u wt: [ Vie)iuPar < too} ‘endowed with the norm given by 4 M.J. Alves, P.C. Carriéo, O.H. Miyagaki +V(z)|ul"|ar. ‘Thus if (Yo) holds, then X is continuously embedded in the space W1?(R). We define the energy functional I: X > R by ) ii {i [lv’P + V(2)|uP?} It is easy to see that I € C1(X,R) and its Fréchet derivative is given by Tu)-v = f [lu' Pu!) oda +[v@ [lua] vdz +o f ¥(p—1) [lump] ods ke fo [lure Pe 2a] vttx -f {lul?-2u] wade and hence . T'(u)-u = jul? + BP Kolu’u 42 — |ulg, we X, where |ul? = fy |u(z)|Pde denotes the usual norm in L(R) We define the Nehari manifold by setting N= {ue X\(0}:1"(u)-u=0}- Ie is well known that the critied points of functional I restricted to the Nehari manifold NV are precisely the weak solutions of problem (1.1). ‘To prove Theorem 1.1 we will neod some prelitsinary rosulls Chat are described by several lemmata established below. Their proofs will be postponed to section 4, Lemma 2.1 Suppose 8 > 1, p> landq> p 8 (q>p). Then (0) N is a manifold and N #0. (b) For any u € X\{0} with I'(u)-u <0, there exists a unique 0 < <1 such that MEN. (0) there exists p > 0 such that |lul] > p for every u € N. (d) If i € N is a constrained critical point of I on N, then I'(ii) = 0 and a is a solution of problem (1.1). (0) m= infyen I(u) > C >0, where C is a constant. Lemma 2.2 Let (um) CN be a minimizing sequence for I (that is, I(tn) —+ m as n+ 00). Then Soliton solutions to a class of elliptic equations 5 (a) ty is bounded in X; thus there exists u € X such that uy, —+ u weakly in X, as n+ 00, (6) Titp+oo |ttn|$ > 0, (0) Let 0 = un —u. Then I'(u)- 0+ iminta-soa E(t) “tn <0, 80 that either I'(u) +4 $0 or liminfn soo I'(0n) +t <0. Lemma 2.3 Let (un) C N be a minimizing sequence for I such that u, — u weakly in X, as n-+00. If u€ N, then I(u) =m. 3 Proof of Theorem 1.1 Let (un) C N be a minimizing sequence for J. From Lemma 2.2-(a), Un is bounded in X and therefore u, — u weakly in X, as mn — oo. We are going to prove Theorem 1.1 by considering several cases of the potential V. Case I (Vj; assumed). From (Vq) and (V;) we have that X is contimuously and compactly embedded into the space L**1(R), for all s > p (see [3]). Thus tn —> win L4(R), as n+ oo. Using Lemma 2.2-(b) it follows that u # 0. We are going to prove that lim, 20 [|Un||” = 0. Notice that it is sufficient to get Uy > u in X, as n + 00. Suppose by contradiction that limy—so0 [|Un |? + 1 > 0. We have, up to a subsequence, T' (en) On = en? + PKaleien 1B beni lel? Loni ‘Taking the lim inf,,-4o0 and using the fact that Tita. [nlf = 0, we conclude that Timinf I’(vn) -Un > 0. From Lemma 2.2-(c) we have that 1’(u)-u <0. Since u ¥ 0, by Lemma 2.1-(b) there exists 0 00. (3.2) ‘Then from (3.1) and (3.2) we obtain m+ o(1) T(un) = Phat +B PB PB pqs a oe & A-PIAull? +. oh [Aulg + 0(1). +o(1) 6 M.J. Alves, P.C. Carrio, O.H. Miyagaki Since 0 < A < 1, it follows that m> Spar + aa 98 pap [ulg = TAu)|n, which a contradiction because Au € N. Hence limy-s.o ||Unl|” = 0. Case IT (Vz assumed). From a result. due to Lions [13] (see also (23, Lemma 1.21]) we have that tu, —+ 0 in L4(R), as n + co, and this is a contradiction because u satisfies the Lemma 2.2-(b). Therefore we infer that there exist 7,r > 0 and Yn €R such that yotr Yimint una} Pda > vy > 0. (3.3) nr ‘We may assume that y, are integer spultiples of the period p of V, if necessary; just take a bigger constant r in (3.3). Define ty (2) = u(x + yn). Since V is periodic, then (W,) is also a minimizing sequence, which is bounded in X and it satisfies (3.3) for y» = 0. This implies that Uy — U weakly in X, as n— 00. Therefore u #0. Now we will prove that I'(u)-u=0 and the result follows from Lemma 2.3 and Lemma 2.1-(d). Otherwise suppose, by contradiction, that I’(u)-u # 0. If I'(u) - u <0, from Lemma 2.1-(b) there exists 0 << 1 such that Au € N. Since (up) is a minimizing sequence, arguing as before we reach a contradiction. If J'(u)-u > 0, from Lemma 2.2-(c) we conclude that lim inf, soo L’(Upn)- Up < 0. Thus, up to a subsequence, limyoo I'(vn) - U <0. Then for n sufficiently large, from Lemma 2.1-(b) there existe 0 < Ay < 1 wich Ghat Anu, € N. Furthermore one has that limsup,,_,.. An < 1. In fact, if limsup,,,..n = 1, taking a subsequence such that, \y + 1 as n + 00, we conclude that T'(&n) + On = T'(An®n)*An®a + 0(1), a8 n+ 00. Thus limpco I'(vp) + Un = 0. But this is a contradiction because limysoo I’(un) «tn < 0. Therefore lim supy.o. An < 1. As before, we find m+ of1) > PEN PL ntnl? + et S—PE Pl + (0) Since limsup,,s9 An < 1, it follows that rm > Fag? + OE lg = Tt) which a contradiction because yt, € N. Hence I'(1) -u Soliton solutions to a class of elliptic equations 7 Case IIT (Va assumed). Set 09 = f {Eis vycerary + 2 opt — Lay} a, ‘We use the notation mp, Np, I, and , for this problem with obvious meanings. From Case IT we know that J, has a critical point 1 € Np where Np= {ue X\ OF: Glu) Oo}. It will be proved that @ > 0 on R. . If V is not constant, from (Vs) we conchide that I'(i)-t# < Ip(t) t= 0. Hence from Lemma 2.1-(b) there exists a unique 0 < Au <1 such that Aut € N, On the other hand, from (V5) we have Tra) = af (la'P + V(2)ol") do + aay < Fhe [iw + yoourlass Sing Hence: m < I(Agit) < mp (3.4) Let (uj) be a minimizing sequence for I. Then tun —* u weakly in X, as n + 00. We are going to prove that u # 0 and J'(u)-u = 0 and the result follows from Lemma 2.3 and Lemma 2.1-(4) Suppose that u = 0. From (V3), given € > 0, there exists R > 0 such that [V(@) - Vol2)| Se if |Z] > R, (3.5) and [V(z)—Vpl < 21Vpl if lz] < R. (3.6) Since u, — u=0 weakly in X, asn —+ 00, then |um|2 + 0, asn — co, ina compact: part of R. One obtains link? < z if |z| < R, for some R>0. (3.7) From (3.5), (3.6) and (3.7) we have Hun) —Iylin) =f 1G) VolluaPae 8 MJ. Alves, P.C. Carriéo, O-H. Miyagaki 2M « SS] tualde + = Paz ? discal! 5 den 4MRe Tt glial? for some M > 0. Hence, Timm, Ip(ttn) = lim, 1( 2 (38) Similarly, we also have He, Fy (ttn) ten = et Teen) «ton (3.9) Thus there exists a wubscqquence (11,) abd Jy > Ooch that Jnatins = Antin € Np. We will prove that 2, is bounded. Notice that 0 = Aelund” + BARE K luiul"p — As ltnl§ + 0(1)- It follows that Ua? = NOY [XEPP eg — BPE Np) + 9(0)- (3.10) Since |Jup||? and |ul,u2—1]2 are bounded, using Lemma 2.2-(b), we deduce that Xn is bounded. 7 On the other hand 3,, does not converge to zero, as n —+ 00; otherwise, O 0, and observing that uy N, we have that TpQAntin) = Tntin) + 0(1) T(x) + o(1) = m+ 0(1), ™p win where in the first inequality we used the fact that, is bounded. But this inequality ig in contradiction with (3.4). Hence u # 0. Set 0}, = uy —u. Now we are going to prove that 9 > liminf Fy(on) va + ¥(u)-u. (3.11) In fact, from Lemma 2.2-(c) we conclude that Soliton solutions to a class of elliptic equations ° 0 > limine (vu) vm + F(a) = tnint [fo -Yoepivaae + J 1e60P+ Volo da Heh" — ble] +10) = taat [oom tf (V(@)—VotedioaPate] +10) By (Vs) and applying Arzel4-Ascoli’s Theorem we obtain [fc voi Pae = of), as nr and by combining this result with the previous inequality we obtain the desired inequality (3.11) If 1'(u)-u <0, from Lemna 2.1-(b) there exists 0 < A< 1 much that Au € N. Since (un) is a minimizing sequence and u 4 0, arguing as Case T, we reach a contradiction. If 1'(u) - u > 0, from (3.11) we conclude that liminfy,4If.(Un) tm <0. Thus, up to a subsequence, limyas Iy(tm) Ya < 0. Now we are going to proceed as in the proof of Case IT. Then for n sulficiently large, from Lemma 2.1-(b) there exists 0 < Ay < Latch that Ay, € Np. Furthermore one has that limsupn son An <1 Otherwise if limsupy sag An = 1, along a subsequence, dy > 1 as n > 00 and we conclude that Fyn) - Ya = F(Antin) “Ant + 01). Thus lim, I,(vn) 04 = 0. But this is a contradiction. Therefore Timsup An <1 Notice that 7B AG? ([Anvnl? + Vo(@)[Awenl?] da + [vr tre PF naval + (0) Since lim supy-soo 4n <1, by (V3) it follows that —¥,(z)|Anvnl?] dx PHN) (Pt IPP enld = [yn > a [Anal + me [Antinlg = LpAntn)|vps 10 MJ. Alves, P.C. Carriio, O.H. Miyagski which is a contradiction because Anti € Np. Hence I(u).u = 0. ‘Thus we have proved that problem (1.1) possesses a solution 1 #0 if one of the conditions Vj, Vz or Vs holds. Tt remains to show that u > 0. We have that (|u|) is also a minimizing vequence for I, because m = Mttx) = H lt). ‘Thess « >'0. We are going to prove the vity of u. Notice that equation (1.1) ean be written in the form — [2 +27 (Wu? 2u)?] (Iu 2a’) —V(z) lulu + jul? Pu + (p— 1)(8 - 1)" (Chan ae Kou Pu")? If there exists some zo € R such that u(zo) = 0, since u > 0, one has u'(z9) = 0. ‘Then from the above equality and by the tniqueness of the solution of the Cauchy problem, we have that u = 0, which is a contradiction because u € X ~ {0}. Hence u>d. 4 Proofs of lemmata Proof of Lemma 2.1. Proof of (a): We will prove that Nis in fact « manifold, For u€ None has BP Kolu’u® "ip = |ulg — ll”. @) Tt follows that T"(u).uu = p( — B)llul? + (8 — @)|ulg. But p> 1, > 1 and q > pi; consequently I"(u)-u-u <0, (42) so that NV is a manifold. ‘To prove that N 70. We fix u € X\{0} and define Y(t) =Htu) = Sule + eK? — ‘Then ¥O=P IO: nee TIE Soliton solutions to a class of elliptic equations ml where F(t) = Yul? + POY Kola’ u? Ip — oP ulg It follows that f(0) = lu? > 0 and f(t) > 0 for small values of t and f(t) <0 for large values of t. Therefore there exists £ > 0 such that f(2) = 0, so that tu € N. Proof of (b): From the proof of item (a) we conclude that 0 < # <1 and it is unique. In fact, since f(0) = Jul!” > 0 and f(1) = (u)-u <0, then 0 pB, since p > q and noticing that SW = (6 — 199°) Kolutw wwe infer that (8 — 1) 9" Kol wg) Ga satisfies J*(f) = 0. Since J"(£) <0 for t > é and f"(t) > 0if 0 0 there exists a function. € W19(R) such that fy |e[Pdr = 1 and fy|#,|Pdr =. It suffices to consider functions @ : R +R such that support of ¢ is contained in [—1,1], / lolz) Par=1 and [g'B=e?, for al zeR. Defining, hae and taking the functions uf == ®,(z), we get ee tatip = 55 and lls Corresponding to such u, we find pyro) J) = Id? + BPO KOSS ). Since J(0) = Iluel? > 0, by choosing < Z;, we et J(t) <0 as t+ oo. Thus f hhas an unique zero. = Id? FE (Koe™ 12 ‘MJ. Alves, P.C, Carriao, O.F1. Miyagali Proof of (c): Since w € N, from the Sobolev embedding of W"?(R) z for $= co or s > p, it follows that * ine 2® {lull? = —Kop? ju’u wut> [2] Proof of (d): Let x(u) = I'(u)-u, Since & € N is a constrained critical point for I on N, from [10, Theorem 1.4.2], there exists 4 € R such that J'(i) = Ax'(ii)- Hence 0 = I'(i)- d= Ay'(a) “i. On the other hand, as I satisfies (4.2), we also have x/(7) -@ = I"(u)-w-% < 0 and thus we conclude that X= 0. ‘Then I"(t) = 0 and i is a solution of the equation ( 1.1). Proof of (e): For each w€ N and from (4.1) we have that p+ lulg < bulg < Chul? and =p>0. (43) ‘Thus In onder to prove the following two Tenuuats we will use the following crucial Proposition, whose the proof will be postponed. When p = 2, it was proved in (18). Proposition 4.1 Let u, — u weakly in W??(R) as n -+ 00 and set Un = tn — U- Then Timm inf full" |p > Timing up + [uu 4h ‘Assuming the Proposition, we conclude the proof of Lemma 2.2 and Lemma 2.3 as follows. Proof of Lemma 2.2. Proof of (a): Let (tin) CN be a minimizing sequence for I. From the Sobolev embedding, it is easy to see that |Iun|? is bounded in X. Hence, wwe can assume that ti, —* u weakly in X, as n+ 00. Proof of (b): Since un € N, from Lemma 2.1-(c) we have that in|? + BPKolunua "Ip 2 [hual? 2 p? > 0. eal Soliton solutions to a class of elliptic equations 1B Proof of (c): Set t= ty ~ ut Since I(t) thy = 0, we obtain that funl§ — fhunll” = BP Kouta yp ‘Taking the limin€ in the above equality and by wing the fact that the sequence (un) € Ny, from (3:1) , (3.2) and Proposition 4.1 we conclude that lira, [lunlG + lelg ~ lvl? — lhul?] > B”Kolirming |vpun Np + BPKolu’u? Also, 0 im |lu”— lim Junlg + A” Ko liming jogo NB + Wu) Hence Tintin 1 (Ug). + Fu). £0. . Proof of Lemma 2.3. Let (uy) C N be a minimizing sequence for J. From Lemma 2.2-(a) one has uy — u weakly in X, as n <> co. By (4.3) we conclude that m+ o(1) = (tn) = SEP fll + (4 PAP hal From Proposition 4.1, using the fact that the norm is lower semicontinuous, we obtain m = liminf I(u,) IP ul? + (q— pA) Kolu uf aa bul? + (a peer *Kolu?- +(q— v8) immint om "al v v QP ray? + (q—pB)61Kolu? tf = Tu). a ‘Thus J(u)

You might also like