Electronic Journal of Differential Equations, Vol. 2014 (2014), No. 199, pp. 1–14. ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu FRACTIONAL POROUS MEDIUM AND MEAN FIELD EQUATIONS IN BESOV SPACES XUHUAN ZHOU, WEILIANG XIAO, JIECHENG CHEN Abstract. In this article, we consider the evolution model ∂t u − ∇ · (u∇P u) = 0, P u = (−∆)−s u, 0 < s ≤ 1, x ∈ Rd , t > 0. We show that when s ∈ [1/2, 1), α > d + 1, d ≥ 2, the equation has a unique α . Moreover, in the critical local in time solution for any initial data in B1,∞ α , 2 ≤ p ≤ ∞, α > d/p, d ≥ 3. case s = 1, the solution exists in Bp,∞ 1. Introduction Let 0 < s ≤ 1, we consider the evolution equation ∂t u − ∇ · (u∇P u) = 0, P u = (−∆)−s u, u(x, 0) = u0 (x). (1.1) d where x ∈ R , t > 0. u = u(x, t) is a real-valued function, representing the density or concentration. P represents the pressure. When 0 < s < 1, we refer the equation as a fractional porous medium equation. It was first introduced by Caffarelli and Vázquez [2]. They proved the existence of a weak solution when u0 is a bounded function and has exponential decay at infinity. Caffarelli, Soria, and Vázquez [1] studied its regularity theory of the weak solution with u0 ∈ L1 ∩ L∞ and the continuity of bounded solutions. When s = 1, the equation leads to a mean field equation ut = ∇ · (u∇P u), P u = (−∆)−1 u, u(x, 0) = u0 (x). (1.2) Equation that was first studied by Lin and Zhang [7]. They proved the existence and uniqueness of positive L∞ solution in two dimensions. When d ≥ 3, Vázquez and Serfaty [11] studied the existence and uniqueness of the weak solution in L∞ spaces for (1.2) by taking the limit s → 1 in the weak solutions of (1.1). Since papers in literature only addressed the weak solution of (1.2), in this article we α show that the strong solution exists in Bp,∞ , α > 0. In this article, we are interested in finding the strong solutions of (1.1) in the α Besov spaces Bp,∞ . We will show that when d ≥ 2, s ∈ [1/2, 1), α > d + 1, α equation (1.1) has a unique local in time solution for any initial data in B1,∞ . α When s = 1, the solution extends to Bp,∞ , α > d/p, 2 ≤ p ≤ ∞, d ≥ 3. The idea 2000 Mathematics Subject Classification. 35K55, 35K65, 76S05. Key words and phrases. Fractional porous medium equation; mean field equation; local solution; Besov space. c 2014 Texas State University - San Marcos. Submitted April 10, 2014. Published September 23, 2014. 1 2 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 of our proof is inspired by the methods used in [3, 12], where the authors studied the quasi-geostrophic equation. In our proof, we construct two commutators, give three estimates of them, and construct a function sequence fitting our equation. The rest of this article is divided in four parts. Section 2 recalls the definition and some properties of the Besov spaces, the Bernstein inequality involving both integer and fractional derivatives, as well as some properties of the fractional Laplacian. In Section 3 we prove three estimates about the constructed commutators and a priori estimate of the solution. Section 4 proves the existence and uniqueness of the fractional porous medium equation. Section 5 is devoted to the mean field equation. The main results are the following two theorems. Theorem 1.1. Let d ≥ 2, s ∈ [1/2, 1], α > d + 1. Assume that the initial data u0 ∈ α α B1,∞ . Then we can find T = T (ku0 kB1,∞ ), such that a unique solution u to (1.1) on β α+2s−2 [0, T ]×Rd exists. And the solution belongs to C 1 ([0, T ]; B1,∞ )∩L∞ ([0, T ]; B1,∞ ), and β ∈ [α + 2s − 2, α]. Theorem 1.2. Let d ≥ 3, α > d/p when 2 ≤ p ≤ ∞. Assume that the initial data α α u0 ∈ Bp,∞ . Then we can find T = T (ku0 kBp,∞ ), such that a unique solution u to the given mean field equation (1.2) on [0, T ] × Rd exists. And the solution belongs α β ), β ∈ (d/p, α), 2 ≤ p ≤ ∞. ) ∩ L∞ ([0, T ]; Bp,∞ to C 1 ([0, T ]; Bp,∞ 2. Preliminaries In this section, we recall the definition of the Besov space. We start with a dyadic decomposition of Rd . Suppose χ ∈ C0∞ (Rd ), φ ∈ C0∞ (Rd \ {0}) satisfying 4 }, 3 3 8 supp φ ⊂ {ξ ∈ Rd : < |ξ| < }, 4 3 X χ(ξ) + φ(2−j ξ) = 1, ξ ∈ Rd , supp χ ⊂ {ξ ∈ Rd : |ξ| ≤ j≥0 X φ(2−j ξ) = 1, ξ ∈ Rd \{0}. j∈Z Define the operators −j Z h(2j y)u(x − y) dy, Z X Sj f = ∆k f = χ(2−j D)u = 2jd g(2j y)u(x − y) dy, ∆j u = φ(2 jd D)u = 2 (2.1) k≤j−1 ∨ where g = χ and h = φ∨ are the inverse Fourier transform of χ and φ, respectively. It can be easily verified that with our choice of φ, ∆j ∆k f ≡ 0, if |j − k| ≥ 2, ∆j (Sk−1 f ∆k f ) ≡ 0, if |j − k| ≥ 5. (2.2) Definition 2.1. For any α ∈ R, and p, q ∈ [1, ∞], the homogeneous Besov spaces r Ḃp,q are defined as α Ḃp,q = {f ∈ Z 0 (Rd ) : kf kḂ α < ∞}. p,q EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION 3 Here, kf kḂ α = hX p,q 2jαq k∆j f kqLp i1/q , where q < ∞. j∈Z When q = ∞, kf kḂ α p,∞ = sup 2jα k∆j f kLp . j∈Z Z (R ) denotes the dual space of Z(R ) = {f ∈ S(Rd ) : ∂ γ fˆ(0) = 0, ∀γ ∈ Nd } and can be identified by the quotient space of S 0 /P with the polynomials space P. 0 d d Definition 2.2. For any α ∈ R, and p, q ∈ [1, ∞], the inhomogeneous Besov space r Bp,q is defined as α α Bp,q = {f ∈ S 0 (Rd ) : kf kBp,q < ∞}. Here, α kf kBp,q = ∞ X 2jαq k∆j f kqLp 1/q + kS0 (f )kLp , when q < ∞. j≥0 When q = ∞, α kf kBp,∞ = sup 2jα k∆j f kLp + kS0 (f )kLp . j≥0 Let us state some basic properties of the Besov spaces. Proposition 2.3. Let s ∈ R, 1 ≤ p ≤ ∞, 1 ≤ q ≤ ∞. α α α ∩ Lp , and kf kBp,q = kf kḂ α + kf kLp ; = Ḃp,q (i) If α > 0, then Bp,q p,q α2 α1 α α (ii) If α1 ≤ α2 , then Bp,q ⊂ Bp,q . If 1 ≤ q1 ≤ q2 ≤ ∞, then Ḃp,q ⊂ Ḃp,q and 1 2 α α Bp,q1 ⊂ Bp,q2 ; α (iii) If α > dp , then Bp,q ,→ L∞ . If p1 ≤ p2 , α1 − pd1 > α2 − pd2 , then Bpα11,q1 ,→ α α α2 ; Bp2 ,q2 , Bp,min(p,2) ,→ Hpα ,→ Bp,max(p,2) α α α (iv) If α > 0, p ≥ 1, then kuvkBp,∞ ≤ CkukL∞ kvkBp,∞ + CkukBp,∞ kvkL∞ . We now turn to Bernstein’s inequalities. When the Fourier transform of a function is supported on a ball or an annulus, the Lp -norms of the derivatives of the function can be bounded in terms of the Lp -norms of the function itself. And it also exists when one replaces the derivatives by the fractional derivatives (see [6, 14]). Proposition 2.4. Let 1 ≤ p ≤ q ≤ ∞, γ ∈ Nd . (1) If α ≥ 0 and supp fˆ ⊂ {ξ ∈ Rd : |ξ| ≤ K2j } for some K > 0 and integer j, then 1 1 k(−∆)α Dγ f kLq ≤ C2j(2α+|γ|)+jd( p − q ) kf kLp . (2) If α ∈ R and supp fˆ ⊂ {ξ ∈ Rd : K1 2j ≤ |ξ| ≤ K2 2j } for some K1 , K2 > 0 and integer j, then 1 1 e j(2α+|γ|)+jd( p1 − q1 ) kf kLp , C2j(2α+|γ|)+jd( p − q ) kf kLp ≤ k(−∆)α Dγ f kLq ≤ C2 e are positive constants independent of j. where C and C Next we state two pointwise inequalities which were proved in [5, 13]. Proposition 2.5. Let 0 ≤ α ≤ 1, f ∈ C 2 (Rd ) decay sufficiently fast at infinity. Then for any x ∈ Rd , 2f (x)(−∆)α f (x) ≥ (−∆)α f 2 (x). 4 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 Proposition 2.6. Let 0 ≤ α ≤ 1, p1 = kl11 ≥ 0, p2 = kl22 ≥ 1 be rational numbers with l1 , l2 odd, and k1 l1 + k2 l2 even. Then for any f ∈ C 2 (Rd ) that decays sufficiently fast at infinity, and for any x ∈ Rd , (p1 + p2 )f p1 (x)(−∆)α f p2 (x) ≥ p2 (−∆)α f p1 +p2 (x). 3. A priori estimate Proposition 3.1. Let α > 0, s ∈ (0, 1), p ∈ [1, ∞] be given. Assume r > d/p. Then there exists some constant C such that r+1 kuk α+1−2s + Ckuk r+2−2s kvkB α 2jα k[∆j , ∂i (−∆)−s u]∂i vkLp ≤ CkvkBp,∞ , (3.1) Bp,∞ Bp,∞ p,∞ where the brackets [, ] represents the commutator, namely [∆j , ∂i (−∆)−s u]∂i v = ∆j (∂i (−∆)−s u)∂i v) − ∂i (−∆)−s u∆j (∂i v). Proof. Using Bony’s para-product decomposition, we have [∆j , ∂i (−∆)−s u]∂i v = L1 + L2 + L3 + L4 + L5 , where X L1 = ∆j [Sk−1 (∂i (−∆)−s u)∆k (∂i v)] − Sk−1 (∂i (−∆)−s u)∆k (∆j (∂i v)), |k−j|≤4 X L2 = ∆j [Sk−1 (∂i v)∆k (∂i (−∆)−s u)], |k−j|≤4 X L3 = e k (∂i v)), ∆j (∆k (∂i (−∆)−s u)∆ k≥j−2 L4 = X Sk−1 (∆j (∂i v))∆k (∂i (−∆)−s u), k X L5 = ∆j 0 (∆j (∂i v))∆j 00 (∂i (−∆)−s u). |j 0 −j 00 |≤1 We shall estimate the above terms separately. First observe Z X L1 = 2jd h(2j (x − y)) Sk−1 (∂i (−∆)−s u)(y) |k−j|≤4 − Sk−1 (∂i (−∆)−s u)(x) ∆k (∆j (∂i v))(y) dy. By Young’s inequality and Bernstein’s inequality, Z X −j −s kL1 kLp ≤ C 2 k∇∂i (−∆) ukL∞ k∆j (∂i v)kLp |y||h(y)| dy |k−j|≤4 ≤ C2−j 2j k(−∆)1−s ukL∞ k∆j vkLp r+2−2s . ≤ Ck∆j vkLp kukBp,∞ Similarly, r+1 kuk α+1−2s . kL2 kLp ≤ C2j(1−2s) k∇vkL∞ k∆j ukLp ≤ C2−jα kvkBp,∞ Bp,∞ We can also estimate, kL3 kLp ≤ C X k≥j−2 k∆k (∂i (−∆)−s u)kLp k∇vkL∞ EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION ≤ C2−jα X 5 r+1 2(j−k)α 2k(α+1−2s) k∆k ukLp kvkBp,∞ k≥j−2 −jα ≤ C2 α+1−2s kvk r+1 . kukBp,∞ Bp,∞ To estimate L4 , by the definition of Sj and ∆j , we can observe that only k satisfying k ≥ j survive. Thus X kL4 kLp ≤ Ck∇vkL∞ 2k(1−2s) k∆k ukLp k≥j ≤ C2−jα X r+1 kuk α+1−2s 2(j−k)α kvkBp,∞ Bp,∞ k≥j ≤ C2 −jα r+1 kuk α+1−2s . kvkBp,∞ Bp,∞ Since ∆k ∆j = 0, for |j − k| ≥ 2, we have r+1 kuk α+1−2s . kL5 kLp ≤ Ck∇vkL∞ 2j(1−2s) k∆j ukLp ≤ C2−jα kvkBp,∞ Bp,∞ Collecting the estimates above, we obtain 2jα k[∆j , ∂i (−∆)−s u]∂i vkLp j(α+1−2s) r+1 k∆j ukLp + kuk α+1−2s kvk r+1 r+2−2s + 2 kvkBp,∞ ≤ C2jα k∆j vkLp kukBp,∞ Bp,∞ Bp,∞ α r+2−2s + kuk α+1−2s kvk r+1 . ≤ kvkBp,∞ kukBp,∞ Bp,∞ Bp,∞ This completes the proof. Proposition 3.2. Let α, s, p, r be as in Proposition 3.1. Then there exists a constant C such that r α+2−2s + Ckuk r+2−2s kvkB α 2jα k[∆j , ∂i (−∆)−s u]∂i vkLp ≤ CkvkBp,∞ kukBp,∞ . (3.2) Bp,∞ p,∞ When p = ∞, this inequality becomes: for any r > 0, r α+2−2s + Ckuk r+2−2s kvkB α 2jα k[∆j , ∂i (−∆)−s u]∂i vkL∞ ≤ CkvkB∞,∞ kukB∞,∞ . B∞,∞ ∞,∞ (3.3) Proof. We want to give a new estimate of the commutator in Proposition 3.1. Following the above proof, the estimate of L1 unchanged, we give different bounds for L2 , L3 , L4 , L5 . First, Z X L2 = 2jd h(2j (x − y))(Sk−1 ∂i v)(y)∆k (∂i (−∆)−s u)(y) dy |k−j|≤4 X = 2jd Z ∂i h(2j (x − y))2j (Sk−1 v)(y)∆k (∂i (−∆)−s u)(y) dy |k−j|≤4 − X 2jd Z h(2j (x − y))(Sk−1 v)(y)∆k (∂ii (−∆)−s u)(y) dy. |k−j|≤4 So we obtain kL2 kLp ≤ C22−2s (k∂i hkL1 + khkL1 )kvkL∞ k∆j ukLp α+2−2s kvkB r ≤ C2−jα kukBp,∞ . p,∞ 6 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 Similarly, X kL3 kLp ≤ C 2jd Z ∂i h(2j (x − y))2j ∆k (∂i (−∆)−s u)(y)∆k v(y) dy k≥j−2 − 2jd Z h(2j (x − y))∆k (∂ii (−∆)−s u)(y)∆k v(y) dy. Hence we obtain, X kL3 kLp ≤ C (2(j−k)(α+1) + 2(j−k)α )2k(α+2−2s) kvkL∞ k∆k ukLp k≥j−2 α+2−2s kvkB r ≤ C2−jα kukBp,∞ . p,∞ Also, kL4 kLp ≤ X C2j k∆j vkL∞ 2k(1−2s) k∆k ukLp k≥j ≤ C2−jα X r 2(j−k)(α+1) 2k(α+2−2s) kvkBp,∞ k∆k ukLp k≥j −jα ≤ C2 r α+2−2s . kvkBp,∞ kukBp,∞ Finally, r α+2−2s . kL5 kLp ≤ Ck∆j vkL∞ 2j(2−2s) k∆j ukLp ≤ C2−jα kvkBp,∞ kukBp,∞ Collecting the estimates above, we can obtain α+2−2s r+2−2s + kvkB r 2jα k[∆j , ∂i (−∆)−s u]∂i vkLp ≤ C2jα k∆j vkLp kukBp,∞ kukBp,∞ p,∞ α r+2−2s + kuk α+2−2s kvkB r . ≤ kvkBp,∞ kukBp,∞ Bp,∞ p,∞ This completes the proof. Proposition 3.3. Let α > 0, s ∈ (0, 1), p ∈ [1, ∞]. Assume r > exists a constant C such that d p. Then there r+1 kuk α+1−2s + Ckuk r+2−2s kvkB α 2jα k[∆j , v](−∆)1−s ukLp ≤ CkvkBp,∞ , Bp,∞ Bp,∞ p,∞ (3.4) where the brackets [, ] represents the commutator, [∆j , v](−∆)1−s u = ∆j (v(−∆)1−s u) − v∆j ((−∆)1−s u). (3.5) Proof. This proposition is proved similarly to Proposition 3.1. We start the proof by writing [∆j , v](−∆)1−s u = I1 + I2 + I3 + I4 + I5 , where X I1 = ∆j (Sk−1 v∆k (−∆)1−s u) − Sk−1 v∆k (∆j (−∆)1−s u), |k−j|≤4 X I2 = ∆j (Sk−1 (−∆)1−s u)∆k v), |k−j|≤4 X I3 = e k (−∆)1−s u), ∆j (∆k v)∆ k≥j−2 I4 = X k Sk−1 (∆j (−∆)1−s u)∆k v, EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION X I5 = 7 ∆j 0 v∆j 00 (∆k (−∆)1−s u). |j 0 −j 00 |≤1 Similar to the proof for Proposition 3.1, first we observe that Z X kI1 kLp = k 2jd h(2j (x − y))(Sk−1 v(y) − Sk−1 v(x))∆k (−∆)1−s u)(y)dykLp |k−j|≤4 ≤C X 2−j k∇vkL∞ k∆j (−∆)1−s ukLp Z |y||h(y)| dy |k−j|≤4 ≤ C2−j 22j(1−s) k∇vkL∞ k∆j ukLp r+1 . ≤ C2j(1−2s) k∆j ukLp kvkBp,∞ Also we obtain r+2−2s k∆j vkLp , kI2 kLp ≤ Ck(−∆)1−s ukL∞ k∆j vkLp ≤ CkukBp,∞ X kI3 kLp ≤ C k∆k vkLp 2k(2−2s) k∆k ukL∞ k≥j−2 X ≤ C2−jα 2(j−k)α 2kα k∆k vkLp 2k(2−2s) k∆k ukL∞ k≥j−2 α r+2−2s . ≤ C2−jα kvkBp,∞ kukBp,∞ Similarly, we estimate kI4 kLp ≤ C X k(−∆)1−s ukL∞ k∆k vkLp k≥j ≤ C2−jα X 2(j−k)α k(−∆)1−s ukL∞ 2kα k∆j vkLp k≥j ≤ C2 −jα r+2−2s kvkB α , kukBp,∞ p,∞ r+2−2s k∆j vkLp . kI5 kLp ≤ Ck(−∆)1−s ukL∞ k∆j vkLp ≤ CkukBp,∞ Collecting the estimates above, we obtain r+1 2jα k[∆j , v](−∆)1−s ukLp ≤ C2j(α+1−2s) k∆j ukLp kvkBp,∞ r+2−2s r+2−2s k∆j vkLp + CkvkB α kukBp,∞ + C2jα kukBp,∞ p,∞ r+1 kuk α+1−2s + CkvkB α r+2−2s . ≤ CkvkBp,∞ kukBp,∞ Bp,∞ p,∞ This completes the proof. Proposition 3.4. Let s ∈ [1/2, 1], p = k/l be a rational number with k even, l odd, α α is a solution of (1.1) with u0 ∈ Bp,∞ and α > dp + 1. Assume that u(x, t) ∈ Bp,∞ for t ∈ [0, T ]. Then, when u(x, t) ≥ 0, we can find some C = C(p, α), that for any t ≤ T, Z t α α α kukBp,∞ ≤ Cku0 kBp,∞ exp{C kukBp,∞ dτ }. (3.6) 0 Proof. Applying ∆j on (1.1), we obtain X ∂t ∆j u = [∆j , ∂i (−∆)−s u]∂i u + ∇(−∆)−s u∆j (∇u) − [∆j , u](−∆)1−s u − u∆j ((−∆)1−s u. 8 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 Multiplying both sides by p∆j u|∆j u|p−2 and integrating over Rd , the equation becomes d k∆j ukpLp dt Z XZ p∆j u|∆j u|p−2 [∆j , ∂i (−∆)−s u]∂i u + p∆j u|∆j u|p−2 ∇(−∆)−s u∆j (∇u) = Z Z p−2 1−s − p∆j u|∆j u| [∆j , u](−∆) u − p∆j u|∆j u|p−2 u∆j ((−∆)1−s u = J1 + J2 + J3 + J4 . From Propositions 3.1 and 3.3, we obtain the estimates p−1 −jα 2 α α+1−2s ≤ C2 J1 ≤ C2−jα k∆j ukp−1 , kukBp,∞ k∆j ukL p kukB α Lp kukBp,∞ p,∞ −jα 2 α α+1−2s ≤ C2 k∆j ukp−1 . kukBp,∞ J3 ≤ C2−jα k∆j ukp−1 α Lp kukBp,∞ Lp kukBp,∞ It is easy to see that Z Z −s p J2 = ∇(−∆) u)∇(|∆j (u)| ) = (−∆)1−s u|∆j u|p dx p p−1 α−1 k∆j uk p ≤ Ckuk α+1−2s k∆j uk p k∆j ukLp ≤ Ck(−∆)1−s ukBp,∞ L L Bp,∞ p−1 α+1−2s kukB α k∆j ukL ≤ C2−jα kukBp,∞ p . p,∞ Using the fact that u ≥ 0 and Propositions 2.5 and 2.6, we estimate Z Z J4 ≤ − p|∆j u|p−2 u(−∆)1−s |∆j u|2 ≤ − u(−∆)1−s |∆j u|p Z p α−1 k∆j uk p = − (−∆)1−s u|∆j u|p ≤ Ck(−∆)1−s ukBp,∞ L α+1−2s kukB α ≤ C2−jα kukBp,∞ k∆j ukp−1 Lp . p,∞ d Combining the above four estimates, we set dt . Since u ≥ 0, kukḂ α ≤ Ckuk2Bp,∞ α p,∞ it follows Z Z d d kukLp = up dx = p up−1 ut dx dt dt Z Z = p up−1 ∇ · (u∇(−∆)−s u) = −(p − 1) ∇up · (∇(−∆)−s u) Z = −(p − 1) up ((−∆)1−s u) α α+1−2s . ≤ Ckukp−1 kukBp,∞ Lp kukBp,∞ This implies d α α α kukBp,∞ ≤ CkukBp,∞ kukBp,∞ , dt which with the Gronwall’s inequality yield (3.6). 4. Proof of Theorem 1.1 From the definition of Riesz potentials (−∆)−s u = c(n, s)|x|−d+2s ∗u, 0 < 2s < d. So when d ≥ 2, 21 ≤ s < 1, we define P u = c(n, s)(|x|−d+2s ∗ σ ) ∗ u = (−∆)−s (σ ∗ u) = (−∆)−s u , u = σ ∗ u . Here σ ∈ Cc∞ , is nonnegative, radially symmetric EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION 9 R and decreasing, σ = 1, σ = ε−d σ(x/). Then we construct a sequence {u(n) }, defined recursively by solving the following equations u(1) = S2 (u0 ) ∂t u(n+1) = ∇ · (u(n+1) ∇(P u(n) )) (n+1) u (x, 0) = un+1 0 (4.1) = Sn+2 u0 . Since u(n) solves the linear system, we can always find the sequence. Similarly to the proof of Proposition 3.4, taking ∆j on (4.1), we obtain X X (n+1) ∂t ∆j u(n+1) = [∆j , ∂i (−∆)−s u(n) + ∂i (−∆)−s u(n) ∆j (∂i u(n+1) ) ]∂i u − [∆j , u(n+1) ](−∆)1−s u(n) − u(n+1) ∆j ((−∆)1−s u(n) ). (4.2) ∆ u(n+1) Multiplying both sides by |∆jj u(n+1) | , integrating over Rd , we denote each corresponding part in the right side by J10 , J20 , J30 , J40 . Now we obtain the estimates α J10 ≤ C2−jα ku(n+1) kB1,∞ ku(n) kB α+1−2s , 1,∞ α J30 ≤ C2−jα ku(n+1) kB1,∞ ku(n) kB α+1−2s , (4.3) 1,∞ (n) where we used the fact that ku kB α+1−2s ≤ ku(n) kB α+1−2s . Also, we have 1,∞ 1,∞ Z J20 = ∇(−∆)−s u(n) )∇(|∆j (u(n+1) )|) Z (n+1) = (−∆)1−s u(n) |dx |∆j u (4.4) α ≤ C2−jα ku(n+1) kB1,∞ ku(n) kB α+1−2s , 1,∞ and J40 = − Z ∆j u(n+1) (n+1) u ∆j ((−∆)1−s u(n) ) |∆j u(n+1) | ∆j u(n+1) (n+1) u ∆j (∆(−∆)−s u(n) ) |∆j u(n+1) | Z ∆j u(n+1) =− ∇u(n+1) ∆j (∇(−∆)−s u(n) ) |∆j u(n+1) | Z = (4.5) α ≤ C2−jα ku(n) kB α+1−2s ku(n+1) kB1,∞ . 1,∞ On the other hand, Z Z Z u(n+1) (n+1) d (n+1) |u |dx = u dx = ∇ · (|u(n+1) |∇(−∆)−s u(n) ) = 0. dt |u(n+1) | t Collecting the estimates above, when s ≥ 1/2, we obtain d (n+1) α α ku kB1,∞ ≤ Cku(n+1) kB1,∞ ku(n) kB α+1−2s 1,∞ dt (n+1) (n) α α ≤ Cku kB1,∞ ku kB1,∞ . (4.6) 10 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 Now from Gronwall’s inequality we obtain (n+1) ku Z (n+1) α ku0 kB1,∞ t α exp{C ku(n) kB1,∞ )dτ } 0 Z t α α ≤ Cku0 kB1,∞ exp{C ku(n) kB1,∞ )dτ }, α kB1,∞ ≤ 0 α with C independent of n. Defining XT := C([0, T ]; B1,∞ ), we have α ku(n+1) kXT ≤ Cku0 kB1,∞ exp(CT ku(n) kXT ). Thus, by the standard induction argument we find α α sup ku(n) kB1,∞ ≤ 2Cku0 kB1,∞ , (4.7) 0≤t≤T0 α for all n ≥ 1, if exp(2CT0 ku0 kB1,∞ ) ≤ 2; namely, T0 ≤ ln 2 2C0 (1+ku0 kB α 1,∞ ). Using Proposition 3.2, 3.3, and (4.4), (4.5), we see d (n+1) α . ku kB α+2s−2 ≤ Cku(n+1) kB α+2s−2 ku(n) kB1,∞ 1,∞ 1,∞ dt Thus, from the uniform estimate (4.7), we have sup k 0≤t≤T0 ∂ (n) u kB α+2s−2 ≤ Cku0 k2B1,∞ . α 1,∞ ∂t (4.8) (4.9) α+2s−2 Let YT := C([0, T ]; B1,∞ ). We will prove that the sequence {u(n) } is Cauchy in YT1 for some T1 ∈ (0, T0 ). Considering the difference u(n+1) − u(n) , ∂t (u(n+1) − u(n) ) = ∇u(n+1) · ∇(−∆)−s u(n) − u(n+1) (−∆)1−s u(n) − ∇u(n) · ∇(−∆)−s u(n−1) + u(n) (−∆)1−s u(n−1) = ∇ · ((u(n+1) − u(n) )∇(−∆)−s u(n) ) + ∇ · (u(n) ∇(−∆)−s (u(n) − u(n−1) )), with initial datum (u(n+1) − u(n) )(x, 0) = ∆n+1 u0 . Proceeding as in the proof of (4.6), we obtain the estimate α k∇ · (u(n) ∇(−∆)−s (u(n) − u(n−1) ))kB α+2s−2 ≤ Cku(n) − u(n−1) kB α−1 ku(n) kB1,∞ . 1,∞ 1,∞ Proceeding as in the proof of (4.8), we obtain the estimate (n+1) α k∇ · ((u(n+1) − u(n) )∇(−∆)−s u(n) − u(n) kB α+2s−2 ku(n) kB1,∞ . )kB α+2s−2 ≤ Cku 1,∞ 1,∞ Then we obtain d (n+1) α ku − u(n) kB α+2s−2 ≤ Cku(n+1) − u(n) kB α+2s−2 ku(n) kB1,∞ 1,∞ 1,∞ dt α + Cku(n) − u(n−1) kB α+2s−2 ku(n) kB1,∞ , 1,∞ and k(u(n+1) − u(n) )(x, 0)kB α+2s−2 1,∞ = k∆n+1 u0 kB α+2s−2 = sup 2j(α+2s−2) k∆j ∆n+1 u0 kL1 1,∞ n(2s−2) ≤ C2 sup n≤j≤n+2 j jα 2 k∆j u0 kL1 EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION 11 α ≤ C2n(2s−2) ku0 kB1,∞ . The above inequalities and Gronwall’s inequality imply that ku(n+1) (t) − u(n) (t)kB α+2s−2 1,∞ α ≤ k(u(n+1) − u(n) )(x, 0)kB α+2s−2 exp(Cku0 kB1,∞ ) 1,∞ Z t α exp(Cku0 kB1,∞ (t − s))ku(n) − u(n−1) kB α+2s−2 (s)ds + 0 0 n(2s−2) ≤C2 (4.10) 1,∞ 0 (n) + C ku (n−1) −u 0 kYT1 (exp C T1 − 1), α where T1 ∈ [0, T0 ], and the constant C 0 = C 0 (ku0 kB1,∞ ). Thus, if C 0 (exp T1 − 1) < α+2s−2 1 (n) ∞ converges to u ∈ L ([0, T1 ]; B1,∞ ) in YT1 . By 2 , we can deduce that u the well-known interpolation inequality in the Besov spaces we have un → u in β L∞ ([0, T1 ]; B1,∞ ) for all β ∈ [α + 2s − 2, α]. Moreover, the estimate (4.8) allows us α+2s−2 to conclude that u ∈ C 1 ([0, T1 ]; B1,∞ ). Letting → 0, n → ∞, we find that u is β a solution to (1.1) in B1,∞ . β Next we prove that the solution is unique. Suppose that u, v in L∞ ([0, T1 ]; B1,∞ ) are two solutions of (1.1) associated with the initial condition u0 , v0 . Then u − v satisfies the equation u − v = ∇ · [(u − v)∇(−∆)−s u] + ∇ · [v∇(−∆)−s (u − v)], (u − v)(x, 0) = u0 − v0 . Working similarly with (4.10), we can obtain k(u − v)(t)kB β 1,∞ ≤ M 2j(β−α) k(u − v)(x, 0)kB β 1,∞ + M ku − vkC([0,T1 ];B β 1,∞ 00 ) (exp C T1 − 1), for M = M (k(u0 kB β , kv0 kB β ). If M (exp C 00 T1 − 1) < 1, then we get the unique1,∞ 1,∞ β ness of solutions in C([0, T1 ]; B1,∞ ). The proof is complete. 5. Proof of Theorem 1.2 When d ≥ 3, we know (−∆)−1 u = c(n, s)|x|−d+2 ∗ u. Let P u = c(n, s)(|x|−d+2 ∗ σ )∗u = (−∆)−1 (σ ∗u) = (−∆)−1 u , where σ is defined in the beginning of Section 4. Consider the successive approximation sequence {u(n) } satisfying u(1) = S2 (u0 ) (n+1) ∂t u(n+1) = ∇ · (u(n+1) ∇(P u(n) · ∇(−∆)−1 u(n) − u(n+1) u(n) )) = ∇u (n+1) u (x, 0) = un+1 0 (5.1) = Sn+2 u0 . Appling ∆j on (5.1), ∂t ∆j u(n+1) = X (n+1) [∆j , ∂i (−∆)−s u(n) ]∂i u (n+1) + ∇(−∆)−s u(n) ) − ∆j (u(n+1) u(n) ∇∆j (u ). (5.2) When p ≥ 2, multiplying both sides of (5.2) by p∆j u(n+1) |∆j u(n+1) |p−2 and integrating over Rd , we obtain XZ d (n+1) p (n+1) k∆j u kLp = p∆j u(n+1) |∆j u(n+1) |p−2 [∆j , ∂i (−∆)−1 u(n) ]∂i u dt 12 X. ZHOU, W. XIAO, J. CHEN Z + Z − EJDE-2014/199 (n+1) p∆j u(n+1) |∆j u(n+1) |p−2 ∇(−∆)−1 u(n) ) ∇∆j (u 00 00 00 p∆j u(n+1) |∆j u(n+1) |p−2 ∆j (u(n+1) u(n) ε ) = J1 + J2 + J3 . By Hölder’s inequality, −1 (n) J100 ≤ Ck∆j u(n+1) kp−1 u ]∂i u(n+1) kLp . Lp k[∆j , ∂i (−∆) From the proof of Proposition 2.3 (iv), (n) J300 ≤ k∆j u(n+1) kp−1 kL∞ k∆j u(n+1) kLp + ku(n+1) kL∞ k∆j u(n) kLp ]. Lp [ku Also we obtain J200 = Z Z = (n+1) p ∇(−∆)−1 u(n) )| ) ∇(|∆j (u (n+1) p u(n) | dx ≤ ku(n) kL∞ k∆j u(n+1) kpLp . |∆j u Now we obtain d k∆j u(n+1) kLp ≤ Ck[∆j , ∂i (−∆)−1 u(n) ]∂i u(n+1) kLp dt + Cku(n) kL∞ k∆j u(n+1) kLp + Cku(n+1) kL∞ k∆j u(n) kLp . When, α > d/p, d (n+1) α α ku kḂ α ≤ Cku(n+1) kBp,∞ ku(n) kBp,∞ . p,∞ dt Letting p → ∞, d k∆j u(n+1) kL∞ ≤ Ck[∆j , ∂i (−∆)−1 u(n) ]∂i u(n+1) kL∞ dt + Cku(n) kL∞ k∆j u(n+1) kL∞ + Ck∆j (u(n+1) u(n) )kL∞ . So for any α > 0, d (n+1) α α ku kḂ α ≤ Cku(n+1) kB∞,∞ ku(n) kB∞,∞ . ∞,∞ dt It is easy to prove that when p = 1, Z Z d u(n+1) (n+1) |u(n+1) |dx = u dx dt |u(n+1) | t Z = ∇ · (|u(n+1) |(|x|−d+2 ∗ σ ) ∗ u(n) ) = 0. Noting u(n+1) (x, 0) = un+1 = Sn+2 u0 , we know ku(n+1) kL1 ≤ C. 0 When p ≥ 2, Z Z d (n+1) |u(n+1) |p dx = p |u(n+1) |p−2 u(n+1) ut dx dt Z = p |u(n+1) |p−2 u(n+1) ∇ · (u(n+1) ∇(−∆)−1 u(n) ) Z = −(p − 1) ∇|u(n+1) |p · (∇(−∆)−1 u(n) ) Z = (p − 1) |u(n+1) |p u(n) ≤ Cku(n+1) kpLp ku(n) kL∞ . EJDE-2014/199 FRACTIONAL POROUS MEDIUM EQUATION 13 It means d (n+1) α α ku kLp ≤ ku(n+1) kLp ku(n) kL∞ ≤ Cku(n+1) kBp,∞ ku(n) kBp,∞ . dt Again letting p → ∞, we obtain d (n+1) α α ku kL∞ ≤ Cku(n+1) kB∞,∞ ku(n) kB∞,∞ . dt Collecting the estimates above, we now obtain d (n+1) d α α α ku kBp,∞ ≤ Cku(n+1) kBp,∞ ku(n) kBp,∞ , p < ∞, α > , dt p d (n+1) α α α ku kB∞,∞ ≤ Cku(n+1) kB∞,∞ ku(n) kB∞,∞ , p = ∞, α > 0. dt For any β ∈ (d/p, α), p < ∞, or β ∈ (0, α), p = ∞ we have d (n+1) ku kBp,∞ ≤ ku(n+1) kBp,∞ ku(n) kBp,∞ , β β β dt and d (n+1) ku − u(n) kBp,∞ ≤ ku(n+1) − u(n) kBp,∞ ku(n) kBp,∞ β β β dt + ku(n) − u(n−1) kBp,∞ ku(n) kBp,∞ . β β The rest of the proof is similar to the proof of Theorem 1.1, we omit it. Acknowledgments. This research was supported by the National Natural Science Foundation of China (11201103). References [1] L. A. Caffarelli, F. Soria, J. L. Vázquez; Regularity of solutions of the fractional porous medium flow, arXiv preprint arXiv:1201.6048, 2012. [2] L. A. Caffarelli, J. L. Vázquez; Nonlinear porous medium flow with fractional potential pressure, Arch.Rational Mech.Anal, 202(2011), 537-565. [3] D. Chae; The quasi-geostrophic equation in the TriebelCLizorkin spaces, Nonlinearity, 16(2003), 479-495. [4] Q. Chen, C. Miao, Z. Zhang; A new Bernstein’s inequality and the 2D dissipative quasigeostrophic equation, Communications in mathematical physics, 271(2007), 821-838. [5] A. Córdoba, D. Córdoba; A maximum principle applied to quasi-geostrophic equations, Communications in mathematical physics, 249(2004), 511-528. [6] P. G. Lemarı̈-Rieusset; Recent development in the navier-stokes problem, CRC Press, 2002. [7] F. Lin, P. Zhang; On the hydrodynamic limit of Ginzburg-Landau wave vortices, Communications on pure and applied mathematics, 55(2002), 831-856. [8] H. Triebel; Theory of Function Spaces. Monograph in Mathematics, Vol.78, Basel: Birkhauser Verlag, 1983. [9] J. L. Vázquez; The Porous Medium Equation: Mathematical Theory, Oxford Mathematical Monographs, 2006. [10] J. L. Vázquez; Nonlinear diffusion with fractional Laplacian operators, Nonlinear Partial Differential Equations. Springer Berlin Heidelberg, 2012: 271-298. [11] J. L. Vázquez, S. Serfaty; A mean field equation as limit of nonlinear diffusions with fractional Laplacian operators, Calculus of Variations and Partial Differential Equations, 49 (2014), 1091-1120. [12] J. Wu; Solutions of the 2D quasi-geostrophic equation in Hölder spaces, Nonlinear Anal,62(2005), 579-594. [13] J. Wu; Lower bounds for an integral involving fractional Laplacians and the generalized Navier-Stokes equations in Besov spaces, Communications in mathematical physics, 263(2006), 803-831. 14 X. ZHOU, W. XIAO, J. CHEN EJDE-2014/199 [14] J. Wu; Existence and uniqueness results for the 2-D dissipative quasi-geostrophic equation, Nonlinear Anal,67(2007), 3013-3036. Xuhuan Zhou Department of Mathematics, Zhejiang University, 310027 Hangzhou, China E-mail address: zhouxuhuan@163.com Weiliang Xiao Department of Mathematics, Zhejiang University, 310027 Hangzhou, China E-mail address: xwltc123@163.com Jiecheng Chen Department of Mathematics, Zhejiang Normal University, 321004 Jinhua, China E-mail address: jcchen@zjnu.edu.cn