Uniform blow-up rate for a porous medium equation with a weighted localized source

  • Weili Zeng1,

    Affiliated with

    • Xiaobo Lu2Email author and

      Affiliated with

      • Qilin Liu3

        Affiliated with

        Boundary Value Problems20112011:57

        DOI: 10.1186/1687-2770-2011-57

        Received: 21 June 2011

        Accepted: 28 December 2011

        Published: 28 December 2011

        Abstract

        In this article, we investigate the Dirichlet problem for a porous medium equation with a more complicated source term. In some cases, we prove that the solutions have global blow-up and the rate of blow-up is uniform in all compact subsets of the domain. Moreover, in each case, the blow-up rate of |u(t)| is precisely determined.

        Keywords

        porous medium equation localized source blow-up, uniform blow-up rate

        1 Introduction

        Let Ω be a bounded domain in ℝ N (N ≥ 1) with smooth boundary ∂Ω. We consider the following parabolic equation with a localized reaction term
        v τ - Δ v m = a ( x ) v q 1 ( x , τ ) v s 1 ( x 0 , τ ) , x Ω , τ > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ1_HTML.gif
        (1.1)
        v ( x , τ ) = 0 , x Ω , τ > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ2_HTML.gif
        (1.2)
        v ( x , 0 ) = v 0 ( x ) , x Ω , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ3_HTML.gif
        (1.3)

        where m ≥ 1, q1 ≥ 0, s1 > 0 and x0 ∈ Ω is a fixed point. Throughout this article, we assume the functions a(x) and v0(x) satisfy the following conditions:

        (A1) a(x) and v0(x) ∈ C2(Ω); a(x), v0(x) > 0 in Ω and a(x) = v0(x) = 0 on ∂Ω.

        When Ω = B = {x ∈ ℝ N ; |x| < R}, we sometimes assume

        (A2) a(x) and v0(x) are radially symmetric; a(r) and v0(r) are non-increasing for r ∈ [0, R].

        Problems (1.1)-(1.3) arise in the study of the flow of a fluid through a porous medium with an internal localized source and in the study of population dynamics (see [13]). Porous medium equations (m > 1) with or without local sources have been studied by many authors [46].

        Concerning (1.1)-(1.3), to the best of authors knowledge, a number of articles have studied it from the point of the view of blow-up and global existence [710]. Many studies have been devoted to the case m = 1 [1013]. The case m = 1, a(x) = 1, q1 = 0, s1 ≥ 1 and m = 1, a(x) = 1, q1, s1 > 1 were studied by Souple [10, 11]. Souple [10] demonstrated that the positive solution blows up in finite time if the initial value v0 is large enough. In the case a(x) = 1, q1 = 0, and s1 > 1, Souple [11] showed that the solution v(x, τ) blows up globally and the blow-up rate is precisely determined. The case q1 = 0 and s1 > 0 was studied by Cannon and Yin [12] and Chandam et al. [13]. Cannon and Yin [12] studied its local solvability and Chandam et al. [13] investigated its blow-up properties.

        The study of this article is motivated by some recent results of related problems (see [14][15][16]. In the case of a(x)(= constant), the global existence and blow-up behavior have been considered by Chen and Xie [15]. It turns out that if q1 + s1 < m or q1 + s1 = m and a(x)(= constant) is sufficiently small, there exists a global solution of problem (1.1)-(1.3); if q1 + s1 > m, the solution of problem (1.1)-(1.3) blows up for large initial datum while it admits a global solution for small initial datum. Furthermore, Du and Xiang [16] obtained the blow-up rate estimates under some appropriate hypotheses on initial datum. For some related localized models arising in physical phenomena, we refer the readers to [1719] and the references therein.

        For the localized semi-linear parabolic equation of the form
        v τ - Δ v = v q 1 ( x , τ ) v s 1 ( x 0 , τ ) , x Ω , τ > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ4_HTML.gif
        (1.4)

        with the Dirichlet boundary condition (1.2) and the initial condition (1.3). In [20], Li and Wang proved that the blow-up set to system (1.2)-(1.4): (a) the system possesses total blow-up when q1 ≤ 1; (b) the system presents single point blow-up patterns when q1 > 1.

        We now restrict ourselves to the problem of the form
        v τ - Δ v m = a ( x ) v q 1 ( x , τ ) v s 1 ( 0 , τ ) , x B , τ > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ5_HTML.gif
        (1.5)
        v ( x , τ ) = 0 , x B , τ > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ6_HTML.gif
        (1.6)
        v ( x , 0 ) = v 0 ( x ) , x B , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ7_HTML.gif
        (1.7)
        where q1 ≥ 0, s1 > 0, and q1 + s1 > m > 0. When m = 1, it was proved in [14] that
        1. (1)

          If 0 ≤ q1 ≤ 1 and q1 + s1 > 1, then the solution of (1.5)-(1.7) blows up in a finite time T.

           
        2. (2)

          If q1 > 1, then x = 0 is the only blow-up point for (1.5)-(1.7).

           
        In the meantime, they obtained the blow-up rate estimate but less precise. Namely,
        1. (i)
          If 0 ≤ q1 < 1, then for any xB
          C 1 ( a ( x ) ) 1 ( 1 - q 1 ) v ( x , τ ) ( T - τ ) 1 ( q 1 + s 1 - 1 ) C 2 , a s τ T , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equa_HTML.gif
           
        where C 1 = ( ( a ( 0 ) ) s 1 ( 1 - q 1 ) ( q 1 + s 1 - 1 ) ) 1 ( 1 - q 1 - s 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq1_HTML.gif, C 2 = ( a ( 0 ) ( p + q - 1 ) ) 1 ( 1 - q 1 - s 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq2_HTML.gif.
        1. (ii)
          If q1 = 1, then for any xB
          a ( x ) a ( 0 ) ln ( T - τ ) - 1 s 1 ln v ( x , τ ) ln ( T - τ ) - 1 s 1 , a s τ T . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equb_HTML.gif
           

        It seems that the results of [14] can be extended to m ≥ 1 and the blow-up rate can be precisely determined. Motivated by this, in this article, we will extend and improve the results of [14].

        The purpose of this article is to determine the blow-up rate of solutions for a nonlinear parabolic equation with a weighted localized source, that is, we investigate how the localized source and the local term affect the blow-up properties of the problem (1.5)-(1.7). Indeed, we find that when q1 ≤ 1, the solution of (1.5)-(1.7) blows up at the whole domain with a uniformly blow-up profile.

        The rest of this article is organized as follows. The results are stated in Section 2. We then prove these results in Section 3.

        2 Preliminaries and Main Results

        The following two theorems are our main results.

        Theorem 2.1 Assume q1 + s1 > m, (A1) and (A2) hold. Let v(x, t) be the solution of problem (1.5)-(1.7), then v(x,t) blows up provided that the initial value v0(x) is sufficiently large.

        The method used in the proof Theorem 2.1 is originally due to [8, 18], and bears much resemblance to that of Theorem 3.2 in [15] and Theorem 1.3 in [16]. Therefore, we omitted them here.

        For the case q1 > 1, we do not know how to deal with the uniform blow-up rate of problem (1.5)-(1.7). In the following, we focus only on the case of 0 ≤ q1 ≤ 1.

        Theorem 2.2 Assume (A1) and (A2) hold. Let v(x, t) be the blow-up solution of (1.5)-(1.7), which blows up in finite time T and v(x, t) is non-decreasing in time, then the following limits hold uniformly in all compact subsets of B.
        1. (i)
          If 0 ≤ q1 < 1, then
          lim τ T ( T - τ ) 1 ( q 1 + s 1 - 1 ) v ( x , τ ) = C ( a ( x ) ) 1 ( 1 - q 1 ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ8_HTML.gif
          (2.1)
           
        where C = ( ( q 1 + s 1 - 1 ) ( a ( 0 ) ) s 1 ( 1 - q 1 ) ) 1 ( 1 - q 1 - s 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq3_HTML.gif.
        1. (ii)
          If q1 = 1, then
          lim τ T ln v ( x , τ ) = a ( x ) a ( 0 ) ln ( T - τ ) - 1 s 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ9_HTML.gif
          (2.2)
           

        Remark 2.1 The domain we considered here is a ball, it seems that the results of Theorem 2.2 remain valid for the general domain. (It is an open problem in this case.)

        To get the blow-up profiles for problem (1.5)-(1.7), we need some transformations. Let u(x, t) = v m (x, τ), t = , then (1.5)-(1.7) becomes
        u t = u p ( Δ u + a ( x ) u q ( x , t ) u s ( 0 , t ) ) , x B , t > 0 , u ( x , t ) = 0 , x B , t > 0 , u ( x , 0 ) = u 0 ( x ) = v 0 m ( x ) , x B , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ10_HTML.gif
        (2.3)

        where 0 ≤ p = (m - 1)/m < 1, q = q1/m, and s = s1/m.

        Under above transformation, assumptions (A1) and (A2) become

        (B1) a(x) and u0(x) ∈ C2(B); a(x), u0(x) > 0 in B and a(x) = u0(x) = 0 on ∂B.

        (B2) a(x) and u0(x) are radially symmetric; a(r) and u0(r) are non-increasing for r ∈ [0, R].

        In our consideration, a crucial role is played by the Dirichlet eigenvalue problem
        - Δ φ = λ φ , in B , φ ( x ) = 0 , on B . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ11_HTML.gif
        (2.4)

        Denote λ be the first eigenvalue and by φ the corresponding eigenfunction with φ(x) > 0 in B, normalized by B a ( x ) φ ( x ) d x = 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq4_HTML.gif.

        3 Proof of Theorem 2.2

        For convenience, we denote
        g ( t ) = u s ( 0 , t ) a n d G ( t ) = 0 t g ( s ) d s . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equc_HTML.gif
        Before proving our result, we would like to give a property of the following problem
        w t = w α ( Δ w + a ( x ) g ( t ) ) , x B , t > 0 , w ( x , t ) = 0 , x B , t > 0 , w ( x , 0 ) = w 0 ( x ) = u 0 1 - q ( x ) , x B , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ12_HTML.gif
        (3.1)

        where 0 ≤ α ≤ 1 and w = u1-q(x, t).

        Lemma 3.1 Assume (B1) and (B2) hold. Let w(x, t) be the solution of Equation (3.1), which blows up in a finite time T* and non-decreasing in time t, then the following limits hold uniformly in all compact subsets of B.
        1. (i)
          If 0 ≤ α < 1, then
          lim t T * w 1 - α ( x , t ) G ( t ) = ( 1 - α ) a ( x ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equd_HTML.gif
           
        2. (ii)
          If α = 1, then
          lim t T * ln w ( x , t ) G ( t ) = a ( x ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Eque_HTML.gif
           
        Proof. (i) Assumption (B2) implies w r ≤ 0 (r = |x|), it then follows that w ( 0 , t ) = max x B ̄ w ( x , t ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq5_HTML.gif and Δw(0, t) ≤ 0 for t > 0. From (3.1), we then get
        d w 1 - α ( 0 , t ) d t ( 1 - α ) a ( 0 ) g ( t ) , 0 < t < T * . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equf_HTML.gif
        .
        Consequently,
        lim t T * sup w 1 - α ( 0 , t ) G ( t ) ( 1 - α ) a ( 0 ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ13_HTML.gif
        (3.2)
        which implies
        lim t T * G ( t ) = a n d lim t T * g ( t ) = . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equg_HTML.gif

        Moreover, it is apparent that limt-T*w(0, t)/g(t) = 0, since s > 1 - q.

        Set R1 ∈ (0, R), B1 = {x ∈ ℝ N , | x |< R1} and b(x) = 1/a(x), xB1. Since a'(r) ≤ 0, we obtain that b'(r) ≥ 0, for 0 ≤ rR1.

        We now introduce the function
        w 1 ( x , t ) = b 1 ( 1 - α ) ( x ) w ( x , t ) , x B 1 , 0 < t < T * . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equh_HTML.gif
        By a simple calculation, and note that ∇w(x, t)∇b(x) = u r (r, t)b'(r) ≤ 0, then there exist m1, m2 > 0 such that
        b ( x ) Δ w ( x , t ) m 1 Δ w 1 ( x , t ) - m 2 w ( x , t ) x B 1 , 0 < t < T * . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equi_HTML.gif

        Setting ε(t) = m2w(0, t)/g(t). From limtT*w(0, t)/g(t) = 0, we infer that there exists t1 ∈ (0, T*) such that 0 < ε(t) ≤ 1/2 for t1t < T*.

        Hence, in view of (3.1), we observe
        1 1 - α ( w 1 1 - α ) t = b ( x ) Δ w + g ( t ) m 1 Δ w 1 + ( 1 - ε ( t ) ) g ( t ) + ε ( t ) g ( t ) - m 2 w ( 0 , t ) = m 1 Δ w 1 + ( 1 - ε ( t ) ) g ( t ) , x B 1 , t 1 < t < T * . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equj_HTML.gif
        Set g1(t) = (1 - ε(t))g(t), G 1 ( t ) = t 1 t g ( s ) d s http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq6_HTML.gif, we then obtain
        lim t T * G 1 ( t ) = and lim t T * G 1 ( t ) G ( t ) = 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equk_HTML.gif
        Obviously, w1(x, t) is a sup-solution of the following equation
        { ( w * ) t = ( w * ) α ( ( m 1 w * + g 1 ( t ) ) , x B 1 , t 1 < t < T * , w * ( x , t ) = 0 , x B 1 , t > t 1 , w * ( x , t 1 ) = b 1 / 1 α ( x ) w ( x , t 1 ) , x B 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equl_HTML.gif
        By the maximum principle, w1(x, t) ≥ w*(x, t) and w r * 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq7_HTML.gif. Similar to the proof of (4.15) in [15] that
        lim t T * ( w * ) 1 - α ( x , t ) G ( t ) = ( 1 - α ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equm_HTML.gif

        uniformly in all compact subsets of B1,

        Therefore, by the arbitrariness of B1, we obtain that the following inequatlity holds uniformly in all compact subsets of B
        lim t T * inf w ( 1 - α ) ( x , t ) G ( t ) ( 1 - α ) a ( x ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ14_HTML.gif
        (3.3)
        In particular,
        lim t T * inf w ( 1 - α ) ( 0 , t ) G ( t ) ( 1 - α ) a ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ15_HTML.gif
        (3.4)
        From (3.2) and (3.4), we deduce
        lim t T * w ( 1 - α ) ( 0 , t ) G ( t ) = ( 1 - α ) a ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ16_HTML.gif
        (3.5)
        Multiplying both sides of (3.1) by φ and integrating over B × (0, t), 0 < t < T*
        1 1 - α B w 1 - α φ d x - B w 0 1 - α φ d x = - λ 0 t B w φ d x d s + G ( t ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equn_HTML.gif
        Since 0 t B w φ d x d s B φ d x 0 t w ( 0 , s ) d s http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq8_HTML.gif, so we have
        lim t T * 0 t B w φ d x d s G ( t ) = 0 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equo_HTML.gif
        It then follows that
        lim t T * B w 1 - α φ d x G ( t ) = ( 1 - α ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ17_HTML.gif
        (3.6)
        Note that w r ≤ 0, (3.3) and (3.6), it is sufficient to prove
        lim t T * sup w 1 - α ( x , t ) G ( t ) ( 1 - α ) a ( x ) , x B . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ18_HTML.gif
        (3.7)
        Assume on the contrary that there exists a point x1B, x1 ≠ 0 such that
        lim t T * sup w 1 - α ( x 1 , t ) G ( t ) = c > ( 1 - α ) a ( x 1 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equp_HTML.gif
        Then there exists a sequence {t n } such that t n T*
        lim t n T * sup w 1 - α ( x 1 , t n ) G ( t n ) = c > ( 1 - α ) a ( x 1 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equq_HTML.gif
        By the continuity of a(x), we deduce that there exists x2B such that (1 - α)a(x) < c for B1 = {x ∈ ℝ n : |x2| ≤ |x| ≤ |x1|}. Using w r ≤ 0, (3.3) and (3.6), it is easy to check that
        lim t n T * B w 1 - α ( x , t n ) φ ( x ) d x G ( t n ) = lim t n T * B \ B 1 w 1 - α ( x , t n ) φ ( x ) d x + B 1 w 1 - α ( x , t n ) φ ( x ) d x G ( t n ) B \ B 1 ( 1 - α ) a ( x ) φ ( x ) d x + lim t n T * c B φ ( x ) d x > ( 1 - α ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equr_HTML.gif

        which is a contradiction to (3.6). Combining (3.3) and (3.7), Lemma 3.1 (i) is proved. Case (ii) can be treated similarly.

        The key step in establishing the result of Theorem 2.2 is the following lemma.

        Lemma 3.2 Under the assumption of Lemma 3.1, let u(x, t) be the blow-up solution of (2.3), which blows up in a finite time T* and non-decreasing in time t, then the following statements hold uniformly in all compact subsets of B:
        1. (i)
          If p + q < 1, then
          lim t T * u 1 - q - p ( x , t ) G ( t ) = ( 1 - q - p ) a ( x ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equs_HTML.gif
           
        2. (ii)
          If p + q = 1, then
          lim t T * ln u ( x , t ) G ( t ) = a ( x ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equt_HTML.gif
           
        Proof. (i) Since u r ≤ 0 and u t ≥ 0, it then follows that u ( 0 , t ) = max x B ̄ u ( x , t ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq9_HTML.gif and Δu(0, t) ≤ 0 for t > 0, which imply limtT*u(0, t) = ∞. Obviously,
        u t ( 0 , t ) a ( 0 ) u p + q ( 0 , t ) g ( t ) , 0 < t < T * , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equu_HTML.gif
        which implies
        lim t T * sup u 1 - p - q ( 0 , t ) G ( t ) ( 1 - p - q ) a ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ19_HTML.gif
        (3.8)

        Notice that p + q < 1 and (3.8), hence limtT*G(t) = ∞ and limtT*g(t) = ∞.

        A simple calculation yields
        1 1 - r Δ u 1 - r = - r u - ( 1 + r ) | u | 2 + u - r Δ u ( if 0 < r < 1 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equv_HTML.gif
        In view of (2.3), we have, for x ∈ Ω, 0 < t < T*
        1 - q 1 - p - q d u 1 - p - q d t = Δ u 1 - q + q ( 1 - q ) u - q - 1 | u | 2 + ( 1 - q ) a ( x ) g ( t ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ20_HTML.gif
        (3.9)
        Multiplying both sides of Equation (3.9) by φ and integrating over B × (0, t), it follows that
        1 1 - p - q B u 1 - p - q φ d x - B u 0 1 - p - q φ d x = - λ 1 - q 0 t B u 1 - q φ d x d s + G ( t ) + 0 t B q u - q - 1 | u | 2 φ d x d s , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ21_HTML.gif
        (3.10)
        for 0 < t < T*. Clearly,
        0 t B u 1 - q φ d x d s 0 t u 1 - q ( 0 , t ) d s B φ ( x ) d x , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ22_HTML.gif
        (3.11)
        which yields
        lim t T * 0 t B u 1 - q φ d x d s G ( t ) = 0 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ23_HTML.gif
        (3.12)
        Setting u1(r, t) = u(1-q)/ 2(r, t)(r = |x|). We may claim that
        lim t T * ( u 1 ( r , t ) ) r ( g ( t ) ) 1 2 = 0 , a . e . r ( 0 , R ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equw_HTML.gif
        Indeed, due to limtT*g(t) = limtT*u s (0, t) = ∞, u r ≤ 0, and s > 1 - q, we then have
        lim t T * 0 R ( u 1 ( r , t ) ) r d r ( g ( t ) ) 1 2 = lim t T * u 1 ( R , t ) - u 1 ( 0 , t ) ( g ( t ) ) 1 2 = 0 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equx_HTML.gif
        Therefore, by Lebesgue's dominated convergence theorem, we infer that
        lim t T * B q u - q - 1 | u | 2 φ ( x ) d x g ( t ) = q ω n lim t T * 0 R u - q - 1 ( r , t ) u r 2 φ ( r ) r n - 1 d r g ( t ) q ω n R n - 1 lim t T * 0 R u - q - 1 ( r , t ) u r 2 φ ( r ) d r g ( t ) = 4 q ( 1 - q ) 2 R n - 1 ω n lim t T * 0 R ( ( u ( 1 - q ) 2 ) r ) 2 φ ( r ) d r g ( t ) C 0 R lim t T * ( ( u ( 1 - q ) 2 ) r ( g ( t ) ) 1 2 ) 2 d r = 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ24_HTML.gif
        (3.13)

        where w n is the surface area of unit ball in ℝ N .

        Now according to (3.10)-(3.12), we obtain
        lim t T * B u 1 - p - q φ d x G ( t ) = ( 1 - p - q ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ25_HTML.gif
        (3.14)
        On the other hand, By (3.9), we find
        d u 1 - q d t u p ( Δ u 1 - q + ( 1 - q ) a ( x ) g ( t ) ) , x B , 0 < t < T * , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equy_HTML.gif
        where γ = p/(1 - q). Consequently, u1-qis a sup-solution of the problem
        d v d t = v γ ( Δ v + ( 1 - q ) a ( x ) g ( t ) ) , x B , 0 < t < T * , v ( x , t ) = 0 , x B , t > 0 , v ( x , 0 ) = u 0 1 - q ( x ) , x B . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equz_HTML.gif
        By the maximum principle, u1-qv in B × (0, T*). Note that 0 ≤ γ < 1, we know from Lemma 3.1 (i) that
        lim t T * v ( 1 - p - q ) ( 1 - q ) ( x , t ) G ( t ) = ( 1 - p - q ) a ( x ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equaa_HTML.gif

        uniformly in all compact subsets of B.

        Thus,
        lim t T * inf u ( 1 - p - q ) ( x , t ) G ( t ) ( 1 - p - q ) a ( x ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ26_HTML.gif
        (3.15)

        uniformly in all compact subsets of B.

        Next, we prove that
        lim t T * sup u ( 1 - p - q ) ( x , t ) G ( t ) ( 1 - p - q ) a ( x ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ27_HTML.gif
        (3.16)

        uniformly in all compact subsets of B.

        We can verify (3.15) by similar means of (3.7). Therefore, we conclude the proof of case (i).
        1. (ii)
          Proceeding as (3.8), we have
          lim t T * sup ln u ( 0 , t ) G ( t ) a ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equab_HTML.gif
           

        For any compact subset B1B, there exists t1 ∈ (0, T*) such that u(x, t1) ≥ 1 for all x B ̄ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq10_HTML.gif, and thus ln u(x, t) ≥ 0 in B ̄ 1 × ( t 1 , T * ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq11_HTML.gif.

        Direct calculation shows
        d ln u d t = 1 1 - q Δ u 1 - q + q u - q - 1 | u | 2 + a ( x ) g ( t ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ28_HTML.gif
        (3.17)

        Let λ1 be the first eigenvalue of -Δ in H 0 1 ( B 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq12_HTML.gif and by φ1 > 0 the corresponding eigenfunction, normalized by B 1 a ( x ) φ 1 ( x ) d x = 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq13_HTML.gif. Set G 1 ( t ) = t 1 t g ( s ) d s http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq14_HTML.gif. Clearly, limtT*G(t)/G1(t) = 1.

        Multiplying both sides of Equation (3.16) by φ1 and integrating over B1 × (t1, t), we get
        B 1 ( ln u ) φ 1 d x - B 1 ( ln u ( x , t 1 ) ) φ 1 d x = - λ t 1 t B 1 u 1 - q φ d x d s + G ( t ) + t 1 t B 1 q u - q - 1 | u | 2 φ d x d s , t 1 < t < T * . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ29_HTML.gif
        (3.18)

        The result of case (ii) follows by analogy with the argument used in the proof of case (i).

        Proof of Theorem 2.2
        1. (i)
          By Lemma 3.2 (i), we infer that
          u ( 0 , t ) ~ ( ( 1 - q - p ) a ( 0 ) ) 1 ( 1 - q - p ) G ( t ) , as  t T * , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equac_HTML.gif
           
        hence
        G ( t ) = g ( t ) = u q ( 0 , t ) ~ ( ( 1 - q - p ) a ( 0 ) ) q ( 1 - q - p ) G ( t ) q ( 1 - q - p ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ30_HTML.gif
        (3.19)
        Integrating equivalence (3.18) between t and T*, we obtain
        G ( t ) ~ ( a ( 0 ) ( 1 - q - p ) ) - 1 ( a ( 0 ) ( p + q + s - 1 ) ( T * - t ) ) ( 1 - q - p ) ( 1 - p - q - s ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ31_HTML.gif
        (3.20)
        Using Lemma 3.2 (i) and substituting p = (m - 1)/m, q = q1/m, s = s1/m, t* = , and u(x, t) = v m (x, τ) into (3.19), we complete the proof of Theorem 2.2 (i).
        1. (ii)
          To obtain the blow-up rate of the exponent type, we need to be more careful in this case, since exponentiation of equivalents is not permitted. Similar to the proof of Theorem 3 in [14] and Lemma 2.3 in [16], we get
          lim t T * G ( t ) = a - 1 ( 0 ) ln ( T * - t ) ( - 1 s ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_Equ32_HTML.gif
          (3.21)
           

        Thanks to Lemma 3.2(ii) and (3.20), we then get the desired result.

        4 Discussion

        This article deals with the porous medium equation with local and localized source terms, represented by two factors v q 1 ( x , τ ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq15_HTML.gif and v s 1 ( 0 , τ ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2011-57/MediaObjects/13661_2011_Article_102_IEq16_HTML.gif, respectively. As we all know that, in the absence of weight function, the solutions of model (1.5)-(1.7) have a global blow-up and the rate of blow-up is uniform in all compact subsets of the domain. A natural question is what happens in the model (1.5)-(1.7), where the source term is the product of localized source, local source, and weight function. It is shown by Theorem 2.2 that if 0 ≤ q1 ≤ 1, this equation possesses uniform blow-up profiles. In other words, the localized term plays a leading role in the blow-up profile for this case. Moreover, the blow-up rate estimates in time and space is obtained.

        Declarations

        Acknowledgements

        The authors thank the anonymous referee for their constructive and valuable comments, which helped in improving the presentation of this study. This study was supported by the National Natural Science Foundation of China (60972001), the National Key Technologies R & D Program of China (2009BAG13A06), the Scientific Innovation Research of College Graduate in Jiangsu Province (CXZZ_0163), and the Scientific Research Foundation of Graduate School of Southeast University (YBJJ1140).

        Authors’ Affiliations

        (1)
        School of Transportation, Southeast University
        (2)
        School of Automation, Southeast University
        (3)
        Department of Mathematics, Southeast University

        References

        1. Diaz J, Kerser R: On a nonlinear degenerate parabolic equation in infiltration or evaporation through a porous medium. J Diff Equ 1987, 69: 368–403. 10.1016/0022-0396(87)90125-2View Article
        2. Furter J, Grinfeld M: Local vs. nonlocal interactions in population dynamics. J Math Biol 1989, 27: 65–80. 10.1007/BF00276081MathSciNetView Article
        3. Okada A, Fukuda I: Total versus single point blow-up of solution of a semilinear parabolic equation with localized reaction. J Math Anal Appl 2003, 281: 485–500. 10.1016/S0022-247X(03)00133-1MathSciNetView Article
        4. Cantrell R, Cosner C: Diffusive logistic equation with indefinite weights: population models in disrupted environments. II SIAM J Math Anal 1991, 22: 1043–1064. 10.1137/0522068MathSciNetView Article
        5. Levine H: The role of critical exponents in blow-up theorem. SIMA Rev 1990, 32: 268–288.
        6. Anderson J: Local existence and uniqueness of solutions of degenerate parabolic equations. Commun Partial Diff Equ 1991, 16: 105–143. 10.1080/03605309108820753View Article
        7. Chen Y, Liu Q, Gao H: Bounedeness of global solutions of a porous medium equation with a localized source. Nonlinear Anal 2006, 64: 2168–2182. 10.1016/j.na.2005.08.004MathSciNetView Article
        8. Fukuda I, Suzuki R: Blow-up bebavior for a nolinear beat equation with a localized source in a ball. J Diff Equ 2005, 218: 273–291. 10.1016/j.jde.2005.02.013MathSciNetView Article
        9. Chen Y, Liu Q, Gao H: Bounedeness of global positive solutions of a porous medium equation with a moving localized source. J Math Anal Appl 2007, 333: 1008–1023. 10.1016/j.jmaa.2006.11.048MathSciNetView Article
        10. Souple P: Blow-up in non-local reaction-diffusion equations. SIAM J Math Anal 1998, 29(6):1301–1334. 10.1137/S0036141097318900MathSciNetView Article
        11. Souple P: Uniform blow-up profiles and boundary for diffusion equations with nonlocal nonlinear source. J Diff Equ 1999, 153: 374–406. 10.1006/jdeq.1998.3535View Article
        12. Cannon R, Yin M: A class of non-linear non-classical parabolic equations. J Diff Equ 1989, 79: 226–288.MathSciNetView Article
        13. Chandam J, Peirce A, Yin H: The blow-up property of solutions to some diffusion equations with localized nonlinear reactions. J Math Anal Appl 1992, 169: 313–328. 10.1016/0022-247X(92)90081-NMathSciNetView Article
        14. Kong L, Wang L, Zheng S: Asymptotic analysis to a parabolic equation with a weighted localized source. Appl Math Comput 2008, 197: 819–827. 10.1016/j.amc.2007.08.016MathSciNetView Article
        15. Chen Y, Xie C: Blow-up for a porous medium equation with a localized source. Appl Math Comput 2004, 159: 79–93. 10.1016/j.amc.2003.10.032MathSciNetView Article
        16. Du L, Xiang Z: A further blow-up analysis for a localized porous medium equation. Appl Math Comput 2006, 179: 200–208. 10.1016/j.amc.2005.11.100MathSciNetView Article
        17. Wang J, Kong L, Zheng S: Asymptotic analysis for a localized nonlinear diffusion equation. Comput Math Appl 2008, 56: 2294–2304. 10.1016/j.camwa.2008.03.057MathSciNetView Article
        18. Friedman A, Mcleod J: Blow-up of positive solutions of semilinear heat equations. Indiana Univ Math J 1985, 34: 425–447. 10.1512/iumj.1985.34.34025MathSciNetView Article
        19. Rouchon P: Boundeness of global solutions of nonlinear diffusion equations with localized reaction term. Diff Integral Equ 2003, 16(9):1083–1092.MathSciNet
        20. Li H, Wang M: Properties of blow-up solutions to a parabolic system with nonlinear localized terms. Discrete Contin Dyn Syst 2005, 13: 683–700.MathSciNetView Article

        Copyright

        © Zeng et al; licensee Springer. 2011

        This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://​creativecommons.​org/​licenses/​by/​2.​0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.