Existence and multiplicity of positive solutions for nonhomogeneous boundary value problems with fractional q-derivatives

  • Yulin Zhao1Email author,

    Affiliated with

    • Haibo Chen2 and

      Affiliated with

      • Qiming Zhang1

        Affiliated with

        Boundary Value Problems20132013:103

        DOI: 10.1186/1687-2770-2013-103

        Received: 17 January 2013

        Accepted: 12 April 2013

        Published: 25 April 2013

        Abstract

        In this paper, we study a class of fractional q-difference equations with nonhomogeneous boundary conditions. By applying the classical tools from functional analysis, sufficient conditions for the existence of single and multiple positive solutions to the boundary value problem are obtained in term of the explicit intervals for the nonhomogeneous term. In addition, some examples to illustrate our results are given.

        MSC:34A08, 34B18, 39A13.

        Keywords

        fractional q-difference equation nonhomogeneous boundary value problem positive solution multiplicity

        1 Introduction

        Fractional differential equations have attracted considerable interest because of its demonstrated applications in various fields of science and engineering including fluid flow, rheology, diffusive transport akin to diffusion, electrical networks, probability [1, 2]. Many researchers have studied the existence of solutions (or positive solutions) to fractional boundary value problems; for example, see [310] and the references therein.

        The early work on q-difference calculus or quantum calculus dates back to Jackson’s papers [11], basic definitions and properties of quantum calculus can be found in the book [12]. For some recent existence results on q-difference equations, we refer to [1315] and the references therein.

        The fractional q-difference calculus had its origin in the works by Al-Salam [16] and Agarwal [17]. More recently, there seems to be new interest in the study of this subject and many new developments were made in this theory of fractional q-difference calculus [1822]. Specifically, fractional q-difference equations have attracted the attentions of several researchers. Some recent work on the existence theory of fractional q-difference equations can be found in [20, 2331]. However, the study of boundary value problems for nonlinear fractional q-difference equations is still in the initial stage and many aspects of this topic need to be explored.

        By using a fixed-point theorem in a cone, M. El-Shahed and F. Al-Askar [25] were concerned with the existence of positive solutions to nonlinear q-difference equation:
        { D q α C u ( t ) + a ( t ) f ( u ( t ) ) = 0 , 0 < t < 1 , 2 < α 3 , u ( 0 ) = D q 2 u ( 0 ) = 0 , a D q u ( 1 ) + b D q 2 u ( 1 ) = 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equa_HTML.gif

        where a , b 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq1_HTML.gif and D q α C http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq2_HTML.gif is the fractional q-derivatives of the Caputo type.

        In [27], Graef and Kong investigated the boundary value problem with fractional q-derivatives
        { ( D q α u ) ( t ) + f ( t , u ( t ) ) = 0 , 0 < t < 1 , n 1 < α n , n N , ( D q i u ) ( 0 ) = 0 , i = 0 , , n 2 , b D q u ( 1 ) = j = 1 m a j D q u ( t j ) + λ , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equb_HTML.gif

        where λ 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq3_HTML.gif is a parameter, and the uniqueness, existence and nonexistence of positive solutions are considered in terms of different ranges of λ.

        By applying the Banach contraction principle, Krasnoselskii’s fixed-point theorem, and the Leray-Schauder nonlinear alternative, Ahmad, Ntouyas and Purnaras [29] studied the existence of solution for the following nonlinear fractional q-difference equation with nonlocal boundary conditions:
        { ( C D q α u ) ( t ) = f ( t , u ( t ) ) , 0 t 1 , 1 < α 2 , a 1 u ( 0 ) b 1 D q u ( 0 ) = c 1 u ( η 1 ) , a 2 u ( 1 ) + b 2 D q u ( 1 ) = c 2 u ( η 2 ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equc_HTML.gif

        where D q α C http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq2_HTML.gif is the fractional q-derivative of the Caputo type, and a i , b i , c i , η i R http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq4_HTML.gif.

        Recently, in [32], the authors investigate the following singular semipositone integral boundary value problem for fractional q-derivatives equation:
        { ( D q α u ) ( t ) + f ( t , u ( t ) ) = 0 , t ( 0 , 1 ) , 2 < α 3 , u ( 0 ) = ( D q u ) ( 0 ) = 0 , u ( 1 ) = μ 0 1 u ( s ) d q s , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equd_HTML.gif

        where 0 < μ < [ α ] q http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq5_HTML.gif, D q α http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq6_HTML.gif is the q-derivative of Riemann-Liouville type of order α, f : [ 0 , 1 ] × ( 0 , + ) ( , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq7_HTML.gif is continuous and semipositone, and may be singular at u = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq8_HTML.gif.

        Since finding positive solutions of boundary value problems is interest in various fields of sciences, fractional q-calculus equations has tremendous potential for applications. In this paper, we will deal with the following nonhomogeneous boundary value problem with fractional q-derivatives:
        { ( D q α u ) ( t ) + f ( t , u ( t ) ) = 0 , t ( 0 , 1 ) , u ( 0 ) = ( D q u ) ( 0 ) = 0 , γ ( D q u ) ( 1 ) + β ( D q 2 u ) ( 1 ) = λ , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ1_HTML.gif
        (1.1)

        where q ( 0 , 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq9_HTML.gif, 2 < α 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq10_HTML.gif, γ 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq11_HTML.gif, β > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq12_HTML.gif, and λ is a parameter, D q α http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq6_HTML.gif is the q-derivative of Riemann-Liouville type of order α, f : [ 0 , 1 ] × R R http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq13_HTML.gif is continuous. In the present work, we gave the corresponding Green’s function of the boundary value problem (1.1) and its properties. By using the generalized Banach contraction principle and Krasnoselskii’s fixed-point theorem, the uniqueness, existence, and multiplicity of positive solution to the BVP (1.1) are obtained in term of the explicit intervals for the nonhomogeneous term. Our results are different from those of [25, 27].

        2 Preliminaries on q-calculus and lemmas

        For the convenience of the reader, below we cite some definitions and fundamental results on q-calculus as well as the fractional q-calculus. The presentation here can be found in, for example, [12, 18, 20, 22].

        Let q ( 0 , 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq9_HTML.gif and define
        [ a ] q = 1 q a 1 q , a R . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Eque_HTML.gif
        The q-analogue of the power function ( a b ) n http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq14_HTML.gif with n N 0 : = { 0 , 1 , 2 , } http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq15_HTML.gif is
        ( a b ) ( 0 ) = 1 , ( a b ) ( n ) = k = 0 n 1 ( a b q k ) , n N , a , b R . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equf_HTML.gif
        More generally, if α R http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq16_HTML.gif, then
        ( a b ) ( α ) = a α k = 0 a b q k a b q α + k , a 0 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ2_HTML.gif
        (2.1)
        Clearly, if b = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq17_HTML.gif, then a ( α ) = a α http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq18_HTML.gif. The q-gamma function is defined by
        Γ q ( x ) = ( 1 q ) ( x 1 ) ( 1 q ) x 1 , x R { 0 , 1 , 2 , } , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equg_HTML.gif

        and satisfies Γ q ( x + 1 ) = [ x ] q Γ q ( x ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq19_HTML.gif.

        The q-derivative of a function f is defined by
        ( D q f ) ( x ) = f ( q x ) f ( x ) ( q 1 ) x , ( D q f ) ( 0 ) = lim x 0 ( D q f ) ( x ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equh_HTML.gif
        and q-derivatives of higher order by
        ( D q 0 f ) ( x ) = f ( x ) , ( D q n f ) ( x ) = D q ( D q n 1 f ) ( x ) , n N . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equi_HTML.gif
        The q-integral of a function f defined in the interval [ 0 , b ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq20_HTML.gif is given by
        ( I q f ) ( x ) = 0 x f ( s ) d q s = x ( 1 q ) k = 0 f ( x q k ) q k , x [ 0 , b ] . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equj_HTML.gif
        If a [ 0 , b ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq21_HTML.gif and f is defined in the interval [ 0 , b ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq20_HTML.gif, then its integral from a to b is defined by
        a b f ( s ) d q s = 0 b f ( s ) d q s 0 a f ( s ) d q s . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equk_HTML.gif
        Similar to that for derivatives, an operator I q n http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq22_HTML.gif is given by
        ( I q 0 f ) ( x ) = f ( x ) , ( I q n f ) ( x ) = I q ( I q n 1 f ) ( x ) , n N . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equl_HTML.gif
        The fundamental theorem of calculus applies to these operators I q http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq23_HTML.gif and D q http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq24_HTML.gif, i.e.,
        ( D q I q f ) ( x ) = f ( x ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equm_HTML.gif
        and if f is continuous at x = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq25_HTML.gif, then
        ( I q D q f ) ( x ) = f ( x ) f ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ3_HTML.gif
        (2.2)
        The following formulas will be used later, namely, the integration by parts formula:
        0 x f ( s ) ( D q g ) ( s ) d q s = [ f ( s ) g ( s ) ] s = 0 s = x 0 x ( D q f ) ( s ) g ( q s ) d q s , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equn_HTML.gif
        and
        http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ4_HTML.gif
        (2.3)
        http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ5_HTML.gif
        (2.4)
        http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ6_HTML.gif
        (2.5)
        http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ7_HTML.gif
        (2.6)

        where D q t http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq26_HTML.gif denotes the derivative with respect to the variable t.

        Definition 2.1 Let α 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq27_HTML.gif and f be a function defined on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq28_HTML.gif. The fractional q-integral of Riemann-Liouville type is ( I q 0 f ) ( x ) = f ( x ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq29_HTML.gif and
        ( I q α f ) ( x ) = 1 Γ q ( α ) 0 x ( x q s ) ( α 1 ) f ( s ) d q s , α > 0 , x [ 0 , 1 ] . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equo_HTML.gif
        Definition 2.2 The fractional q-derivative of the Riemann-Liouville type of order α 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq27_HTML.gif is defined by ( D q 0 f ) ( x ) = f ( x ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq30_HTML.gif and
        ( D q α f ) ( x ) = ( D q [ α ] I q [ α ] α f ) ( x ) , α > 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equp_HTML.gif

        where [ α ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq31_HTML.gif is the smallest integer greater than or equal to α.

        Lemma 2.3 ([20])

        Assume that α 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq27_HTML.gif and a b t http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq32_HTML.gif, then ( t a ) ( α ) ( t b ) ( α ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq33_HTML.gif.

        Lemma 2.4 Let α , β 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq34_HTML.gif and f be a function defined on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif. Then the following formulas hold:
        1. (1)

          ( I q β I q α f ) ( x ) = ( I q α + β f ) ( x ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq36_HTML.gif,

           
        2. (2)

          ( D q α I q α f ) ( x ) = f ( x ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq37_HTML.gif.

           

        Lemma 2.5 ([20])

        Let α > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq38_HTML.gif and n be a positive integer. Then the following equality holds:
        ( I q α D q n f ) ( x ) = ( D q n I q α f ) ( x ) k = 0 n 1 x α n + k Γ q ( α + k n + 1 ) ( D q k f ) ( 0 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equq_HTML.gif

        Lemma 2.6 ([22])

        Let α R + http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq39_HTML.gif, λ ( 1 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq40_HTML.gif, the following is valid:
        I q α ( ( t a ) ( λ ) ) = Γ q ( λ + 1 ) Γ q ( α + λ + 1 ) ( t a ) ( α + λ ) , 0 < a < t < b . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equr_HTML.gif
        Particularly, for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq41_HTML.gif, a = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq42_HTML.gif, using q-integration by parts, we have
        ( I q α 1 ) ( t ) = 1 Γ q ( α ) 0 t ( t q s ) ( α 1 ) d q s = 1 Γ q ( α ) 0 t D q s ( ( t s ) ( α ) ) [ α ] q d q s = 1 Γ q ( α + 1 ) 0 t s D q ( ( t s ) ( α ) ) d q s = 1 Γ q ( α + 1 ) t ( α ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equs_HTML.gif
        Obviously, we have 0 t ( t q s ) ( α 1 ) d q s = 1 [ α ] q t ( α ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq43_HTML.gif, and
        0 t ( 1 q s ) ( α 1 ) d q s = 0 t D q s ( ( 1 s ) ( α ) ) [ α ] q d q s = 1 [ α ] q 0 t s D q ( ( 1 s ) ( α ) ) d q s = 1 [ α ] q [ 1 ( 1 t ) ( α ) ] . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equt_HTML.gif

        In order to define the solution for the problem (1.1), we need the following lemmas.

        Lemma 2.7 For given y C [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq44_HTML.gif, the unique solution of the boundary value problem
        ( D q α u ) ( t ) + y ( t ) = 0 , t ( 0 , 1 ) , 2 < α 3 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ8_HTML.gif
        (2.7)
        subject to the boundary conditions
        u ( 0 ) = ( D q u ) ( 0 ) = 0 , γ ( D q u ) ( 1 ) + β ( D q 2 u ) ( 1 ) = λ , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ9_HTML.gif
        (2.8)
        is given by
        u ( t ) = 0 1 G ( t , q s ) y ( s ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ10_HTML.gif
        (2.9)
        where
        G ( t , s ) = { γ t α 1 ( 1 s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 s ) ( α 3 ) ( γ + [ α 2 ] q β ) Γ q ( α ) ( t s ) ( α 1 ) Γ q ( α ) , 0 s t 1 , γ t α 1 ( 1 s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 s ) ( α 3 ) ( γ + [ α 2 ] q β ) Γ q ( α ) , 0 t s 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ11_HTML.gif
        (2.10)
        Proof Since 2 < α 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq10_HTML.gif, we put n = 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq45_HTML.gif. In view of Definition 2.1 and Lemma 2.4, we see that
        ( D q α u ) ( t ) = y ( t ) ( I q α D q 3 I q 3 α u ) ( t ) = ( I q α y ) ( t ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equu_HTML.gif
        Then it follows from Lemma 2.5 that the solution u ( t ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq46_HTML.gif of (2.7) and (2.8) is given by
        u ( t ) = c 1 t α 1 + c 2 t α 2 + c 3 t α 3 0 t ( t q s ) ( α 1 ) Γ q ( α ) y ( s ) d q s , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ12_HTML.gif
        (2.11)

        for some constants c 1 , c 2 , c 3 R http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq47_HTML.gif. From u ( 0 ) = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq48_HTML.gif, we have c 3 = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq49_HTML.gif.

        Differentiating both sides of (2.11) and with the help of (2.4) and (2.6), we obtain,
        ( D q u ) ( t ) = [ α 1 ] q c 1 t α 2 + [ α 2 ] q c 2 t α 3 0 t [ α 1 ] q ( t q s ) ( α 2 ) Γ q ( α ) y ( s ) d q s , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equv_HTML.gif
        and
        ( D q 2 u ) ( t ) = [ α 1 ] q [ α 2 ] q c 1 t α 3 + [ α 2 ] q [ α 3 ] q c 2 t α 4 0 t [ α 1 ] q [ α 2 ] q ( t q s ) ( α 3 ) Γ q ( α ) y ( s ) d q s . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equw_HTML.gif
        Then by the boundary condition ( D q u ) ( 0 ) = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq50_HTML.gif, we get c 2 = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq51_HTML.gif. Using the boundary condition γ ( D q u ) ( 1 ) + β ( D q 2 u ) ( 1 ) = λ http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq52_HTML.gif, we get
        c 1 = 1 γ + [ α 2 ] q β × ( γ 0 1 ( 1 q s ) ( α 2 ) Γ q ( α ) y ( s ) d q s + β 0 1 [ α 2 ] q ( 1 q s ) ( α 3 ) Γ q ( α ) y ( s ) d q s + λ [ α 1 ] q ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equx_HTML.gif
        Hence, we have
        u ( t ) = t α 1 γ + [ α 2 ] q β ( γ 0 1 ( 1 q s ) ( α 2 ) Γ q ( α ) y ( s ) d q s + β 0 1 [ α 2 ] q ( 1 q s ) ( α 3 ) Γ q ( α ) y ( s ) d q s + λ [ α 1 ] q ) 0 t ( t q s ) ( α 1 ) Γ q ( α ) y ( s ) d q s = 0 1 G ( t , q s ) y ( s ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equy_HTML.gif

        This completes the proof of the lemma. □

        Lemma 2.8 The function G ( t , s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq53_HTML.gif defined by (2.10) satisfies the following conditions:
        1. (i)

          G ( t , q s ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq54_HTML.gif, and G ( t , q s ) G ( 1 , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq55_HTML.gif for all 0 t , s 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq56_HTML.gif.

           
        2. (ii)

          G ( t , q s ) t α 1 G ( 1 , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq57_HTML.gif for all 0 t , s 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq56_HTML.gif.

           
        Proof We start by defining the following two functions:
        g 1 ( t , s ) = γ t α 1 ( 1 s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 s ) ( α 3 ) ( γ + [ α 2 ] q β ) Γ q ( α ) , 0 t s 1 , g 2 ( t , s ) = g 1 ( t , s ) ( t s ) ( α 1 ) Γ q ( α ) , 0 s t 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equz_HTML.gif
        Obviously, g 1 ( t , q s ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq58_HTML.gif. Now g 2 ( 0 , q s ) = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq59_HTML.gif, and for t 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq60_HTML.gif
        g 2 ( t , q s ) = γ t α 1 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) Γ q ( α ) ( t q s ) ( α 1 ) Γ q ( α ) = t α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s t ) ( α 1 ) ) ( γ + [ α 2 ] q β ) Γ q ( α ) t α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) ) ( γ + [ α 2 ] q β ) Γ q ( α ) = t α 1 ( γ ( ( 1 q s ) ( α 2 ) ( 1 q s ) ( α 1 ) ) + [ α 2 ] q β ( ( 1 q s ) ( α 3 ) ( 1 q s ) ( α 1 ) ) ) ( γ + [ α 2 ] q β ) Γ q ( α ) 0 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaa_HTML.gif

        Therefore, G ( t , q s ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq54_HTML.gif.

        Moreover, for s ( 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq61_HTML.gif, it follows from (2.4) and Lemma 2.3 that
        D q t g 2 ( t , q s ) = [ α 1 ] q ( γ + [ α 2 ] q β ) Γ q ( α ) ( γ t α 2 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 2 ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( t q s ) ( α 2 ) ) [ α 1 ] q t α 2 ( γ + [ α 2 ] q β ) Γ q ( α ) ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 2 ) ) = [ α 1 ] q [ α 2 ] q β t α 2 ( γ + [ α 2 ] q β ) Γ q ( α ) ( ( 1 q s ) ( α 3 ) ( 1 q s ) ( α 2 ) ) 0 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equab_HTML.gif

        which implies that g 2 ( t , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq62_HTML.gif is an increasing function with respect to t. It is clear that g 1 ( t , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq63_HTML.gif is increasing in t. Therefore, G ( t , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq64_HTML.gif is an increasing function of t for all s ( 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq61_HTML.gif, and so G ( t , q s ) G ( 1 , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq55_HTML.gif.

        When 0 t q s 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq65_HTML.gif, then
        G ( t , q s ) = γ t α 1 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) Γ q ( α ) G ( q s , q s ) G ( 1 , q s ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equac_HTML.gif
        Finally, we prove part (ii). When 0 q s t 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq66_HTML.gif, we have
        G ( t , q s ) G ( 1 , q s ) = γ t α 1 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( t q s ) ( α 1 ) γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) γ t α 1 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) t α 1 ( 1 q s ) ( α 1 ) γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) = t α 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equad_HTML.gif
        If 0 t q s 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq65_HTML.gif, then we have
        G ( t , q s ) G ( 1 , q s ) = γ t α 1 ( 1 q s ) ( α 2 ) + [ α 2 ] q β t α 1 ( 1 q s ) ( α 3 ) γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) t α 1 [ γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) ] γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ( γ + [ α 2 ] q β ) ( 1 q s ) ( α 1 ) = t α 1 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equae_HTML.gif

        which implies that part (ii) holds. This completes the proof of the lemma. □

        Remark 2.9 If we let 0 < τ < 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq67_HTML.gif, then
        min t [ τ , 1 ] G ( t , q s ) τ α 1 G ( 1 , q s ) , for  s [ 0 , 1 ] . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaf_HTML.gif

        According to [20], we may take τ = q n http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq68_HTML.gif, n N http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq69_HTML.gif.

        3 The main results

        Let X = C ( [ 0 , 1 ] ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq70_HTML.gif be a Banach space endowed with the norm u X = max 0 t 1 | u ( t ) | http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq71_HTML.gif. Define the cone P X http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq72_HTML.gif by P = { u X : u ( t ) 0 , 0 t 1 } http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq73_HTML.gif.

        Define the operator T : P X http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq74_HTML.gif as follows:
        ( T u ) ( t ) = 0 1 G ( t , q s ) f ( s , u ( s ) ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ13_HTML.gif
        (3.1)
        Theorem 3.1 Assume that f : [ 0 , 1 ] × [ 0 , + ) [ 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq75_HTML.gif is continuous and there exists a nonnegative function h C [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq76_HTML.gif such that
        | f ( t , u ) f ( t , v ) | h ( t ) | u v | , t [ 0 , 1 ] , u , v [ 0 , + ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ14_HTML.gif
        (3.2)
        Then the BVP (1.1) has a unique positive solution for any λ ( 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq77_HTML.gif, provided
        0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s < ( γ + [ α 2 ] q β ) Γ q ( α ) 2 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ15_HTML.gif
        (3.3)

        If, in addition, f ( t , 0 ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq78_HTML.gif on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif, then the conclusion is true for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif.

        Proof We will show that under the assumptions (3.2) and (3.3), T m http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq80_HTML.gif is a contraction operator for m sufficiently large.

        By the definition of G ( t , q s ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq64_HTML.gif, for u , v P http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq81_HTML.gif, we have
        | ( T u ) ( t ) ( T v ) ( t ) | 0 1 G ( t , q s ) | f ( s , u ( s ) ) f ( s , v ( s ) ) | d q s 0 1 t α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) ( γ + [ α 2 ] q β ) Γ q ( α ) h ( s ) d q s u v X = t α 1 u v X ( γ + [ α 2 ] q β ) Γ q ( α ) 0 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s = Λ 1 t α 1 ( γ + [ α 2 ] q β ) Γ q ( α ) u v X , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equag_HTML.gif

        where Λ 1 = 0 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq82_HTML.gif.

        Consequently,
        | ( T 2 u ) ( t ) ( T 2 v ) ( t ) | 0 1 G ( t , q s ) | f ( s , ( T u ) ( s ) ) f ( s , ( T v ) ( s ) ) | d q s Λ 1 u v X ( γ + [ α 2 ] q β ) Γ q ( α ) 0 1 G ( t , q s ) s α 1 h ( s ) d q s Λ 1 t α 1 u v X ( γ + [ α 2 ] q β ) 2 [ Γ q ( α ) ] 2 0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s = Λ 1 Λ 2 t α 1 ( γ + [ α 2 ] q β ) 2 [ Γ q ( α ) ] 2 u v X , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equah_HTML.gif

        where Λ 2 = 0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq83_HTML.gif.

        By introduction, we get
        | ( T m u ) ( t ) ( T m v ) ( t ) | Λ 1 Λ 2 m 1 t α 1 ( γ + [ α 2 ] q β ) m [ Γ q ( α ) ] m u v X . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equai_HTML.gif
        From the condition (3.3), we have
        Λ 1 Λ 2 m 1 ( γ + [ α 2 ] q β ) m [ Γ q ( α ) ] m = Λ 1 Λ 2 [ Λ 2 ( γ + [ α 2 ] q β ) Γ q ( α ) ] m Λ 1 Λ 2 ( 1 2 ) m < 1 4 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaj_HTML.gif
        for m sufficiently large. So, we get
        ( T m u ) ( t ) ( T m v ) ( t ) X < 1 4 u v X . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equak_HTML.gif

        Hence, it follows from the generalized Banach contraction principle that the BVP (1.1) has a unique positive solution for any λ ( 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq77_HTML.gif. If λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif, then the condition f ( t , 0 ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq84_HTML.gif on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif and Lemma 2.8 imply that u ( t ) > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq85_HTML.gif in ( 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq86_HTML.gif. This completes the proof of the theorem. □

        Remark 3.2 When h ( t ) h 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq87_HTML.gif is a constant, the condition (3.2) reduces to a Lipschitz condition.

        Our next existence results is based on Krasnoselskii’s fixed-point theorem [33].

        Lemma 3.3 (Krasnoselskii’s)

        Let E be a Banach space, and let P E http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq88_HTML.gif be a cone. Assume Ω 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq89_HTML.gif, Ω 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq90_HTML.gif are open subsets of E with θ Ω 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq91_HTML.gif, Ω ¯ 1 Ω 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq92_HTML.gif and let T : P ( Ω ¯ 2 Ω 1 ) P http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq93_HTML.gif be a completely continuous operator such that, either
        1. (1)

          T u u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq94_HTML.gif, u P Ω 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq95_HTML.gif and T u u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq96_HTML.gif, u P Ω 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq97_HTML.gif, or

           
        2. (2)

          T u u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq96_HTML.gif, u P Ω 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq95_HTML.gif and T u u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq94_HTML.gif, u P Ω 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq97_HTML.gif.

           

        Then T has at least one fixed point in P ( Ω ¯ 2 Ω 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq98_HTML.gif.

        Define a cone K X http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq99_HTML.gif by
        K = { u X : u ( t ) 0 , u ( t ) t α 1 u , t [ 0 , 1 ] } . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equal_HTML.gif

        Obviously, K is a cone of nonnegative functions in X.

        Lemma 3.4 The operator T : K K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq100_HTML.gif is completely continuous.

        Proof Firstly, we prove that T ( K ) K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq101_HTML.gif. By (2.9) and Lemma 2.8, we have
        T u = max 0 t 1 { 0 1 G ( t , q s ) f ( s , u ( s ) ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q } 0 1 G ( 1 , q s ) f ( s , u ( s ) ) d q s + λ ( γ + [ α 2 ] q β ) [ α 1 ] q . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equam_HTML.gif
        On the other hand,
        ( T u ) ( t ) t α 1 ( 0 1 G ( 1 , q s ) f ( s , u ( s ) ) d q s + λ ( γ + [ α 2 ] q β ) [ α 1 ] q ) t α 1 T u . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equan_HTML.gif

        Hence, we have T ( K ) K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq101_HTML.gif.

        Next, we show that T is uniformly bounded. For fixed r > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq102_HTML.gif, consider a bounded subset K r http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq103_HTML.gif of K defined by K r = { u K : u r , r > 0 } http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq104_HTML.gif, and let M = max 0 u r | f ( t , u ) | + 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq105_HTML.gif. Then for u K r http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq106_HTML.gif, we get
        | ( T u ) ( t ) | 0 1 G ( t , q s ) | f ( s , u ( s ) ) | d q s + | λ | t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q M 0 1 G ( 1 , q s ) d q s + | λ | ( γ + [ α 2 ] q β ) [ α 1 ] q < + , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equao_HTML.gif

        which implies that T ( K r ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq107_HTML.gif is bounded.

        Finally, we show that T is equicontinuous. For all ε > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq108_HTML.gif, setting
        δ = min { ε ω ( α 1 ) , 1 2 ( ε ω ) 1 α 1 } , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equap_HTML.gif
        where
        ω = M ( γ + [ α 2 ] q β ) + | λ | Γ q ( α ) ( γ + [ α 2 ] q β ) Γ q ( α ) [ α 1 ] q . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaq_HTML.gif
        For any u K r http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq106_HTML.gif, we can prove that if t 1 , t 2 [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq109_HTML.gif and 0 < t 2 t 1 < δ http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq110_HTML.gif, then
        | ( T u ) ( t 2 ) ( T u ) ( t 2 ) | < ε . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equar_HTML.gif
        In fact, we have
        | ( T u ) ( t 2 ) ( T u ) ( t 1 ) | 0 1 | G ( t 2 , q s ) G ( t 1 , q s ) | f ( s , u ( s ) ) d q s + | λ | ( t 2 α 1 t 1 α 1 ) ( γ + [ α 2 ] q β ) [ α 1 ] q M 0 1 | G ( t 2 , q s ) G ( t 1 , q s ) | d q s + | λ | ( t 2 α 1 t 1 α 1 ) ( γ + [ α 2 ] q β ) [ α 1 ] q M ( 0 t 1 | G ( t 2 , q s ) G ( t 1 , q s ) | d q s + t 1 t 2 | G ( t 2 , q s ) G ( t 1 , q s ) | d q s + t 2 1 | G ( t 2 , q s ) G ( t 1 , q s ) | d q s ) + | λ | ( t 2 α 1 t 1 α 1 ) ( γ + [ α 2 ] q β ) [ α 1 ] q M γ + [ α 2 ] q β ( 0 t 1 γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) Γ q ( α ) ( t 2 α 1 t 1 α 1 ) d q s + t 1 t 2 γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) Γ q ( α ) ( t 2 α 1 t 1 α 1 ) d q s + t 2 1 γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) Γ q ( α ) ( t 2 α 1 t 1 α 1 ) d q s ) + | λ | ( t 2 α 1 t 1 α 1 ) ( γ + [ α 2 ] q β ) [ α 1 ] q = M ( t 2 α 1 t 1 α 1 ) ( γ + [ α 2 ] q β ) Γ q ( α ) 0 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) d q s + | λ | ( γ + [ α 2 ] q β ) [ α 1 ] q ( t 2 α 1 t 1 α 1 ) = M ( γ + [ α 2 ] q β ) + | λ | Γ q ( α ) ( γ + [ α 2 ] q β ) Γ q ( α ) [ α 1 ] q ( t 2 α 1 t 1 α 1 ) = ω ( t 2 α 1 t 1 α 1 ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equas_HTML.gif
        If δ t 1 < t 2 < 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq111_HTML.gif, then
        | ( T u ) ( t 2 ) ( T u ) ( t 1 ) | ω ( t 2 α 1 t 1 α 1 ) < ω ( α 1 ) ( t 2 t 1 ) < ω ( α 1 ) δ ε . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equat_HTML.gif
        If 0 t 1 < δ http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq112_HTML.gif, t 2 < 2 δ http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq113_HTML.gif, then
        | ( T u ) ( t 2 ) ( T u ) ( t 1 ) | ω ( t 2 α 1 t 1 α 1 ) < ω t 2 α 1 < ω ( 2 δ ) α 1 ε . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equau_HTML.gif

        By means of Arzela-Ascoli theorem, T : K K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq100_HTML.gif is completely continuous.

        For the sake of convenience, we introduce the following weight functions:
        ϕ ( r ) = max { f ( t , u ( t ) ) : ( t , u ) [ 0 , 1 ] × [ 0 , r ] } , φ ( r ) = min { f ( t , u ( t ) ) : ( t , u ) [ τ , 1 ] × [ τ α 1 r , r ] } , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equav_HTML.gif
        and set
        l = ( 0 1 G ( 1 , q s ) d q s ) 1 , L = ( τ α 1 τ 1 G ( 1 , q s ) d q s ) 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaw_HTML.gif

         □

        Theorem 3.5 Suppose that there exists two positive numbers ξ 1 < ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq114_HTML.gif such that one of the following conditions is satisfied

        ( H 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq115_HTML.gif) φ ( ξ 1 ) ξ 1 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq116_HTML.gif, ϕ ( ξ 2 ) l 2 ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq117_HTML.gif;

        ( H 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq118_HTML.gif) ϕ ( ξ 1 ) l 2 ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq119_HTML.gif, φ ( ξ 2 ) ξ 2 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq120_HTML.gif.

        Then the BVP (1.1) has at least one positive solution u K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq121_HTML.gif, such that ξ 1 u ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq122_HTML.gif for λ ( 0 , ( γ + [ α 2 ] q β ) [ α 1 ] q 2 ξ 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq123_HTML.gif. If, in addition, f ( t , 0 ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq78_HTML.gif on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif, then the conclusion is true for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif.

        Proof Because the proofs are similar, we prove only the case ( H 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq115_HTML.gif). Denote Ω ξ 1 = { u X : u < ξ 1 } http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq124_HTML.gif. Then for any u K Ω ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq125_HTML.gif, we get u = ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq126_HTML.gif, 0 u ( t ) ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq127_HTML.gif, 0 t 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq128_HTML.gif, and τ α 1 ξ 1 min τ t 1 t α 1 u u ( t ) u = ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq129_HTML.gif, τ t 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq130_HTML.gif. By assumption ( H 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq115_HTML.gif), we have
        f ( t , u ) φ ( ξ 1 ) ξ 1 L , τ t 1 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equax_HTML.gif
        In view of (2.9) and Lemma 2.8, we have
        T u min τ t 1 { 0 1 G ( t , q s ) f ( s , u ( s ) ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q } min τ t 1 0 1 G ( t , q s ) f ( s , u ( s ) ) d q s τ α 1 τ 1 G ( 1 , q s ) f ( s , u ( s ) ) d q s τ α 1 τ 1 G ( 1 , q s ) d q s ξ 1 L = ξ 1 = u . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equay_HTML.gif
        On the other hand, define Ω ξ 2 = { u X : u < ξ 2 } http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq131_HTML.gif. For any t [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq132_HTML.gif and u K Ω ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq133_HTML.gif, we have u = ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq134_HTML.gif and 0 u ( t ) ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq135_HTML.gif, 0 t 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq128_HTML.gif. Thus,
        f ( t , u ) ϕ ( ξ 2 ) l 2 ξ 2 , for  0 t 1 , 0 u ξ 2 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equaz_HTML.gif
        It follows
        T u = max 0 t 1 { 0 1 G ( t , q s ) f ( s , u ( s ) ) d q s + λ t α 1 ( γ + [ α 2 ] q β ) [ α 1 ] q } 0 1 G ( 1 , q s ) d q s l 2 ξ 2 + λ ( γ + [ α 2 ] q β ) [ α 1 ] q l 1 l 2 ξ 2 + ξ 1 2 < ξ 2 2 + ξ 2 2 = u . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equba_HTML.gif

        By Lemma 3.3, the operator T has at least one fixed point u K ( Ω ¯ ξ 2 Ω ξ 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq136_HTML.gif, and ξ 1 u ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq122_HTML.gif. Since u ( t ) t α 1 u ξ 1 t α 1 > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq137_HTML.gif, 0 < t < 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq138_HTML.gif, then, the solution u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq139_HTML.gif is positive for λ > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq140_HTML.gif. As in the proof of Theorem 3.1, u ( t ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq141_HTML.gif is a positive solution for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif. This completes the proof of the theorem. □

        Theorem 3.6 Suppose that there exists three positive numbers ξ 1 < ξ 2 < ξ 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq142_HTML.gif such that one of the following conditions is satisfied

        ( H 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq143_HTML.gif) φ ( ξ 1 ) ξ 1 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq116_HTML.gif, ϕ ( ξ 2 ) < l 2 ξ 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq144_HTML.gif, φ ( ξ 3 ) ξ 3 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq145_HTML.gif;

        ( H 4 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq146_HTML.gif) ϕ ( ξ 1 ) l 2 ξ 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq119_HTML.gif, φ ( ξ 2 ) > ξ 2 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq147_HTML.gif, ϕ ( ξ 3 ) l 2 ξ 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq148_HTML.gif.

        Then the BVP (1.1) has at least two positive solutions u 1 , u 2 K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq149_HTML.gif such that ξ 1 u 1 < ξ 2 u 2 ξ 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq150_HTML.gif for λ ( 0 , ( γ + [ α 2 ] q β ) [ α 1 ] q 2 ξ 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq151_HTML.gif. If, in addition, f ( t , 0 ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq78_HTML.gif on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif, then the conclusion is true for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif.

        Proof We prove only the case ( H 4 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq146_HTML.gif). Since φ : [ 0 , + ) [ 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq152_HTML.gif is continuous and φ ( ξ 2 ) > ξ 2 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq153_HTML.gif, there exist two positive numbers η 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq154_HTML.gif, η 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq155_HTML.gif such that ξ 1 < η 1 < ξ 2 < η 2 < ξ 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq156_HTML.gif and φ ( η 1 ) η 1 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq157_HTML.gif, φ ( η 2 ) η 2 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq158_HTML.gif. Thus, it follows from the assumption ( H 4 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq146_HTML.gif) that
        ϕ ( ξ 1 ) l 2 ξ 1 , φ ( η 1 ) η 1 L , and φ ( η 2 ) η 2 L , ϕ ( ξ 3 ) l 2 ξ 3 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbb_HTML.gif

        From Theorem 3.5, the operator T has two fixed point u 1 K ( Ω ¯ η 1 Ω ξ 1 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq159_HTML.gif, u 2 K ( Ω ¯ ξ 3 Ω η 2 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq160_HTML.gif with ξ 1 u 1 < ξ 2 < u 2 ξ 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq161_HTML.gif. Therefore, the BVP (1.1) has at least two positive solutions for λ ( 0 , ( γ + [ α 2 ] q β ) [ α 1 ] q 2 ξ 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq162_HTML.gif. As in the proof of Theorem 3.1, u 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq163_HTML.gif, u 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq164_HTML.gif are two positive solutions for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif. This completes the proof of the theorem. □

        Denote the integer part of m by [ m ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq165_HTML.gif. Generally, we have the following theorem.

        Theorem 3.7 Suppose that there exists m + 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq166_HTML.gif positive numbers ξ 1 < ξ 2 < < ξ m + 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq167_HTML.gif such that one of the following conditions is satisfied:

        ( H 5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq168_HTML.gif) φ ( ξ 2 j 1 ) > ξ 2 j 1 L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq169_HTML.gif, ϕ ( ξ 2 j ) < l 2 ξ 2 j http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq170_HTML.gif, j = 1 , 2 , , [ m + 2 2 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq171_HTML.gif;

        ( H 6 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq172_HTML.gif) ϕ ( ξ 2 j 1 ) < l 2 ξ 2 j 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq173_HTML.gif, φ ( ξ 2 j ) > ξ 2 j L http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq174_HTML.gif, j = 1 , 2 , , [ m + 2 2 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq171_HTML.gif.

        Then the BVP (1.1) has at least m positive solutions u i K http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq175_HTML.gif, i = 1 , 2 , , m http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq176_HTML.gif, such that ξ i < u i < ξ i + 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq177_HTML.gif for λ ( 0 , ( γ + [ α 2 ] q β ) [ α 1 ] q 2 ξ 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq123_HTML.gif. If, in addition, f ( t , 0 ) 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq78_HTML.gif on [ 0 , 1 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq35_HTML.gif, then the conclusion is true for λ = 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq79_HTML.gif.

        4 Examples

        Example 4.1 The fractional q-difference boundary value problem
        { D 0.5 2.5 u ( t ) + 2 e t 5 ( 1 + e t ) ( tan 1 u + t 2 + t sin 2 t + 1 ) = 0 , 0 < t < 1 , u ( 0 ) = ( D 0.5 u ) ( 0 ) = 0 , 0.25 ( D 0.5 u ) ( 1 ) + 0.75 ( D 0.5 2 u ) ( 1 ) = λ http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ16_HTML.gif
        (4.1)

        has a unique positive solution for any λ ( 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq178_HTML.gif.

        Proof In this case, α = 2.5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq179_HTML.gif, q = 0.5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq180_HTML.gif, γ = 0.25 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq181_HTML.gif, β = 0.75 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq182_HTML.gif, λ > 0 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq140_HTML.gif. Let
        f ( t , u ) = 2 e t 5 ( 1 + e t ) ( tan 1 u + t 2 + t sin 2 t + 1 ) , ( t , u ) [ 0 , 1 ] × ( 0 , + ) , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbc_HTML.gif
        and h ( t ) = 2 e t 5 ( 1 + e t ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq183_HTML.gif. It is easy to prove that
        | f ( t , u ) f ( t , v ) | h ( t ) | tan 1 u tan 1 v | h ( t ) | u v | , for  ( t , u ) , ( t , v ) [ 0 , 1 ] × [ 0 , + ) . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbd_HTML.gif
        A simple computation showed
        ( γ + [ α 2 ] q β ) Γ q ( α ) = ( 0.25 + [ 0.5 ] 0.5 × 0.75 ) Γ 0.5 ( 2.5 ) 0.9168 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Eqube_HTML.gif
        and
        Λ 2 = 0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) 2 e s 5 ( 1 + e s ) d q s 2 5 0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) d q s 2 5 0 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) d q s 0.3774 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbf_HTML.gif
        which implies that
        0 1 s α 1 ( γ ( 1 q s ) ( α 2 ) + [ α 2 ] q β ( 1 q s ) ( α 3 ) ) h ( s ) d q s < ( γ + [ α 2 ] q β ) Γ q ( α ) 2 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbg_HTML.gif
        Obviously, for any m 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq184_HTML.gif, we have
        Λ 1 Λ 2 m 1 ( γ + [ α 2 ] q β ) m [ Γ q ( α ) ] m 0.3774 0.9168 × 2 m 1 < 0.2058 < 1 4 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbh_HTML.gif

        Thus, Theorem 3.1 implies that the boundary value problem (4.1) has a unique positive solution for any λ ( 0 , + ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq178_HTML.gif. □

        Example 4.2 Consider the following fractional boundary value problem:
        { D 0.5 2.5 u ( t ) + u 2 ( 1 2 cos ( π t π 2 ) + 1 4 ) = 0 , 0 < t < 1 , u ( 0 ) = ( D 0.5 u ) ( 0 ) = 0 , 2 ( D 0.5 u ) ( 1 ) + 7 ( D 0.5 2 u ) ( 1 ) = λ , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equ17_HTML.gif
        (4.2)

        where α = 2.5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq179_HTML.gif, q = 0.5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq180_HTML.gif, γ = 2 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq185_HTML.gif, β = 7 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq186_HTML.gif. Choosing n = 1 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq187_HTML.gif, then τ = q n = 0.5 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq188_HTML.gif.

        By calculation, we get ( γ + [ α 2 ] q β ) Γ q ( α ) 8.1132 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq189_HTML.gif. By Lemma 2.6, Lemma 2.8 and with the aid of a computer, we obtain that
        l = ( 0 1 G ( 1 , 0.5 s ) d 0.5 s ) 1 1.6756 , http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbi_HTML.gif
        and
        L = ( ( 0.5 ) α 1 0.5 1 G ( 1 , 0.5 s ) d 0.5 s ) 1 0.0657 . http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_Equbj_HTML.gif
        Let f ( t , u ) = u 2 ( 1 2 cos ( π t π 2 ) + 1 4 ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq190_HTML.gif. Take ξ 1 = 3 4 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq191_HTML.gif, ξ 2 = 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq192_HTML.gif, then ( γ + [ α 2 ] q β ) [ α 1 ] q 2 ξ 1 2.9578 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq193_HTML.gif, and f ( t , u ) http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq194_HTML.gif satisfies
        1. (i)

          ϕ ( 3 4 ) = max { u 2 ( 1 2 cos ( π t π 2 ) + 1 4 ) : ( t , u ) [ 0 , 1 ] × [ 0 , 3 4 ] } = 27 64 < l 2 ξ 1 0.6283 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq195_HTML.gif;

           
        2. (ii)

          φ ( 3 ) = min { u 2 ( 1 2 cos ( π t π 2 ) + 1 4 ) : ( t , u ) [ 0.5 , 1 ] × [ 1.0608 , 3 ] } 0.2652 > L ξ 2 0.1971 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq196_HTML.gif.

           

        So, by Theorem 3.5, the problem (4.2) has one positive solution u http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq139_HTML.gif such that 1 u X 3 http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq197_HTML.gif for λ ( 0 , 2.9578 ] http://static-content.springer.com/image/art%3A10.1186%2F1687-2770-2013-103/MediaObjects/13661_2013_Article_357_IEq198_HTML.gif.

        Declarations

        Acknowledgements

        Dedicated to Professor Hari M Srivastava.

        The authors are highly grateful for the referees’ careful reading and comments on this paper. The research is supported by the National Natural Science Foundation of China (Grant No. 11271372, 11201138); it is also supported by the Hunan Provincial Natural Science Foundation of China (Grant No. 13JJ3106, 12JJ2004), and the Scientific Research Fund of Hunan Provincial Education Department (Grant No. 12B034).

        Authors’ Affiliations

        (1)
        School of Science, Hunan University of Technology
        (2)
        Department of Mathematics, Central South University

        References

        1. Podlubny I: Fractional Differential Equations. Academic Press, San Diego; 1999.
        2. Kibas AA, Srivastava HM, Trujillo JJ: Theory and Applications of Fractional Differential Equations. Elsevier, Amsterdam; 2006.
        3. Al-Refai M, Ali Hajji M: Monotone iterative sequences for nonlinear boundary value problems of fractional order. Nonlinear Anal. 2011, 74: 3531-3539. 10.1016/j.na.2011.03.006MathSciNetView Article
        4. Ahmad B, Sivasundaram S: On four-point nonlocal boundary value problems of nonlinear integro-differential equations of fractional order. Appl. Math. Comput. 2010, 217: 480-487. 10.1016/j.amc.2010.05.080MathSciNetView Article
        5. Stanék S: The existence of positive solutions of singular fractional boundary value problems. Comput. Math. Appl. 2011, 62: 1379-1388. 10.1016/j.camwa.2011.04.048MathSciNetView Article
        6. Zhao Y, Chen H, Huang L: Existence of positive solutions for nonlinear fractional functional differential equation. Comput. Math. Appl. 2012, 64: 3456-3467. 10.1016/j.camwa.2012.01.081MathSciNetView Article
        7. Agarwal RP, O’Regan D, Stanek S: Positive solutions for Dirichlet problems of singular nonlinear fractional differential equations. J. Math. Anal. Appl. 2010, 371: 57-68. 10.1016/j.jmaa.2010.04.034MathSciNetView Article
        8. Goodrich CS: On discrete sequential fractional boundary value problems. J. Math. Anal. Appl. 2012, 385: 111-124. 10.1016/j.jmaa.2011.06.022MathSciNetView Article
        9. Zhang S: Positive solutions to singular boundary value problem for nonlinear fractional differential equation. Comput. Math. Appl. 2010, 59: 1300-1309. 10.1016/j.camwa.2009.06.034MathSciNetView Article
        10. Zhang X, Liu L, Wu Y: Multiple positive solutions of a singular fractional differential equation with negatively perturbed term. Math. Comput. Model. 2012, 55: 1263-1274. 10.1016/j.mcm.2011.10.006MathSciNetView Article
        11. Jackson FH: q -Difference equations. Am. J. Math. 1970, 32: 305-314.View Article
        12. Kac V, Cheung P: Quantum Calculus. Springer, New York; 2002.View Article
        13. Ernst, T: The history of q-calculus and a new method. UUDM Report 2000:16, Department of Mathematics, Uppsala University, 2000, ISSN:1101-3591
        14. Ahmad B: Boundary-value problems for nonlinear third-order q -difference equations. Electron. J. Differ. Equ. 2011., 2011: Article ID 94
        15. Goodrich CS: Existence of a positive solution to a system of discrete fractional boundary value problems. Appl. Math. Comput. 2011, 217: 4740-4753. 10.1016/j.amc.2010.11.029MathSciNetView Article
        16. Al-Salam WA: Some fractional q -integrals and q -derivatives. Proc. Edinb. Math. Soc. 1966/1967, 15(2):135-140. 10.1017/S0013091500011469MathSciNetView Article
        17. Agarwal RP: Certain fractional q -integrals and q -derivatives. Proc. Camb. Philos. Soc. 1969, 66: 365-370. 10.1017/S0305004100045060View Article
        18. Annaby MH, Mansour IS: q-Fractional Calculus and Equations. Springer, Berlin; 2012.View Article
        19. Atici FM, Eloe PW: Fractional q -calculus on a time scale. J. Nonlinear Math. Phys. 2007, 14: 333-344.MathSciNetView Article
        20. Ferreira RAC: Nontrivial solutions for fractional q -difference boundary value problems. Electron. J. Qual. Theory Differ. Equ. 2010, 70: 1-10.View Article
        21. Goodrich CS: Continuity of solutions to discrete fractional initial value problem. Comput. Math. Appl. 2010, 59: 3489-3499. 10.1016/j.camwa.2010.03.040MathSciNetView Article
        22. Rajković PM, Marinković SD, Stanković MS: Fractional integrals and derivatives in q -calculus. Appl. Anal. Discrete Math. 2007, 1: 311-323. 10.2298/AADM0701311RMathSciNetView Article
        23. Atici FM, Eloe PW: Two-point boundary value problems for finite fractional difference equations. J. Differ. Equ. Appl. 2011, 17: 445-456. 10.1080/10236190903029241MathSciNetView Article
        24. Ferreira RAC: Positive solutions for a class of boundary value problems with fractional q -differences. Comput. Math. Appl. 2011, 61: 367-373. 10.1016/j.camwa.2010.11.012MathSciNetView Article
        25. El-Shahed M, Al-Askar F: Positive solutions for boundary value problem of nonlinear fractional q -difference equation. ISRN Math. Anal. 2011., 2011: Article ID 385459
        26. Liang S, Zhang J: Existence and uniqueness of positive solutions for three-point boundary value problem with fractional q -differences. J. Appl. Math. Comput. 2012, 40: 277-288. 10.1007/s12190-012-0551-2MathSciNetView Article
        27. Graef JR, Kong L: Positive solutions for a class of higher order boundary value problems with fractional q -derivatives. Appl. Math. Comput. 2012, 218: 9682-9689. 10.1016/j.amc.2012.03.006MathSciNetView Article
        28. Ma J, Yang J: Existence of solutions for multi-point boundary value problem of fractional q -difference equation. Electron. J. Qual. Theory Differ. Equ. 2011, 92: 1-10.MathSciNetView Article
        29. Ahmad B, Ntouyas S, Purnaras I: Existence results for nonlocal boundary value problems of nonlinear fractional q -difference equations. Adv. Differ. Equ. 2012., 2012: Article ID 140
        30. El-Shahed M, Gaber M: Two-dimensional q -differential transformation and its application. Appl. Math. Comput. 2011, 217: 9165-9172. 10.1016/j.amc.2011.03.152MathSciNetView Article
        31. Zhao Y, Chen H, Zhang Q: Existence results for fractional q -difference equations with nonlocal q -integral boundary conditions. Adv. Differ. Equ. 2013., 2013: Article ID 48
        32. Zhao Y, Ye G, Chen H: Multiple positive solutions of a singular semipositone integral boundary value problem for fractional q -derivatives equation. Abstr. Appl. Anal. 2013., 2013: Article ID 643571
        33. Guo DJ, Lakshmikantham V: Nonlinear Problems in Abstract Cones. Academic Press, San Diego; 1988.

        Copyright

        © Zhao et al.; licensee Springer. 2013

        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.