We propose an efficient numerical method for a class of fractional diffusion-wave equations with the Caputo fractional derivative of order α. This approach is based on the finite difference in time and the global sinc collocation in space. By utilizing the collocation technique and some properties of the sinc functions, the problem is reduced to the solution of a system of linear algebraic equations at each time step. Stability and convergence of the proposed method are rigorously analyzed. The numerical solution is of 3-α order accuracy in time and exponential rate of convergence in space. Numerical experiments demonstrate the validity of the obtained method and support the obtained theoretical results.
1. Introduction
Fractional diffusion-wave equations have attracted considerable attention as generalization of the classical diffusion/wave equation by replacing the integer-order time derivative with a fractional derivative of order α (1<α<2) [1]. It can be derived from the physical case of a general random time-dependent velocity function equipped with an algebraic time-correlation function (but with a Gaussian space correlation) [2]. Owing to the anomalous superdiffusion of particles, many electromagnetic, acoustic, mechanical, and biological responses can be modelled accurately by fractional diffusion-wave equations, for example, the propagation of stress waves in viscoelastic solids [3, 4]. There exist some analytical methods to find exact solutions of fractional diffusion-wave equations [5–9]. However, it is usually difficult or even impossible to achieve the exact solutions in many cases, and ones have to resort to numerical methods.
Compared with the considerable literature on numerical solutions of fractional diffusion equations [10–15], only a few works have been carried out in the numerical methods of fractional diffusion-wave equations. The homotopy analysis method [16] and the Adomian decomposition method [17] were used to construct analytical approximate solutions of fractional diffusion-wave equations, respectively. Sun and Wu [18] presented a full discrete difference scheme for the initial-boundary-value problem of a diffusion-wave equation by introducing two new variables and obtained the stability and convergence properties by using energy method. In [19], under the weak smoothness conditions, two finite difference schemes with first-order accuracy in temporal direction and second-order accuracy in spatial direction were proposed to solve the same problem. Murillo and Yuste [20] developed an explicit difference method where the L2 discretization formula is used for solving fractional diffusion-wave equations. Due to the nonlocal nature of fractional derivatives, all previous solutions have to be saved to compute the solution at the current time level. It makes the storage very expensive when low-order methods are employed for spatial discretization.
In this paper, we propose a numerical method mixing finite difference with sinc collocation to solve a class of fractional diffusion-wave equations. The sinc method is widely used to solve integral, integro-differential, ordinary differential, and partial differential equations [21–27]. As far as we know, there are very limited papers on solving fractional differential equation by the sinc method [28, 29]. The present method has great advantage in storage requirement because the sinc-collocation method needs fewer grid points to produce highly accurate solution when it is compared with a low-order method.
The remainder of this paper is organized as follows. In Section 2, we introduce some necessary definitions and relevant results for developing this method. Section 3 is devoted to the finite difference and sinc-collocation discretization for the fractional diffusion-wave equations. As a result, a system of linear algebraic equations is formed, and the solutions of the considered problems are obtained. Section 4 is concerned with the stability and convergence analysis. In Section 5, some numerical experiments are given to demonstrate the effectiveness of the proposed method and the obtained theoretical results. A brief conclusion ends this paper in the final section.
2. Notations and Some Preliminary Results
In this section, we introduce some basic definitions and relevant results of the fractional calculus [30, 31] and sinc functions.
Definition 1 (see [32]).
Let α∈R+. The operator Jaα defined on L1[a,b] by
(1)Jaαf(t)=1Γ(α)∫at(t-s)α-1f(s)ds
for a≤t≤b is called the Riemann-Liouville fractional integral operator of order α. For α=0, one sets Ja0:=I, that is, the identity operator.
Definition 2 (see [32]).
Let α∈R+ and n=⌈α⌉. The Caputo fractional differential operator DCaα for a≤t≤b is defined as
(2)DCaαf(t)=Jan-αDnf(t)=1Γ(n-α)∫at(t-s)n-α-1f(n)(s)ds.
The sinc functions and its properties are discussed thoroughly in [21, 23]. For any h>0, the translated sinc functions with equidistant space nodes are given as
(3)S(k,h)(z)=sinc(z-khh),k=0,±1,±2,…,
where the sinc functions are defined on the whole real line by
(4)sinc(x)={sin(πx)πx,x≠0,1,x=0.
If f is defined on R, then for any h>0 the series
(5)C(f,h)(z)=∑k=-∞∞f(kh)S(k,h)(z)
is called the Whittaker cardinal expansion of f whenever this series converges. f can be approximated by truncating (5). To construct our needed approximations on the interval [a,b], we choose
(6)ϕ(x)=ln(x-ab-x)
which maps a finite interval [a,b] onto R. ϕ is a conformal map which maps the eye-shaped domain in the complex z-plane
(7)DE={z∈C:|arg(z-ab-z)|<d≤π2}
onto the infinite strip DS of the complex z~-plane
(8)DS={z~∈C:|Im(z~)|<d≤π2}.
The basis functions on [a,b] are taken to be the composite translated sinc functions
(9)Sϕ(k,h)(x)=S(k,h)(ϕ(x))=sinc(ϕ(x)-khh).
Thus one may define the inverse image of the equidistant space node {ih} as
(10)xi=ϕ-1(ih)=a+beih1+eih,i=0,±1,±2,….
The class of functions such that the known exponential convergence rate exists for the sinc interpolation is denoted by B(DE) and defined in the following.
Definition 3 (see [21, 23]).
Let B(DE) be the class of functions f which are analytic in DE and satisfy
(11)∫ϕ-1(x+L)|f(z)dz|⟶0,x⟶±∞,
where L={iυ:|υ|<d≤(π/2)} and
(12)∫∂DE|f(z)dz|<∞
on the boundary of DE (denoted ∂DE).
The following theorem gives the error which results from differentiating the truncated cardinal series.
Theorem 4 (see [21, 22]).
Let ϕ′f/g∈B(DE) and let
(13)supx∈[a,b],-π/h≤t≤π/h|(ddx)n[g(x)eitϕ(x)]|≤C1h-n,
for n=0,1,…,m; here g is a weight function and C1 is a constant depending only on m, ϕ, and g. Assume that there exists a constant C2 such that
(14)f(x)g(x)≤C2eβϕ(x)[1+eϕ(x)]2β;
here β is a positive constant. Then taking h=πd/βN, one has
(15)|dnf(x)dxn-∑k=-NNf(xk)g(xk)dndxn[g(x)Sϕ(k,h)(x)]|≤C3N(m+1)/2exp(-πdβN)
for all n=0,1,…,m; here C3 is a constant depending only on m, ϕ, g, d, β, and f.
The above expressions show that the sinc interpolation on B(DE) converges exponentially. We also require the following derivatives of the composite translated sinc functions evaluated at the nodes:
(16)δk,i(0)=[Sϕ(k,h)(x)]|[Sϕ(k,h)(x)]x=xi={1,k=i,0,k≠i.(17)δk,i(1)=ddϕ[Sϕ(k,h)(x)]|[Sϕ(k,h)(x)]x=xi={0,k=i,(-1)i-k(i-k)h,k≠i.(18)δk,i(2)=d2dϕ2[Sϕ(k,h)(x)]|[Sϕ(k,h)(x)]x=xi={-π23h2,k=i,-2(-1)i-k(i-k)2h2,k≠i.
3. The Finite Difference and Sinc-Collocation Discretization
We consider the following fractional diffusion-wave equations with a nonhomogeneous field [33]:
(19)1c∂αu(x,t)∂tα=∂2u(x,t)∂x2+1Kf(x,t),a<x<b,t>0
with the initial conditions
(20)u(x,0)=φ(x),∂u(x,0)∂t=ψ(x),a<x<b
and the boundary conditions
(21)u(a,t)=0,u(b,t)=0,t>0,
where x∈[a,b] and t>0 are space and time variables, c and K are constants, and f(x,t) denotes the field variable. Here the time-fractional derivative is defined as the Caputo fractional derivative, and 1<α<2. Many authors refer to the fractional equation (19) as the fractional diffusion-wave equation when 1<α<2, which is expected to interpolate the diffusion equation and the wave equation [7, 18].
3.1. Temporal Discretization by a Finite Difference Scheme
First, we derive a finite difference scheme for temporal discretization of this equation. Let tk:=kτ, k=0,1,2,…, where τ is the time stepsize. In order to discretize the time-fractional derivative by using a finite difference approximation [18, 34], we introduce the following lemmas.
Lemma 5 (see [18]).
Suppose f(t)∈C2[0,tn]. Then
(22)|∫0tnf′(t)dt(tn-t)α-1-τ1-α2-α×[b0f(tn)-∑k=1n-1(bk-1-bk)f(tn-k)-bn-1f(t0)]|≤12-α[2-α12+23-α3-α-(1+21-α)]max0≤t≤tn|f′′(t)|τ3-α,
where 1<α<2 and
(23)bk=(k+1)2-α-k2-α,k=0,1,2,….
It is direct to check that
(24)1=b0>b1>⋯>bn>0,bn⟶0asn⟶∞.
Let
(25)ϕ(x,t)=∂αu(x,t)∂tα=1Γ(2-α)∫0t∂2u(x,s)∂s2ds(t-s)α-1,
and define the temporal grid functions
(26)Uk=u(x,tk),Φk=ϕ(x,tk),k=0,1,2,….
By the Taylor expansion, it follows that
(27)Utk-(1/2)=1τ(Uk-Uk-1)+r1kτ2,(28)1cΦk-(1/2)=(Uk-(1/2))xx+1Kfk-(1/2)+r2kτ2,
where r1k, r2k are constants. Based on Lemma 5, one has
(29)Φk=τ1-αΓ(3-α)[b0Utk-∑l=1k-1(bl-1-bl)Utk-l-bk-1Ut0]+O(τ3-α),k=1,2,3,….
Hence,
(30)Φk-(1/2)=12(Φk+Φk-1)=τ1-αΓ(3-α)×[b0Utk-(1/2)-∑l=1k-1(bl-1-bl)Utk-l-(1/2)-bk-1Ut0]+r3kτ3-α,
where r3k are constants. Let δtUk-(1/2)=(1/τ)(Uk-Uk-1). In view of (27), one has
(31)Φk-(1/2)=τ1-αΓ(3-α)×[b0δtUk-(1/2)-∑l=1k-1(bl-1-bl)δtUk-l-(1/2)-bk-1Ut0]+τ3-αΓ(3-α)[b0r1k-∑l=1k-1(bl-1-bl)r1k-l]+r3kτ3-α.
Then substituting (31) into (28) and observing Ut0=ψ(x), one obtains
(32)τ1-αcΓ(3-α)[b0δtUk-(1/2)-∑l=1k-1(bl-1-bl)δtUk-l-(1/2)-bk-1ψ]=(Uk-(1/2))xx+1Kfk-(1/2)+RTk,
where
(33)RTk=-τ3-αcΓ(3-α)[b0r1k-∑l=1k-1(bl-1-bl)r1k-l]-1cr3kτ3-α+r2kτ2.
It follows from (32) that one can construct easily the following semidiscrete finite difference scheme for (19):
(34)τ1-αcΓ(3-α)[b0δtuk-(1/2)-∑l=1k-1(bl-1-bl)δtuk-l-(1/2)-bk-1ψ]=(uk-(1/2))xx+1Kfk-(1/2),
which is equivalent to
(35)1cταΓ(3-α)(Δ1+Δ2)=12[(uk)xx+(uk-1)xx]+1Kfk-(1/2),k=2,3,…,
where
(36)Δ1=b0uk-(2b0-b1)uk-1,Δ2=∑l=2k-1(bl-2-2bl-1+bl)uk-l+(bk-2-bk-1)u0-bk-1τψ.
According to (33), this scheme is of (3-α)-order accuracy. A rigorous analysis of the convergence rate will be provided later. The above scheme can be rewritten into
(37)b0cταΓ(3-α)uk-12(uk)xx=2b0-b1cταΓ(3-α)uk-1-1cταΓ(3-α)Δ2+12(uk-1)xx+1Kfk-(1/2).
Specially, for the case k=1, the scheme simply reads
(38)b0cταΓ(3-α)u1-12(u1)xx=b0cταΓ(3-α)(u0+τψ)+12(u0)xx+1Kf(1/2).
So (37) and (38) together with the initial condition
(39)u0=u(x,0)=φ(x),a<x<b
and the boundary conditions
(40)uak=u(a,tk)=0,ubk=u(b,tk)=0,k≥1
form a complete semidiscrete problem.
3.2. Space Discretization by the Sinc-Collocation Method
Next we consider space discretization to (37) by the sinc-collocation method. We select the collocation points xi by (10). The space discretization proceeds by approximating the solution based on the composite translated sinc functions (9)
(41)uk≃uNk=∑j=-nnujkSϕ(j,h)(x).
The unknown parameters ujk will be determined by the collocation method. It is noted that the approximation in (41) satisfies the boundary conditions in (40) since Sϕ(j,h)(x), j=-n,-n+1,…,n are zero when x tends to a and b.
Now substituting (41) into (37) and collocating in 2n+1 points xi, we obtain
(42)b0cταΓ(3-α)∑j=-nnujkSϕ(j,h)(xi)-12∑j=-nnujkSϕ′′(j,h)(xi)=2b0-b1cταΓ(3-α)∑j=-nnujk-1Sϕ(j,h)(xi)+12∑j=-nnujk-1Sϕ′′(j,h)(xi)-1ταΓ(3-α)×[∑l=2k-1(bl-2-2bl-1+bl)kkkkkkiik×∑j=-nnujk-lSϕ(j,h)(xi)+(bk-2-bk-1)u0(xi)kikkkkkk∑l=2k-1-τbk-1ψ(xi)]+1Kfk-(1/2)(xi),
where i=-n,-n+1,…,n. Based on (17) and (18), we let
(43)qji:=Sϕ′′(j,h)(xi)=ϕ′′(xi)δj,i(1)+(ϕ′(xi))2δj,i(2).
Hence, (42) is reduced immediately by (16) and (43) to
(44)b0cταΓ(3-α)uik-12∑j=-nnqjiujk=2b0-b1cταΓ(3-α)uik-1-1cταΓ(3-α)×[∑l=2k-1(bl-2-2bl-1+bl)uik-l∑l=2k-1kkk+(bk-2-bk-1)u0(xi)-τbk-1ψ(xi)]+12∑j=-nnqjiujk-1+1Kfk-(1/2)(xi),
where i=-n,-n+1,…,n. To obtain a matrix representation of the above equation, we let
(45)A=b0cταΓ(3-α)I(2n+1)×(2n+1),B=12(qji)(2n+1)×(2n+1),C(1)=2b0-b1cταΓ(3-α)I(2n+1)×(2n+1),C(l)=-bl-2-2bl-1+blcταΓ(3-α)I(2n+1)×(2n+1),C(k)=-bk-2-bk-1cταΓ(3-α)I(2n+1)×(2n+1),D=bk-1cτα-1Γ(3-α)I(2n+1)×(2n+1),Uk=[u-nk,u-n+1k,…,unk]T,Ψ=[ψ(x-n),ψ(x-n+1),…,ψ(xn)]T,Fk-(1/2)=1K[fk-(1/2)(x-n),fk-(1/2)(x-n+1),…,fk-(1/2)(xn)]T,
where the matrix I is the identity matrix and l=2,3,…,k-1. Therefore, at each time step, we get the following system of 2n+1 linear equations with 2n+1 unknown parameters ujk, and this system can be expressed in a matrix form
(46)QUk=P,
where
(47)Q=A-B,P=BUk-1+∑l=1kC(l)Uk-l+DΨ+Fk-(1/2).
For the initial condition (39), we have
(48)U0=[φ(x-n),φ(x-n+1),…,φ(xn)]T.
Consequently (46) can be solved easily for the unknown coefficients Uk. Hence the approximation solution given in (41) can be obtained.
4. Stability and Convergence Analysis of the Derived Method
To analyse the stability and convergence of the derived method, the inner product is defined by 〈u,υ〉=∫abuυdx with the corresponding norm ∥u∥=(u,u)(1/2), which will be used thereafter. We first give the following lemmas.
Lemma 6 (see [18]).
For any Q=Q1,Q2,Q3,… and υ, here Qi,υ∈L2(a,b), it can be verified that
(49)∑i=1k〈b0Qi-∑l=1i-1(bl-1-bl)Qi-l-bi-1υ,Qi〉≥12(2-α)k1-α∑i=1k∥Qi∥2-12k2-α∥υ∥2,kkkkkkkkkkkkkkkkkkkkk=1,2,3,…,
where bl is defined in (23).
Lemma 7.
Suppose that uNk, k=1,2,3,…, satisfy
(50)τ1-αcΓ(3-α)[b0δtuNk-(1/2)-∑l=1k-1(bl-1-bl)δtuNk-l-(1/2)-bk-1υ]=(uNk-(1/2))xx+fk-(1/2),(51)uN0=φ,0<x<1,(52)uNk(a)=0,uNk(b)=0.
Then one has the estimate
(53)∥(uNk)x∥2≤∥(uN0)x∥2+tk2-αcΓ(3-α)∥υ∥2+cΓ(2-α)tkα-1τ∑l=1k∥fk-(1/2)∥2.
Proof.
Based on (50), we have
(54)τ1-αcΓ(3-α)∑k=1m〈b0δtuNk-(1/2)∑l=1k-1-∑l=1k-1(bl-1-bl)δtuNk-l-(1/2)ikkkkkkkkkkkk∑l=1k-1-bk-1υ,τδtuNk-(1/2)〉=∑k=1m((uNk-(1/2))xx,τδtuNk-(1/2))+∑k=1m(fk-(1/2),τδtuNk-(1/2)).
Applying Lemma 6 to the term in left side of (54), we obtain
(55)τ1-αcΓ(3-α)∑k=1m〈b0δtuNk-(1/2)-∑l=1k-1(bl-1-bl)δtuNk-l-(1/2)kkkkkkkkkkkki∑l=1k-1-bk-1υ,τδtuNk-(1/2)〉≥12cΓ(2-α)tm1-ατ∑k=1m∥δtuNk-(1/2)∥2-12cΓ(3-α)tm2-α∥υ∥2.
On the other side, based on the fact that δtuNk-(1/2)(a)=δtuNk-(1/2)(b)=0, a straightforward calculation of the right terms in (54) gives
(56)∑k=1m〈(uNk-(1/2))xx,τδtuNk-(1/2)〉=-τ∑k=1m〈(uNk-(1/2))x,δt(uNk-(1/2))x〉=-12∑k=1m〈(uNk)x+(uNk-1)x,(uNk)x-(uNk-1)x〉=-12[∥(uNm)x∥2-∥(uN0)x∥2].∑k=1m〈fk-(1/2),τδtuNk-(1/2)〉≤12τ∑k=1m[cΓ(2-α)tmα-1∥fk-(1/2)∥2∥δtuNk-(1/2)∥21cΓ(2-α)tm1-α∥δtuNk-(1/2)∥2lllllllllllllllllii+1cΓ(2-α)tm1-α∥δtuNk-(1/2)∥2].
Substituting (55) and (56) into (54) yields inequality (53).
Suppose that uk, k=0,1,2,…, satisfy the conditions of Theorem 4. Substituting the approximation solution uNk given in (41) into (34) leads to
(57)τ1-αcΓ(3-α)[b0δtuNk-(1/2)-∑l=1k-1(bl-1-bl)δtuNk-l-(1/2)-bk-1ψ]=(uNk-(1/2))xx+1Kfk-(1/2)+RNk,k=1,2,3,….
Subtracting (57) from (34), we obtain
(58)τ1-αcΓ(3-α)×(b01τ[(uk-uNk)-(uk-1-uNk-1)]∑l=1k-1ii-1τ∑l=1k-1(bl-1-bl)[(uk-l-uNk-l)-(uk-l-1-uNk-l-1)])=(uk-(1/2))xx-(uNk-(1/2))xx-RNk,k=1,2,3,….
Based on Theorem 4, we have
(59)RNk=-1cΓ(3-α)ταne-πdβn×[b0(dk-dk-1)-∑l=1k-1(bl-1-bl)(dk-l-dk-l-1)]+dk+1n(3/2)e-πdβn,
where d0,d1,…,dk+1 are constants.
Theorem 8.
Let uNk, k=0,1,2,…, be the approximation solution given in (41) and let uk satisfy the conditions of Theorem 4. Then one has
(60)∥uNk∥2≤C[∥(uN0)x∥2+tk2-αcΓ(3-α)∥ψ∥2∑l=1k∥1Kfl-(1/2)+RNl-(1/2)∥2kkkkk+cΓ(2-α)tkα-1τlllllllli×∑l=1k∥1Kfl-(1/2)+RNl-(1/2)∥2],
where C is constant.
Proof.
Considering (57) and noting that uN0=φ and uNk(a)=uNk(b)=0, k=0,1,2,…, we can easily obtain this result from Lemma 7 and the Poincaré inequality.
Theorem 9.
Let the problem (19)–(21) have the exact solution Uk and let uNk be the approximation solution given in (41), where k=0,1,2,…. If uk satisfy the conditions of Theorem 4, then for kτ≤T, one has
(61)∥ek∥≤Γ(2-α)Tα(b-a)(c1τ3-α+c2τ-αn(3/2)e-πdβn),
where ek=Uk-uNk.
Proof.
Subtracting (57) from (32), we obtain the error equation
(62)τ1-αcΓ(3-α)[b0δtek-(1/2)-∑l=1k-1(bl-1-bl)δtek-l-(1/2)]=(ek-(1/2))xx+RTk-RNk,
where k=1,2,3,…. Noting that e0=0 and ek(a)=ek(b)=0, k=1,2,…, and applying Lemma 7, we have
(63)∥(ek)x∥2≤cΓ(2-α)tkα-1τ∑l=1k∥RTl-RNl∥2.
Based on (33) and (59), we obtain
(64)∥(ek)x∥≤c3Γ(2-α)Tα(b-a)×[τ3-α+c4τ-αn(3/2)e-πdβn].
Applying the Poincaré inequality yields the needed result.
Remark 10.
In Theorem 9, the error estimate formula contains c2τ-αn(3/2)e-πdβn (the second term in the right-hand side), where the error contribution from the spatial approximation is affected by the inverse of the time step. It is worthwhile noting that similar results are also found for the classical diffusion/wave equation. However for large n, n(3/2)e-πdβn can be much smaller than τ; therefore, this affection generally would not reduce the global accuracy.
5. Numerical Examples
To validate the effectiveness of the proposed method for the problem (19)–(21), we consider the example given in [18]. Consider the following:
(65)∂αu(x,t)∂tα=∂2u(x,t)∂x2+sin(πx),0<x<1,0<t≤1,u(x,0)=0,∂u(x,0)∂t=0,0<x<1,u(0,t)=0,u(1,t)=0,0<t≤1.
The exact solution of the above problem is [33]
(66)u(x,t)=1π2[1-Eα(-π2tα)]sin(πx),
where Eα(z)=∑k=0∞zk/Γ(αk+1) which is one-parameter Mittag-Leffler function.
For solving the above problem (with α=1.3 and 1.7, resp.) by using the method described in Section 3, we choose β=1 and d=π/2, and this leads to h=π/2n. We will report the efficiency and accuracy of the given method based on the L2-errors and L∞-errors. Figure 1 gives the 3D diagrams of the numerical solutions uN(x,t) on the whole computational domain [0,1]×[0,1] with τ=0.001, n=20. In Figure 2, we plot the curves of the numerical solutions uN(x,t) and exact solutions u(x,t) for several fixed time instants. A good agreement of the numerical solution with the exact one is achieved. Furthermore, Figure 3 shows the absolute error |uN(x,t)-u(x,t)| obtained by the present method with τ=0.001, n=20.
The numerical solutions with α=1.3 (a) and α=1.7 (b).
Comparison of the numerical and exact solutions for several fixed time instants with α=1.3 (a) and α=1.7 (b).
Plot of the absolute errors with α=1.3 (a) and α=1.7 (b).
To check the convergence behavior of the numerical solution with respect to the time step τ and the parameter n about the number of collocation points, we represent the errors ∥Uk-uNk∥ in two discrete norms: L∞ and L2. All the numerical results reported in Figures 4 and 5, and Tables 1 and 2 are evaluated at t=0.8.
Temporal errors and convergence orders with α=1.3.
τ
L∞-error
Order
L2-error
Order
1/10
7.589394E-04
5.363813E-04
1/20
2.352470E-04
1.6898084
1.662599E-04
1.6898189
1/40
7.279201E-05
1.6923243
5.144424E-05
1.6923586
1/80
2.252242E-05
1.6924183
1.591604E-05
1.6925284
1/160
6.961152E-06
1.6939638
4.918054E-06
1.6943215
1/320
2.144986E-06
1.6983580
1.514220E-06
1.6995127
1/640
6.553662E-07
1.7105947
4.614370E-07
1.7143693
Temporal errors and convergence orders with α=1.7.
τ
L∞-error
Order
L2-error
Order
1/10
5.571817E-03
3.937904E-03
1/20
2.625563E-03
1.0855211
1.855628E-03
1.0855201
1/40
1.148662E-03
1.1926720
8.118237E-04
1.1926694
1/80
4.844558E-04
1.2455178
3.423934E-04
1.2455116
1/160
2.006883E-04
1.2714088
1.418397E-04
1.2713934
1/320
8.239443E-05
1.2843377
5.823513E-05
1.2843000
1/640
3.367394E-05
1.2909143
2.380175E-05
1.2908220
Temporal errors in the L∞ and L2 norms versus τ for several α.
Spatial errors in the L∞ and L2 norms versus n for several α.
Firstly, the computational investigation is concerned with the temporal convergence rate. We choose the parameter n=60 and it is a value large enough such that the errors coming from the spatial approximation are negligible [35]. In Tables 1 (α=1.3) and 2 (α=1.7), we list the temporal errors when τ decreases from 1/10 to 1/320 and the convergence order which is very close to the expectation order 3-α. We also plot the errors in the L∞ and L2 norms as a function of the time stepsize τ for several α in Figure 4, where a logarithmic scale has been used for both τ-axis and error axis. From Figure 4, it is clear that, for α=1.3 and 1.7, the slopes of the error curves in these log-log plots approach 1.7 and 1.3, respectively. So the proposed method yields a temporal approximation order which is close to 3-α as forecasted by the theoretical estimate.
Now we check the spatial accuracy with respect to the parameter n about the number of collocation points. For the similar reason mentioned above, we fix the time step τ sufficiently small to avoid contamination of the temporal error. In Figure 5, we present the spatial errors as a function of n with τ=0.0001, where a logarithmic scale is now used for the spatial errors axis. It is clearly observed that the spatial errors appear in an exponential decay. In this semilog representations, we observe that the error variations are essentially linear versus n [36].
6. Conclusion
In this paper, we have developed and analyzed an efficient numerical method for a class of the initial-boundary-value problems of fractional diffusion-wave equations. Based on a finite difference scheme in time and a global sinc-collocation method in space, the problem is reduced to the solution of a system of linear algebraic equations at each time step. The proposed method leads to 3-α order accuracy in time and exponential convergence in space. The theoretical results are perfectly confirmed by the numerical experiments.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
The present research is supported by the National Natural Science Foundation of China (Grant nos. 11271311 and 11371016), the Chinese Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT) (Grant no. IRT1179), and the Hunan Province Innovation Foundation for Postgraduate (Grant no. CX2013B252). The authors are grateful to the anonymous reviewers for their careful comments and valuable suggestions on this paper.
MainardiF.WegnerJ. L.NorwoodF. R.Fractional diffusive waves in viscoelastic solids1995Fairfield, NJ, USAASME/AMR9397BalescuR.V-Langevin equations, continuous time random walks and fractional diffusion2007341628010.1016/j.chaos.2007.01.0502-s2.0-34047203146MainardiF.Fractional relaxation-oscillation and fractional diffusion-wave phenomena1996791461147710.1016/0960-0779(95)00125-52-s2.0-0030464353KilbasA. A.SrivastavaH. M.TrujilloJ. J.2006Amsterdam, The NetherlandElsevierMainardiF.The fundamental solutions for the fractional diffusion-wave equation199696232810.1016/0893-9659(96)00089-4MR14198112-s2.0-30244460855MomaniS.OdibatZ.ErturkV. S.Generalized differential transform method for solving a space- and time-fractional diffusion-wave equation20073705-637938710.1016/j.physleta.2007.05.083MR24190632-s2.0-35349007529AgrawalO. P.Solution for a fractional diffusion-wave equation defined in a bounded domain2002291–414515510.1023/A:1016539022492MR19264712-s2.0-0036650559LuchkoY.Fractional wave equation and damped waves20135431603150510.1063/1.4794076MR30594272-s2.0-84875877772PovstenkoY.Axisymmetric solutions to fractional diffusion-wave equation in a cylinder under Robin boundary condition201322281767177710.1140/epjst/e2013-01962-4PodlubnyI.ChechkinA.SkovranekT.ChenY.Vinagre JaraB. M.Matrix approach to discrete fractional calculus II: partial fractional differential equations200922883137315310.1016/j.jcp.2009.01.014MR25093112-s2.0-61349186917YusteS. B.Weighted average finite difference methods for fractional diffusion equations2006216126427410.1016/j.jcp.2005.12.006MR2223444ZBLl1094.650852-s2.0-33646128485ZhuangP.LiuF.AnhV.TurnerI.New solution and analytical techniques of the implicit numerical method for the anomalous subdiffusion equation20084621079109510.1137/060673114MR2383224ZBLl1173.260062-s2.0-55549107511KhaderM. M.On the numerical solutions for the fractional diffusion equation20111662535254210.1016/j.cnsns.2010.09.007MR2765206ZBLl1221.652632-s2.0-78651444144PiretC.HanertE.A radial basis functions method for fractional diffusion equations2013238718110.1016/j.jcp.2012.10.041MR3028344ZBLl1286.651352-s2.0-84873162810MohebbiA.AbbaszadehM.DehghanM.A high-order and unconditionally stable scheme for the modified anomalous fractional sub-diffusion equation with a nonlinear source term2013240364810.1016/j.jcp.2012.11.052MR3039243ZBLl1287.650642-s2.0-84874387562JafariH.GolbabaiA.SeifiS.SayevandK.Homotopy analysis method for solving multi-term linear and nonlinear diffusion-wave equations of fractional order20105931337134410.1016/j.camwa.2009.06.020ZBLl1189.652502-s2.0-74249121146OdibatZ. M.MomaniS.Approximate solutions for boundary value problems of time-fractional wave equation2006181176777410.1016/j.amc.2006.02.004MR2270533ZBLl1148.651002-s2.0-33749512364SunZ.WuX.A fully discrete difference scheme for a diffusion-wave system200656219320910.1016/j.apnum.2005.03.003MR22009382-s2.0-30744474991HuangJ.TangY.VázquezL.YangJ.Two finite difference schemes for time fractional diffusion-wave equation201364470772010.1007/s11075-012-9689-0MR3132960MurilloJ. Q.YusteS. B.An explicit difference method for solving fractional diffusion and diffusion-wave equations in the Caputo form201162602101410.1115/1.40026872-s2.0-78649924907StengerF.1993New York, NY, USASpringerStengerF.2011New York, NY, USACRC PressLundJ.BowersK. L.1992Philadelphia, Pa, USASIAMParandK.DehghanM.PirkhedriA.The sinc-collocation method for solving the Thomas-Fermi equation2013237124425210.1016/j.cam.2012.08.001MR2966903ZBLl1253.651112-s2.0-84865437859Al-KhaledK.Numerical study of Fisher's reaction-diffusion equation by the Sinc collocation method2001137224525510.1016/S0377-0427(01)00356-9MR18652302-s2.0-0035892569SecerA.Numerical solution and simulation of second-order parabolic PDEs with Sinc-Galerkin method using Maple2013201310686483MR309029110.1155/2013/686483DehghanM.Emami-NaeiniF.The Sinc-collocation and Sinc-Galerkin methods for solving the two-dimensional Schrödinger equation with nonhomogeneous boundary conditions201337229379939710.1016/j.apm.2013.04.043MR31154052-s2.0-84878800254SaadatmandiA.DehghanM.AziziM.The Sinc-Legendre collocation method for a class of fractional convection-diffusion equations with variable coefficients201217114125413610.1016/j.cnsns.2012.03.003MR29303122-s2.0-84861716404BaumannG.StengerF.Fractional calculus and Sinc methods201114456862210.2478/s13540-011-0035-3MR28463772-s2.0-84856328476PodlubnyI.1999New York, NY, USAAcademic PressMR1658022BaleanuD.DiethelmK.SE.TrujilloJ. J.2012SingaporeWorld ScientificWeilbeerM.2006Clausthal-Zellerfeld, GermanyPapierfliegerAgrawalO. P.Response of a diffusion-wave system subjected to deterministic and stochastic fields200383426527410.1002/zamm.200310033MR19770022-s2.0-21144451936DiethelmK.2010Berlin, GermanySpringerMR2680847LinY.XuC.Finite difference/spectral approximations for the time-fractional diffusion equation200722521533155210.1016/j.jcp.2007.02.001MR2349193ZBLl1126.651212-s2.0-34547548712LiX.XuC.A space-time spectral method for the time fractional diffusion equation20094732108213110.1137/080718942MR2519596ZBLl1193.35243