Abstract
In this paper, we compare solutions of q-order fractional differential equations of Caputo type for q near 1 with solutions of the corresponding 1-order ordinary differential equations. By establishing the explicit lower and upper bounds of Mittag-Leffler functions, we obtain the effective convergence results. It is shown that the limit cases \(q\to1_{+}\) and \(q\to1_{-}\) are different. A simple illustrative example is also presented.
Similar content being viewed by others
1 Introduction
Fractional differential equations (FDEs) are a rapidly developing area of mathematics with many stimulating applications [1,2,3,4]. Recently, plenty interesting existence and controllability results on the theory of solutions of FDEs or fractional inclusions have been given in [5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22]. Mathematical modeling approaches using fractional derivatives are presented in [17,18,19,20,21,22] with numerical simulations on various challenging topics.
On the one hand, several properties of ordinary or partial differential equations (DEs) appear in FDEs as well, like asymptotic properties of solutions or equilibria. On the other hand, unlike to DEs, FDEs have no nonconstant periodic solutions and they do not create dynamical systems, which is one of the most obvious characteristics in studying FDEs. So there is a natural question to study the relationship between solutions of FDEs and DEs when the order q of FDEs is near to a natural number \(n\in \mathbb {N}\). Here, we call such FDEs weakly fractional, which can be used to seek numerically the solutions of DEs.
In this paper, we investigate for simplicity the case when q is near to \(n=1\), but our method can be directly extended to any n. We study two cases: \(q\to1_{-}\) in Sect. 2 and \(q\to1_{+}\) in Sect. 3. We derive error estimates in both cases. A simple numerical illustrative example is given to demonstrate theoretical results. Our next step will be to extend this paper for weakly fractional semilinear evolution equations in Banach spaces.
2 The case \(q\to1_{-}\)
Consider a fractional differential equation
where \(D_{0}^{q}\) is the Caputo fractional derivative of order \(q\in (0,1)\) with the lower limit at zero,
and \(f\in C(\mathbb {R}_{+}\times \mathbb {R}^{n},\mathbb {R}^{n})\) along with an ordinary differential equation
where \(x_{0},y_{0}\in \mathbb {R}^{n}\). We suppose
-
(H)
There are nonnegative constants M and L such that \(\Vert f(t,x)\Vert \le M\) and \(\Vert f(t,x)-f(t,y)\Vert \le L\Vert x-y\Vert \) for any \(t\in \mathbb {R}_{+}\) and \(x,y\in \mathbb {R}^{n}\), where \(\Vert \cdot\Vert \) is a norm on \(\mathbb {R}^{n}\).
It is well known [4] that problem (1) is equivalent to the following integral equation:
Then we derive
Thus by the Henry–Gronwall inequality (see [23, Corollary 2]), we get
for any \(t\in \mathbb {R}_{+}\), where \(E_{q}\) is the Mittag-Leffler function [24]. We continue with the case \(x_{0}=y_{0}\). Then we get
for any \(t\in \mathbb {R}_{+}\). The equation
has the only solution \(s_{0}>0\) given by
Note that the function \(s_{0}(q)\) is increasing on \((0,1)\) with \(\lim_{q\to0_{+}}s_{0}(q)=0\) and
for the Euler constant γ. Next, clearly, we have
Consequently, we obtain
We can check numerically that \(\lambda''(q)>0\) for \(q\in(0,1)\), then that \(\lambda'(q)\) is increasing from −∞ to \(-e^{-\gamma }\doteq-0.561459\), and then that \(\lambda(q)\) is decreasing from 1 to 0. So we consider \(q\in(1/2,1)\) and then \(-0.751988\le\lambda '(q)\le-0.561459\). This implies that
for \(q\in(1/2,1)\). Next, by [25, Lemma 2], we have the following.
Lemma 2.1
For all \(t\in \mathbb {R}_{+}\), \(q\in(0,1)\), and \(\kappa>0\), it holds
Furthermore, (5) implies
for \(t\in \mathbb {R}_{+}\). So if \(q\in(1/2,1)\), then by Lemma 2.1 we get
for \(\tilde{L}=\max\{L,L^{2}\}\).
Now we are ready to deal with (3). First, (3) immediately implies the following expected result.
Theorem 2.2
Under assumption (H), the solution \(x(t)\) of (1) uniformly converges on any finite interval \([0,T]\), \(T>0\), of \(\mathbb {R}_{+}\) to the solution \(y(t)\) of (2) if \(q\to1_{-}\) and \(x_{0}=y_{0}\).
Proof
The proof follows directly from (3), (5) and by
uniformly for \(t\in[0,T]\) and any fixed \(T>0\). □
Next, we take any \(\epsilon>0\) and consider an equation
Clearly, \(\theta_{q}(t)\) is increasing on \(\mathbb {R}_{+}\) from 0 to ∞. Thus (8) has the only solution \(\bar{t}(\epsilon,q)\in \mathbb {R}_{+}\). By the above observations we can easily see that \(\lim_{\epsilon\to 0_{+}}\bar{t}(\epsilon,q)=0\) and \(\lim_{q\to1_{-}}\bar{t}(\epsilon ,q)=\infty\).
Furthermore, the function \(t\mapsto t-\frac{t^{q}}{\varGamma(q+1)}\) is nonpositive on \([0,r_{0}]\) and nonnegative on \([r_{0},\infty)\) for
Note that the function \(r_{0}(q)\) is increasing on \((0,1)\) from \(\lim_{q\to0_{+}}r_{0}(q)=1\) to \(\lim_{q\to1_{-}}r_{0}(q)=e^{1-\gamma}\doteq1.526205\).
Next, we study the function \(\phi_{t}(q):=\frac{t^{q}}{\varGamma(q+1)}\) on \((0,1)\) for \(t>0\). We have
For \(t\in(0,1]\) and \(q\in(1/2,1)\), we get
while for \(1\le t\le T\) and \(q\in(1/2,1)\), we get
for \(T>1\). Consequently, we have
for \(t\in(0,T]\), \(T>1\), and \(q\in(1/2,1)\). This implies
for \(t\in[0,T]\), \(T>1\), and \(q\in(1/2,1)\). Using (6), (7), and (11), we arrive at
for \(t\in[0,T]\), \(T>1\), and \(q\in(1/2,1)\). Now, we consider instead of (8) the following one:
The function \(\eta_{L,q}(T)\) is increasing from \(8e^{\tilde{L}}\) to ∞ on \([1,\infty)\). So, for any
(12) has a unique solution \(T_{L}(q)>1\). Note
Summarizing, we have the following result.
Theorem 2.3
Under assumption (H) and for any q fulfilling (13), the solutions \(x(t)\) and \(y(t)\) of (1) and (2) with \(x_{0}=y_{0}\), respectively, satisfy
for any \(t\in[0,T_{L}(q)]\), where \(T_{L}(q)>1\) is the unique solution of (12).
3 The case \(q\to1_{+}\)
Consider a fractional differential equation
where \(q\in(1,2)\) and \(f\in C(\mathbb {R}_{+}\times \mathbb {R}^{n},\mathbb {R}^{n})\) along with an ordinary differential equation
where \(x_{0},x_{1},y_{0},y_{1}\in \mathbb {R}^{n}\). Again, we suppose assumption (H). It is known [2, Theorem 3.24] that initial value problem (15) is equivalent to the integral equation
Analogously to the previous section, we derive
and the Henry–Gronwall inequality yields
Hence, for \(x_{0}=y_{0}\), \(x_{1}=y_{1}\), estimation (3) follows for any \(t\in \mathbb {R}_{+}\). Function \(s_{0}(q)\) of (4) is increasing on \((1,2)\) from
to 1. So this time,
Consequently, we have (compare with (5))
where \(\lambda(q)\) is given by (5). One can check numerically that \(-\lambda''(q)<0\) for \(q\in(1,2)\). So, \(-\lambda'(q)\) is decreasing from \(e^{-\gamma}\doteq0.561459\) to \(\frac{3}{4}-\frac {\gamma}{2}\doteq0.461392\), and \(-\lambda(q)\) is increasing from 0 to \(1/2\). Hence for \(q\in(1,2)\), we can estimate
Next, we need the following analog to Lemma 2.1.
Lemma 3.1
For all \(t\in \mathbb {R}_{+}\), \(q\in(1,4/3)\), and \(\kappa>0\), it holds
Proof
Using Dzherbashyan’s recursion formula [26],
for \(\alpha,\beta>0\), \(z\in \mathbb {R}\), \(m\in \mathbb {N}\), where \(\imath=\sqrt {-1}\), we can write
for any \(z>0\). Next, from [27, Theorem 2.1] we know
for any \(\alpha>0\), \(z<0\). So, using \(\cos\frac{\pi q}{2}\geq-1/2\) for \(q\in(1,4/3)\), we get
Finally, applying this estimation and Lemma 2.1 to (19) results in
□
Since by (17),
for all \(t\in \mathbb {R}_{+}\), Lemma 3.1 implies
for \(q\in(1,4/3)\), where \(\overline{L}=\max\{L,L^{3/4}\}\). So we obtain a result on the uniform convergence.
Theorem 3.2
Under assumption (H), the solution \(x(t)\) of (15) uniformly converges on any finite interval \([0,T]\), \(T>0\), of \(\mathbb {R}_{+}\) to the solution \(y(t)\) of (16) if \(q\to1_{+}\) and \(x_{0}=y_{0}\), \(x_{1}=y_{1}\).
Proof
The statement can be proved analogously to Theorem 2.2. □
Next, we consider equation (8) for an arbitrary \(\epsilon>0\) and \(q\in(1,4/3)\). Clearly, \(\theta_{q}(t)\) is increasing on \(\mathbb {R}_{+}\) from 0 to ∞, implying that (8) has the only solution \(\bar{t}(\epsilon,q)\in \mathbb {R}_{+}\) for which \(\lim_{\epsilon\to0_{+}}\bar{t}(\epsilon,q)=0\) and \(\lim_{q\to 1_{+}}\bar{t}(\epsilon,q)=\infty\) hold. Moreover, the function \(t\mapsto t-\frac{t^{q}}{\varGamma(q+1)}\) is nonnegative on \([0,r_{0}]\) and nonpositive on \([r_{0},\infty)\) for \(r_{0}\) given by (9). Note that \(r_{0}(q)\) is increasing on \((1,\infty )\) from \(\lim_{q\to1_{+}}r_{0}(q)=e^{1-\gamma}\doteq1.526205\) to ∞.
Next, we consider the function \(\phi_{t}(q):=\frac{t^{q}}{\varGamma(q+1)}\) on \((1,4/3)\) for \(t>0\). From (10), we obtain
for \(t\in(0,1]\), and
for \(t\in(1,T]\), \(T>1\). As a consequence, we have
for all \(t\in(0,T]\), \(T>1\), \(q\in(1,4/3)\). This implies
for \(t\in(0,T]\), \(T>1\), \(q\in(1,4/3)\). Using (18), (20), and (21), we arrive at
for \(t\in(0,T]\), \(T>1\), \(q\in(1,4/3)\). Now, we consider the equation
The function \(\mu_{L,q}(T)\) is increasing from \(4(e^{\overline {L}}+4\sqrt{3}/9)\) to ∞ on \([1,\infty)\). So, for any
(22) has a unique solution \(T_{L}(q)>1\). Note that \(\lim_{q\to 1_{+}}T_{L}(q)=\infty\). Summarizing, we have the following result.
Theorem 3.3
Under assumption (H) and for any q fulfilling (23), the solutions \(x(t)\) and \(y(t)\) of (15) and (16) with \(x_{0}=y_{0}\), \(x_{1}=y_{1}\), respectively, satisfy
for any \(t\in[0,T_{L}(q)]\), where \(T_{L}(q)>1\) is the unique solution of (22).
Next, we present a simple example illustrating the convergence results when the order q is close to 1.
Example 3.4
Let us consider the following initial-value problems:
where \(q\in(0,1)\) in (25) and \(q\in(1,2)\) in (27). The ODEs have the solutions \(y(t)=y_{0}e^{pt}\), \(v(t)=e^{pt}(v_{0}+v_{1}/p)-v_{1}/p\). From [2, Theorem 4.3], the other solutions are \(x(t)=x_{0}E_{q}(pt^{q})\) and \(u(t)=u_{0}E_{q}(pt^{q})+u_{1}tE_{q,2}(pt^{q})\). To see the convergence, we set all the initial conditions and the parameter equal to 1, i.e., \(x_{0}=y_{0}=u_{0}=v_{0}=u_{1}=v_{1}=p=1\). Figure 1 depicts the convergences \(x\to y\) and \(u\to v\) as \(q\to1_{-}\) and \(q\to1_{+}\), respectively.
The physical significance of Fig. 1 relies on demonstration of transition of q through 1. Since (25) is a one-dimensional system depending just on \(x_{0}\), its limit (26) is also one-dimensional. But passing to (27), we get a two-dimensional system depending on \(u_{0}\) and \(u_{1}\). Then its limit (28) as \(q\to1_{+}\) is also two-dimensional. This makes the difference. Note that (28) is equivalent to a second order ODE
The above arguments are more visible for \(p<0\). Then by [27, Formula (7)] we see that solutions of (25), (26), and (27) asymptotically tend to zero, while the one of (28) tends to \(-\frac{v_{1}}{p}\). So all these equations are dissipative. But the limit of (27) as \(q\to2_{-}\) is
which has all solutions oscillating for \(p<0\). Consequently, the dissipation of (25)–(28) is changing to oscillation on finite intervals as \(q\to2_{-}\). This is presented in Figs. 2 and 3.
These figures also support the fact that comparison estimates can be done in general only on finite intervals.
4 Conclusion
Solutions of q-order fractional differential equations of Caputo type for q near 1 are compared to solutions of the corresponding 1-order ordinary differential equations, by establishing the effective convergence results. As a result we get that the limit cases \(q\to1_{+}\) and \(q\to1_{-}\) are different. Theoretical results are demonstrated on a simple illustrative example. Our method can be directly extended to any order q near a natural number.
References
Fečkan, M., Wang, J.R., Pospíšil, M.: Fractional-Order Equations and Inclusions. de Gruyter, Berlin (2017)
Kilbas, A.A., Srivastava, H.M., Trujillo, J.J.: Theory and Applications of Fractional Differential Equations. North-Holland Mathematics Studies, vol. 204. Elsevier, Amsterdam (2006)
Li, C., Zeng, F.: Numerical Methods for Fractional Calculus. CRC Press, Boca Raton (2015)
Zhou, Y.: Fractional Evolution Equations and Inclusions: Analysis and Control. Academic Press, San Diego (2016)
Wang, Y., Liu, L., Wu, Y.: Positive solutions for a nonlocal fractional differential equation. Nonlinear Anal. 74, 3599–3605 (2011)
Zhang, X., Liu, L., Wu, Y.: Existence results for multiple positive solutions of nonlinear higher order perturbed fractional differential equations with derivatives. Appl. Math. Comput. 219, 1420–1433 (2012)
Wang, Y., Liu, L., Zhang, X., Wu, Y.: Positive solutions of a fractional semipositone differential system arising from the study of HIV infection models. Appl. Math. Comput. 258, 312–324 (2015)
Li, M., Wang, J.: Exploring delayed Mittag-Leffler type matrix functions to study finite time stability of fractional delay differential equations. Appl. Math. Comput. 324, 254–265 (2018)
Wang, J., Ibrahim, A.G., O’Regan, D.: Topological structure of the solution set for fractional non-instantaneous impulsive evolution inclusions. J. Fixed Point Theory Appl. 20(2), Article ID 59 (2018)
Wang, J., Ibrahim, A.G., O’Regan, D.: Hilfer type fractional differential switched inclusions with noninstantaneous impulsive and nonlocal conditions. Nonlinear Anal., Model. Control 23(6), 921–941 (2018)
Li, M., Debbouche, A., Wang, J.: Relative controllability in fractional differential equations with pure delay. Math. Methods Appl. Sci. 41, 8906–8914 (2018)
Zhang, X., Mao, C., Liu, L., Wu, Y.: Exact iterative solution for an abstract fractional dynamic system model for bioprocess. Qual. Theory Dyn. Syst. 16, 205–222 (2017)
Zhang, X., Liu, L., Wu, Y., Wiwatanapataphee, B.: Nontrivial solutions for a fractional advection dispersion equation in anomalous diffusion. Appl. Math. Lett. 66, 1–8 (2017)
Jiang, J., Liu, L., Wu, Y.: Multiple positive solutions of singular fractional differential system involving Stieltjes integral conditions. Electron. J. Qual. Theory Differ. Equ. 43, 1–18 (2012)
Klimek, M.: Sequential fractional differential equations with Hadamard derivative. Commun. Nonlinear Sci. Numer. Simul. 16, 4689–4697 (2011)
Ma, Q., Wang, R., Wang, J., Ma, Y.: Qualitative analysis for solutions of a certain more generalized two-dimensional fractional differential system with Hadamard derivative. Appl. Math. Comput. 257, 436–445 (2014)
Singh, J., Kumar, D., Baleanu, D.: New aspects of fractional Biswas–Milovic model with Mittag-Leffler law. Math. Model. Nat. Phenom. 14, 303 (2019)
Singh, J., Kumar, D., Baleanu, D., Rathored, S.: An efficient numerical algorithm for the fractional Drinfeld–Sokolov–Wilson equation. Appl. Math. Comput. 335, 12–24 (2018)
Kumar, D., Singh, J., Baleanu, D., Rathore, S.: Analysis of a fractional model of Ambartsumian equation. Eur. Phys. J. Plus 133, 259 (2018)
Kumar, D., Singh, J., Baleanu, D., Sushila: Analysis of regularized long-wave equation associated with a new fractional operator with Mittag-Leffler type kernel. Physica A 492, 155–167 (2018)
Kumar, D., Tchier, F., Singh, J., Baleanu, D.: An efficient computational technique for fractal vehicular traffic flow. Entropy 20, 259 (2018)
Heymans, N., Podlubny, I.: Physical interpretation of initial conditions for fractional differential equations with Riemann–Liouville fractional derivatives. Rheol. Acta 45, 765–771 (2006)
Ye, H., Gao, J., Ding, Y.: A generalized Gronwall inequality and its application to a fractional differential equation. J. Math. Anal. Appl. 328, 1075–1081 (2007)
Gorenflo, R., Kilbas, A.A., Mainardi, F., Rogosin, S.V.: Mittag-Leffler Functions, Related Topics and Applications. Springer, Berlin (2014)
Cong, N.D., Doan, T.S., Siegmund, S., Tuan, H.T.: On stable manifolds for planar fractional differential equations. Appl. Math. Comput. 226, 157–168 (2014)
Dzherbashyan, M.M.: Integral Transforms and Representations of Functions in the Complex Plane. Nauka, Moscow (1966) (in Russian)
Gorenflo, R., Loutchko, J., Luchko, Y.: Computation of the Mittag-Leffler function \(E_{\alpha,\beta }(z)\) and its derivative. Fract. Calc. Appl. Anal. 5(4), 491–518 (2002) Correction: Fract. Calc. Appl. Anal. 6(1), 111–112 (2003)
Acknowledgements
The authors are grateful to the referees for their careful reading of the manuscript and their valuable comments.
Availability of data and materials
Not applicable.
Funding
This work is supported by the National Natural Science Foundation of China (11661016), Training Object of High Level and Innovative Talents of Guizhou Province ((2016)4006), Science and Technology Program of Guizhou Province ([2017]5788-10), Major Research Project of Innovative Group in Guizhou Education Department ([2018]012), the Slovak Research and Development Agency under the contract No. APVV-14-0378, and the Slovak Grant Agency VEGA No. 2/0153/16 and No. 1/0078/17.
Author information
Authors and Affiliations
Contributions
All authors contributed equally to this paper. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Fečkan, M., Pospíšil, M. & Wang, J. Note on weakly fractional differential equations. Adv Differ Equ 2019, 143 (2019). https://doi.org/10.1186/s13662-019-2086-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13662-019-2086-4