Skip to main content
Log in

Numerical stability of Grünwald–Letnikov method for time fractional delay differential equations

  • Published:
BIT Numerical Mathematics Aims and scope Submit manuscript

Abstract

This paper is concerned with the numerical stability of time fractional delay differential equations (F-DDEs) based on Grünwald–Letnikov (GL) approximation for the Caputo fractional derivative. In particular, we focus on the numerical stability region and the Mittag–Leffler stability. Using the boundary locus technique, we first derive the exact expression of the numerical stability region in the parameter plane, and show that the fractional backward Euler scheme is not \(\tau (0)\)-stable, which is different from the backward Euler scheme for integer DDE models. Secondly, we prove the numerical Mittag–Leffler stability for the numerical solutions provided that the parameters fall into the numerical stability region, by employing the singularity analysis of generating function. Our results show that the numerical solutions of F-DDEs are completely different from the classical integer order DDEs, both in terms of \(\tau (0)\)-stability and the long-time decay rate. Numerical examples are given to confirm the theoretical results.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

Similar content being viewed by others

Notes

  1. The linear multistep method (LMM) with generating polynomials \(\rho (\zeta )\) and \(\sigma (\zeta )\) applying to the test equation gives the characteristic equation \(\rho (\zeta )-\mu \sigma (\zeta )=0\), where \(\mu =h\lambda \). For backward Euler method, we have \(\rho (\zeta )=\zeta -1\) and \(\sigma (\zeta )=\zeta \). The numerical stability region \(S_{h}=\{\mu \in \mathbb {C}: \text { all roots }\zeta _j(\mu ) \text { of the characteristic equation satisfy }|\zeta _j(\mu )|<1\}.\) Sometimes one defines \(S_{h}=\{\mu \in \mathbb {C}: \text { all roots }\zeta _j(\mu ) \text { of the characteristic equation satisfy }|\zeta _j(\mu )|\le 1, \text { multiple roots satisfy } |\zeta _j(\mu )|< 1\}\), which is generally called the weak linear stability region. The LMM is called A-stable if \(S_h \supset S_{*}\), i.e., \(S_h \supset \mathbb {C}^{-}\). See more details in [16, Chapter V].

References

  1. Bellen, A., Zennaro, M.: Numerical Methods for Delay Differential Equations. Oxford University Press (2013)

  2. Brunner, H.: Volterra Integral Equations: An Introduction to Theory and Applications, vol. 30. Cambridge University Press (2017)

  3. Bu, S.: Well-posedness of vector-valued fractional differential equations with delay. Sci. Sin. Math. 49, 1–10 (2019)

    MATH  Google Scholar 

  4. Burov, S., Barkai, E.: Fractional Langevin equation: overdamped, underdamped, and critical behaviors. Phys. Rev. E 78(3), 031112 (2008)

    Article  MathSciNet  Google Scholar 

  5. Čermák, J., Došlá, Z., Kisela, T.: Fractional differential equations with a constant delay: stability and asymptotics of solutions. Appl. Math. Comput. 298, 336–350 (2017)

    MathSciNet  MATH  Google Scholar 

  6. Čermák, J., Horníček, J., Kisela, T.: Stability regions for fractional differential systems with a time delay. Commun. Nonlinear Sci. Numer. Simul. 31(1–3), 108–123 (2016)

    Article  MathSciNet  Google Scholar 

  7. Čermák, J., Nechvátal, L.: On exact and discretized stability of a linear fractional delay differential equation. Appl. Math. Lett. 105, 106296 (2020)

    Article  MathSciNet  Google Scholar 

  8. Chen, H., Stynes, M.: Blow-up of error estimates in time-fractional initial-boundary value problems. IMA J. Numer. Anal. 41(2), 974–997 (2021)

    Article  MathSciNet  Google Scholar 

  9. Chen, Z.Q.: Time fractional equations and probabilistic representation. Chaos Solitons Fractals 102, 168–174 (2017)

    Article  MathSciNet  Google Scholar 

  10. Dabiri, A., Butcher, E.: Numerical solution of multi-order fractional differential equations with multiple delays via spectral collocation methods. Appl. Math. Mod. 56, 424–448 (2018)

    Article  MathSciNet  Google Scholar 

  11. Diethelm, K.: The Analysis of Fractional Differential Equations: An Application-Oriented Exposition Using Differential Operators of Caputo Type. Springer (2010)

  12. Flajolet, P., Odlyzko, A.: Singularity analysis of generating functions. SIAM J. Discret. Math. 3(2), 240–261 (1990)

    Article  MathSciNet  Google Scholar 

  13. Flajolet, P., Sedgewick, R.: Analytic Combinatorics. Cambridge University Press (2009)

  14. Garrappa, R., Kaslik, E.: On initial conditions for fractional delay differential equations. Commun. Nonlinear Sci. Numer. Simul. 90, 105359 (2020)

    Article  MathSciNet  Google Scholar 

  15. Guglielmi, N.: Delay dependent stability regions of \(\theta \)-methods for delay differential equations. IMA J. Numer. Anal. 18(3), 399–418 (1998)

    Article  MathSciNet  Google Scholar 

  16. Hairer, E., Wanner, G.: Solving Ordinary Differential Equations II Stiff and Differential-Algebraic Problems. Sciencep, Beijing (2006)

    MATH  Google Scholar 

  17. Huang, C., Hu, Y., Tian, H.: Delay-dependent stability analysis of multistep methods for delay differential equations. Acta Math. Appl. Sin. Engl. Ser. 25(4), 607–616 (2009)

    Article  MathSciNet  Google Scholar 

  18. Kaslik, E., Sivasundaram, S.: Analytical and numerical methods for the stability analysis of linear fractional delay differential equations. J. Comput. Appl. Math. 236(16), 4027–4041 (2012)

    Article  MathSciNet  Google Scholar 

  19. Kou, S.C.: Stochastic modeling in nanoscale biophysics: subdiffusion within proteins. Ann. Appl. Stat. 2(2), 501–535 (2008)

    Article  MathSciNet  Google Scholar 

  20. Lahrouz, A., Hajjami, R., El Jarroudi, M., Settati, A.: Mittag–leffler stability and bifurcation of a nonlinear fractional model with relapse. J. Comput. Appl. Math. 386, 113247 (2021)

    Article  MathSciNet  Google Scholar 

  21. Li, L., Liu, J.-G.: A generalized definition of Caputo derivatives and its application to fractional ODEs. SIAM J. Math. Anal. 50(3), 2867–2900 (2018)

    Article  MathSciNet  Google Scholar 

  22. Li, L., Liu, J.-G.: Some compactness criteria for weak solutions of time fractional PDEs. SIAM J. Math. Anal. 50(4), 3963–3995 (2018)

    Article  MathSciNet  Google Scholar 

  23. Li, L., Liu, J.-G., Lu, J.: Fractional stochastic differential equations satisfying fluctuation-dissipation theorem. J. Stat. Phys. 169(2), 316–339 (2017)

    Article  MathSciNet  Google Scholar 

  24. Li, L., Wang, D.: Complete monotonicity-preserving numerical methods for time fractional ODEs. Commun. Math. Sci. 19(5), 1301–1336 (2021)

    Article  MathSciNet  Google Scholar 

  25. Lubich, C.: A stability analysis of convolution quadraturea for Abel–Volterra integral equations. IMA J. Numer. Anal. 6(1), 87–101 (1986)

    Article  MathSciNet  Google Scholar 

  26. Maleki, M., Davari, A.: Fractional retarded differential equations and their numerical solution via a multistep collocation method. Appl. Numer. Math. 143, 203–222 (2019)

    Article  MathSciNet  Google Scholar 

  27. Matignon, D.: Stability results for fractional differential equations with applications to control processing. Comput. Eng. Syst. Appl. 2(1), 963–968 (1996)

    Google Scholar 

  28. Metzler, R., Klafter, J.: The random walk’s guide to anomalous diffusion: a fractional dynamics approach. Phys. Rep. 339(1), 1–77 (2000)

  29. Tuan, H., Siegmund, S.: Stability of scalar nonlinear fractional differential equations with linearly dominated delay. Fract. Calc. Appl. Anal. 23(1), 250–267 (2020)

    Article  MathSciNet  Google Scholar 

  30. Tuan, H., Trinh, H.: A qualitative theory of time delay nonlinear fractional-order systems. SIAM J. Control Optim. 58(3), 1491–1518 (2020)

    Article  MathSciNet  Google Scholar 

  31. Wang, D., Xiao, A., Liu, H.: Dissipativity and stability analysis for fractional functional differential equations. Fract. Calc. Appl. Anal. 18(6), 1399–1422 (2015)

    Article  MathSciNet  Google Scholar 

  32. Wang, D., Zou, J.: Dissipativity and contractivity analysis for fractional functional differential equations and their numerical approximations. SIAM J. Numer. Anal. 57(3), 1445–1470 (2019)

    Article  MathSciNet  Google Scholar 

  33. Wang, D., Zou, J.: Mittag–leffler stability of numerical solutions for fractional ODEs. arXiv preprint arXiv:2108.09620 (2021)

  34. Widder, D.V.: The Laplace Transform. Princeton University Press (1941)

  35. Zayernouri, M., Cao, W., Zhang, Z., Karniadakis, G.: Spectral and discontinuous spectral element methods for fractional delay equations. SIAM J. Sci. Comput. 36(6), B904–B929 (2014)

    Article  MathSciNet  Google Scholar 

  36. Zhang, Q., Li, T.: Asymptotic stability of compact and linear \(\theta \)-methods for space fractional delay generalized diffusion equation. J. Sci. Comput. 81(3), 2413–2446 (2019)

    Article  MathSciNet  Google Scholar 

  37. Zhao, J., Jiang, X., Xu, Y.: Generalized adams method for solving fractional delay differential equations. Math. Comput. Simul. 180, 401–419

Download references

Acknowledgements

The authors would like to thank two anonymous referees for their very careful reading and insightful comments that have helped us improve the presentation of the paper significantly. The work of L. Li was partially supported by NSFC 11901389, 12031013, Shanghai Sailing Program 19YF1421300. The work of D. L. Wang was partially supported by NSFC 11871057, 11931013, 12071403.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Dongling Wang.

Additional information

Communicated by Mihaly Kovacs.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Proof of Lemma 4.1

Proof of Lemma 4.1

For (i), we need to show that

$$\begin{aligned} 2^{\alpha }<2\frac{[(1-\alpha )\pi ]^{\alpha }\sin (\alpha \pi /2)}{\sin (\alpha \pi )} =\frac{[(1-\alpha )\pi ]^{\alpha }}{\sin (\pi (1-\alpha )/2)}, \end{aligned}$$

by Lemma 2.2. This is equivalent to

$$\begin{aligned} \sin \left( \frac{(1-\alpha )\pi }{2}\right) <\left[ \frac{(1-\alpha )\pi }{2}\right] ^{\alpha }. \end{aligned}$$

When \(\frac{(1-\alpha )\pi }{2}>1\), the inequality clearly holds. If \(\frac{(1-\alpha )\pi }{2}\in (0, 1]\), then

$$\begin{aligned} \sin \left( \frac{(1-\alpha )\pi }{2}\right) <\frac{(1-\alpha )\pi }{2} \le \left[ \frac{(1-\alpha )\pi }{2}\right] ^{\alpha }\quad \text {as}\quad \alpha \in (0, 1). \end{aligned}$$

(ii). By Lemma 2.2, we only need to show that the straight lines are below the boundary,

$$\begin{aligned} 2^{\alpha }k^{\alpha }\ge \frac{[(1-\alpha )\pi ]^{\alpha }}{\sin (\pi (1-\alpha )/2)},~~k\ge 3. \end{aligned}$$
(A.1)

It suffices to show that

$$\begin{aligned} \begin{aligned} \sin \left( \frac{\pi (1-\alpha )}{2}\right) - \left[ \frac{(1-\alpha )\pi }{2k} \right] ^{\alpha }\ge 0. \end{aligned} \end{aligned}$$
(A.2)

It’s worth noting that the function \(h_1(x)= \frac{\sin x}{x}\) is decreasing with \(\frac{2}{\pi }\le h_1(x)\le 1\) and it is concave on \([0, \pi /2]\) (note that the sign of the second order derivative is determined by the sign of \((1-\frac{x^2}{2})\frac{\sin x}{x}-\cos x\), which is negative for \(x\in [0, \pi /2]\)).

For \(k\ge 5\), since \(\sin (\pi (1-\alpha )/2)\ge \frac{\pi }{2}(1-\alpha )\frac{2}{\pi }\), it suffices to show that

$$\begin{aligned} \frac{\pi }{2}(1-\alpha )\frac{2}{\pi }- \left[ \frac{(1-\alpha )\pi }{2k}\right] ^{\alpha }\ge 0. \end{aligned}$$

The latter is equivalent to show that (setting \(\beta =1-\alpha \)) \(h_2(\beta )=\beta \log \beta -(1-\beta )\log \left( \frac{\pi }{2k} \right) \ge 0.\) The derivative \(h'_2(\beta )\) is negative for \(k\ge 5\) and the value at \(\beta =1\) is zero. Hence, the inequality (A.2) holds for \(k\ge 5\).

If \(k=3\), consider directly (setting \(\beta =1-\alpha \)) that \( g(\beta ):=\log \left( \sin (\frac{\pi }{2}\beta ) \right) +(\beta -1)\log \left( \frac{\beta \pi }{6}\right) . \) As \(\beta \rightarrow 1^-\), \(g(\beta )\) tends to 0. Hence, to show \(g(\beta )>0\) for \(\beta \in (0,1)\), we only need to show that the first order derivative

$$\begin{aligned} g'(\beta )=\frac{\pi }{2}\frac{\cos (\pi \beta /2)}{\sin (\pi \beta /2)}+\log \left( \frac{\pi }{6}\right) +\log (\beta )+(\beta -1)/\beta \end{aligned}$$

is always negative on \(\beta \in (0, 1)\).

It is easy to see that \(g'(\beta )<0\) for \(\beta \in (0.8, 1)\) (considering only the first two terms). On the other hand, we have \( g''(\beta )=\beta ^{-1}+\beta ^{-2}-(\pi /2)^2\frac{1}{\sin ^2(\pi \beta /2)}. \) This equation \(g''(\beta )=0\) only has one root \(\beta _*\) with \(\beta \in (0, 1)\) and the root satisfies \( \frac{\sin (\pi \beta _*/2)}{(\pi \beta _*/2)}=\frac{1}{\sqrt{1+\beta _*}}. \) This can be seen from the fact \(h_1(x)=\frac{\sin x}{x}\) is decreasing and concave while the function \(\frac{1}{\sqrt{1+x}}\) is convex on \([0, \pi /2]\). At the same time, we can check \(g''(0.8)>0\) and \(g''(1)<0\). Hence, we know that the root \(\beta _*\in (0.8, 1)\) and \(g''(\beta )\) is positive on [0, 0.8]. Together with \(g'(0.8)<0\), we find that \(g'(\beta )\) is negative on (0, 0.8]. Then the first derivative is always negative on \(\beta \in (0, 1)\). Therefore, \(k=3\) is also proved. \(\square \)

Proof for the properties of the \(\Gamma _0\) curve

Proof of Lemma 4.3

Define \(\phi =\theta /2\in \left( \frac{(1-\alpha )\pi }{2(1-\alpha /k)}, \frac{\pi }{2} \right) \). Consider that

$$\begin{aligned} h(\phi ):=\frac{\lambda _k(\theta )}{2^{\alpha }k^{\alpha }}=\frac{\sin ^{\alpha } \left( \frac{\phi }{k} \right) \sin \left( \phi +\frac{\alpha \pi }{2} -\frac{\alpha \phi }{k}\right) }{\sin (\phi )}. \end{aligned}$$

To show this function is decreasing, it is sufficient to prove that

$$\begin{aligned} \frac{d}{d\phi }\ln h(\phi )=\frac{\alpha }{k}\frac{\cos (\phi /k)}{\sin (\phi /k)}+\frac{(1-\alpha /k)\cos \left( \phi +\frac{\alpha \pi }{2}-\frac{\alpha \phi }{k}\right) }{\sin \left( \phi +\frac{\alpha \pi }{2} -\frac{\alpha \phi }{k} \right) }-\frac{\cos \phi }{\sin \phi } \le 0.\nonumber \\ \end{aligned}$$
(B.1)

Separate the middle term on the left side of the above equation and use the trigonometric function formula, the above equation can be equivalent to \( \frac{\alpha }{k}\frac{\sin (\phi +\alpha \pi /2-\alpha \phi /k-\phi /k)}{\sin (\phi /k)\sin (\phi +\alpha \pi /2-\alpha \phi /k)} -\frac{\sin (\alpha \pi /2-\alpha \phi /k)}{\sin (\phi )\sin (\phi +\alpha \pi /2-\alpha \phi /k)}\le 0. \) Noting that \(\sin (\phi +\alpha \pi /2-\alpha \phi /k)>0\) under the assumption \(\phi \in \left( \frac{(1-\alpha )\pi }{2(1-\alpha /k)}, \frac{\pi }{2} \right) \). Hence, we need

$$\begin{aligned} \frac{\alpha \sin (\phi +\alpha \pi /2-\alpha \phi /k-\phi /k)}{\sin (\alpha \pi /2 -\alpha \phi /k)}\le \frac{k\sin (\phi /k)}{\sin \phi }. \end{aligned}$$
(B.2)

We are going to prove that (B.2) is true for \(k\ge 2\). Since

$$\begin{aligned} \frac{\alpha \sin (\phi +\alpha \pi /2-\alpha \phi /k-\phi /k)}{\sin (\alpha \pi /2-\alpha \phi /k)}\le \frac{\alpha }{\sin (\alpha \pi /2-\alpha \phi /k)}\le \frac{1}{\cos (\phi /k)}, \end{aligned}$$
(B.3)

where the second inequality is due to the fact \(\frac{\alpha }{\sin (\alpha \pi /2-\alpha \phi /k)}\) is increasing function for \(\alpha \in (0, 1)\). By the concavity of \(\sin (x)\) on \(x\in (0, \pi /2)\), we know that \( \sin (\phi )\le \frac{k}{2}\sin (2\phi /k ) \) is true for \(k\ge 2\), which implies that \( \frac{1}{\cos (\phi /k)}\le \frac{k\sin (\phi /k)}{\sin (\phi )}. \) This together with (B.3) shows that the inequality (B.2) is true for \(k\ge 2\). This completes the first part of the proof.

Using the above fact, to show that the line is below \(\Gamma _0\), we only have to show that \(\left( a_k^{(0)},-a_k^{(0)} \right) \) is above the line, where \(a_k^{(0)}\) is defined in Lemma 4.2. In other words, we need

$$\begin{aligned} 2^{\alpha }k^{\alpha }\ge 2^{\alpha }k^{\alpha }\frac{\cos ^{\alpha }(\phi _1)}{\cos (\alpha \phi _1)}, \quad \text {where}\quad \phi _1=\frac{\pi (k-1)}{2(k-\alpha )}. \end{aligned}$$
(B.4)

Since the right hand side is decreasing in \(\phi \), the largest value is achieved at \(\phi _1=0\), which is \(2^{\alpha }k^{\alpha }\). Hence, the inequality (B.4) is true. \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Li, L., Wang, D. Numerical stability of Grünwald–Letnikov method for time fractional delay differential equations. Bit Numer Math 62, 995–1027 (2022). https://doi.org/10.1007/s10543-021-00900-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10543-021-00900-0

Keywords

MSC Codes

Navigation