Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Analysis of the L1 scheme for fractional wave equations with nonsmooth data

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read For nonsmooth data, the L1 scheme reaches O(tau^{3-alpha}) in L2 at positive times, if tau^alpha/h^2 stays bounded.

desk verdict Genuine new result for the standard L1 scheme; the modified-scheme O(τ^2) claim is asserted, not proved. read the letter →

arxiv 1908.09145 v2 pith:QPLR5MAX submitted 2019-08-24 math.NA cs.NA

classification math.NAcs.NA MSC 65M6065M1235R1165M15
keywords L1schemefractionalwaveequationnonsmoothdatafiniteelementmethoddiscreteLaplacetransformconvergenceanalysismodified
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper pins down what the well-known L1 time-stepping scheme actually delivers for fractional wave equations when the initial data are only square-integrable, not smooth. The central result is that a full discretization combining the L1 scheme in time with piecewise-linear finite elements in space has L2 error at positive times of order O($tau^{{3-alpha}}$), and a slightly modified L1 scheme attains the rate O($tau^{2}$). Both rates hold with nonsmooth data, and the inhomogeneous case with nonzero forcing is covered as well. The practical price is a grid condition: the time step tau and the smallest spatial mesh diameter h_min must satisfy tau^$\alpha$/$h_min^{2}$ uniformly bounded. If that condition fails, the paper's analysis says temporal accuracy deteriorates, and its Experiment 3 shows exactly that.

What carries the argument

The load-bearing object is the discrete Laplace transform of the time-stepping kernel. For the standard L1 scheme the kernel is b_j = $j^{{2-alpha}}$/Gamma(3-$\alpha$), whose transform satisfies \hat b(z) = \sum_{k=-\infty}^\infty (z + 2k\pi i)^{\$\alpha$-3} in a strip, and the companion symbol \psi(z) = $e^{{-z}}$(e^z-1)^3 \hat b(z) approximates z^\$\alpha$. The analysis compares \psi(z)+\mu(e^z+1) with the continuum symbol z^\$\alpha$ + 2\mu, where \mu = \$\lambda$ \tau^\$\alpha$/2, on a pair of rays avoiding the eigenvalues. The standard kernel leaves a nonzero constant in \hat b(z) - $z^{{\alpha-3}}$ at z=0, which is what costs the 3-$\alpha$ rate. The modified scheme changes only beta_1, adding a constant that forces (\hat\$\beta$(z)-$z^{{\alpha-3}}$)(0)=0 and removes that leading error, yielding second order.

What would settle it

Run the standard L1 scheme on the nonsmooth-data test u0(x)=$x^{{-0.49}}$ in (0,1), keep tau fixed, and halve h repeatedly so tau^$\alpha$/$h^{2}$ leaves the bounded regime; the observed temporal error at t=1 should stop decaying like $tau^{{3-alpha}}$ and should worsen as h shrinks. Alternatively, repeat with a nonuniform time grid: a clean O($tau^{{3-alpha}}$) rate there would contradict the paper's stated restriction, since the analysis does not cover that case.

Watch

Extended reading notes

Core claim

The paper's target is the L2($\Omega$)-norm convergence of two fully discrete time-stepping methods for the fractional wave equation. For the standard L1 scheme, Theorem 4.2 gives, at each positive time t_k, an error bounded by ($t_k^{{alpha-3}}$ $tau^{{3-alpha}}$ + $t_k^{{-alpha}}$ $h^{2}$) times the L2 norm of u0, plus analogous contributions from u1 and the forcing term; in particular, the temporal order is 3-$\alpha$. For the modified L1 scheme, Theorem 4.3 gives temporal order 2. The proofs write both the exact and numerical solutions as contour integrals over a pair of rays, using the discrete Laplace transform of the time-stepping kernel, and then compare the integrands. The modified kernel is constructed so that the leading symbol error at z=0 vanishes, which upgrades the temporal rate from 3-$\alpha$ to 2.

Load-bearing premise

The claimed rates collapse if the time grid is not uniform or if tau^$\alpha$/$h_min^{2}$ is not uniformly bounded, since the proof constants blow up as that ratio grows.

Editorial extensions

If this is right

  • Practitioners can use the standard L1 scheme for fractional wave equations with nonsmooth initial values and still expect temporal convergence in L2 of order O(tau^{3-alpha}), rather than only first order.
  • The modified L1 scheme gives a genuine O(tau^2) temporal rate with nonsmooth data, which is useful when long-time accuracy is limited by the initial-data singularity.
  • For inhomogeneous problems, the error is controlled by f(0) and the L1 norm of f', with the same temporal exponents and a logarithmic factor at alpha = 3/2.
  • Spatial discretization with piecewise-linear elements contributes the expected O(h^2) term up to a logarithmic factor, so temporal and spatial errors balance naturally when the two contributions are comparable.
  • A user must monitor tau^alpha/h_min^2; refining space alone while keeping tau fixed can destroy the temporal accuracy rather than improve the total error.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: the symbol-correction trick that fixes beta_1 could be tried on nonuniform time grids by replacing the global correction with a local one, though the paper explicitly says its techniques do not apply there.
  • Beyond the paper: the tau^alpha/h^2 condition behaves like a CFL-type restriction, so one could test whether local time stepping with tau roughly proportional to h^{2/alpha} preserves the proved rates at lower computational cost.
  • Beyond the paper: an analogous modified L1 scheme might yield O(tau^2) for variable-coefficient or nonlinear fractional wave equations, but the present analysis gives no evidence beyond the linear constant-coefficient setting.
  • Beyond the paper: because the limiting step is a one-dimensional symbol comparison, the same Laplace-transform machinery could be reused to derive sharp constants or to predict when the 3-alpha rate can be improved for smoother data.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper analyzes the L1 scheme and a modified L1 scheme for a fractional wave equation D_{0+}^{\alpha-1}(u'-u_1)-\Delta u=f with 1<\alpha<2 and nonsmooth initial data. For a full discretization consisting of the L1 rule in time and P1 finite elements in space, the authors claim a new stability estimate and temporal accuracy O(\tau^{3-\alpha}) in the L2 norm at positive times under a uniform temporal grid and a uniformly bounded ratio \tau^\alpha/h_{\min}^2. A modified L1 scheme with one altered coefficient is proposed and claimed to achieve temporal accuracy O(\tau^2). Convergence in the inhomogeneous case is also stated. The proofs for the standard L1 scheme use Laplace transform representations, sector arguments, and contour estimates following the framework of Jin-Lazarov-Zhou. Numerical experiments in Section 5 verify the predicted orders and also demonstrate the deterioration of accuracy for large \tau^\alpha/h^2.

Significance. If the main results are valid, the paper fills a genuine gap: it establishes L2-norm temporal accuracy O(\tau^{3-\alpha}) for the L1 scheme for fractional wave equations with nonsmooth data, a result not previously available in the literature. The paper also identifies the condition \tau^\alpha/h_{\min}^2\lesssim 1 as important for the full discretization, and it documents numerically that large values of this ratio degrade accuracy, which is a useful practical warning. The analysis of the standard L1 scheme is detailed and follows the established Laplace-transform route; the stability proofs for both schemes are given, and the numerical experiments are extensive and match the predicted rates, including the predicted suboptimal behavior in the large-ratio regime. The modified-scheme result, however, is a headline claim that is currently not actually proved in the manuscript: Theorem 3.3 is asserted by reference to a 'simple modification' of earlier proofs, with no verification that the modified symbol satisfies the necessary sector, lower-bound, and cancellation estimates.

major comments (3)
  1. [Section 3.3, Theorem 3.3] The modified L1 scheme's O(\tau^2) error estimate is not proved. The text says only that 'by a simple modification of the proofs of Theorems 3.1 and 3.2, we readily obtain the following error estimate.' The proof of Theorem 3.1 relies on Lemma 3.3 (a zero-free sector for \psi(z)+\mu(1+e^z)), Lemma 3.7 (the lower bound |\psi(z)+\mu(1+e^z)| \gtrsim \mu+|z|^\alpha), and a delicate cancellation estimate bounding the integrand difference by |z|^{\alpha+2}+\mu|z|^{3-\alpha}+\mu^2|z|^{2-\alpha}. For the modified symbol \hat\beta, the corresponding kernel \Psi(z)=e^{-z}(e^z-1)^3\hat\beta(z) must be shown to satisfy the same sector condition, the same lower bound, and the improved cancellation error \mu|z|^2+|z|^{2+\alpha}. Remarks 3.8 and 3.9 only state the desired bounds; no proof is supplied for any of these three requirements. The numerical tables in Section 5 are consistent with O(\tau^2), but they do not replace the missing proof. Because Theorem 4.3 and the abstract's modified-scheme claim rest on Theorem 3.3, the full proof must be provided, or the claim must be explicitly downgraded to a conjecture supported by numerics.
  2. [Section 4, Theorems 4.2 and 4.3] The passage from the scalar ODE theorems to the full discretizations is compressed into the sentence 'By the above procedure, we have the following two theorems.' For the standard L1 scheme this is probably routine but should still be written out at least as a lemma: one needs to apply Theorems 3.1 and 3.2 mode by mode with \lambda=\lambda_i, to verify that the constants are uniform under the standing assumption \tau^\alpha/h_{\min}^2\le \mu_0, and to justify summing the f-dependent terms in Theorem 3.2 over the eigenbasis of -\Delta_h. For the modified scheme the procedure is not routine at all, because Theorem 3.3 itself is unproved. The paper should either include the detailed eigenvalue decomposition argument or state clearly which part of the proof is being deferred.
  3. [Section 4.2, Lemma 4.1] The proof of the semidiscrete spatial estimate (56) uses the resolvent bound \|(z^\alpha-\Delta)^{-1}-(z^\alpha-\Delta_h)^{-1}P_h\|_{\mathcal{L}(L^2)} \lesssim h^2 for all z\in\Upsilon\setminus\{0\}, citing [21, Theorem 2.1]. The uniformity in z on the whole contour \Upsilon is essential for the subsequent integration (\int e^{sz} h^2 dz \sim s^{-1}h^2). Please state the exact form of the resolvent bound that is being quoted and explain why it holds uniformly along the contour, including near the origin. If the known bound has an additional factor such as |z|^{-\alpha}, the integrals in the proof of Lemma 4.1 would need to be re-examined.
minor comments (5)
  1. [Abstract and Introduction] The abstract and introduction contain grammatical slips, e.g. 'The convergence of these schemes in inhomogeneous case are also established' and 'As a extension of integer order equation'; these should be corrected in a final pass.
  2. [Section 3.2.3] After defining \mathcal{E}(t):=\int_0^t(E-\tilde E)(s)ds, the proof of Lemma 3.9 immediately writes E(t) for the same quantity, which conflicts with the kernel E(t) defined in (30). Please use a distinct symbol for the integrated error consistently throughout the proof.
  3. [Section 3.2.2, Lemma 3.7] In the proof of Lemma 3.7 the expansion of (1+re^{i\theta})^{-1}\psi(re^{i\theta}) is written with a factor 1/2, namely r^\alpha e^{i\alpha\theta}/2 + r^{\alpha+1}g(r), whereas (19) states \psi(re^{i\theta})/(r^\alpha(\cos\alpha\theta+i\sin\alpha\theta))\to 1, which suggests the leading term should be r^\alpha e^{i\alpha\theta} without the factor 1/2. Please check and correct this, or explain the convention used for the expansion.
  4. [Section 5.2, Table 6] In Table 6, for problem (e) with \alpha=1.2 and h=2^{-8}, the reported Error2 value 3.33e-5 appears inconsistent with the previous row (1.29e-5 at h=2^{-7}) and the stated order 1.96; it should probably be 3.33e-6. Please verify the entry.
  5. [Section 3 and Section 4] The theorems involving f(0) assume f'\in L^1; in that case f is absolutely continuous and f(0) is well defined as a trace, but this should be stated explicitly for the reader, especially in the PDE setting where f takes values in L^2(\Omega).

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the O(tau^{3-alpha}) bound is derived from contour-integral estimates, while the modified O(tau^2) bound is asserted without proof (a completeness gap, not a circular reduction).

full rationale

The central derivation is self-contained. Theorem 3.1 represents the exact and numerical solutions by contour integrals (Lemmas 3.5 and 3.6), proves symbol lower bounds (Lemmas 3.3 and 3.7), and bounds the integrand differences, producing O(tau^{3-alpha}); Theorem 4.2 then applies this mode-by-mode to the discrete Laplacian eigenbasis. No error term is fitted from data and no target bound is used as an input. The modified scheme in Discretization 2 is constructed analytically: beta_1 is chosen in Remark 3.3 so that (hat beta(z) - z^{alpha-3})(0) = 0, which is a design condition, not a fit to numerical output. However, the paper's second headline result, Theorem 3.3, is not proved: Section 3.3 states 'Finally, by a simple modification of the proofs of Theorems 3.1 and 3.2, we readily obtain the following error estimate' and then asserts the O(tau^2) bound; the required sector and cancellation estimates for the modified symbol are not shown. This is an omitted proof and an unsupported claim, but it is not a circular equivalence, so it does not raise the circularity score. Self-citations to [22] and [16] occur for technical lemmas and equivalence remarks and are not the load-bearing argument of the main theorems.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The central claims rest on standard results in fractional calculus and Laplace transform theory, plus explicit regularity and mesh assumptions. No free parameters are fitted to data, and no new entities are introduced. The main load-bearing assumptions are the uniform grid and bounded tau^alpha/h^2_min, which the paper tests numerically.

assumptions (7)
  • standard math Lemma 2.1 fractional calculus inequalities: for 0<gamma<1/2, C1||D^gamma_{a+}v||^2 <= Integral (D^gamma_{a+}v, D^gamma_{b-}v) <= C2||D^gamma_{a+}v||^2, and the duality relation for D^{2gamma}.
    Used throughout to bound the energy terms in stability proofs (Section 2, Lemma 2.1; proof cited to Ervin and Roop [8]).
  • standard math Exact solution representation via Laplace transform: y(t) = (1/2pi i) Integral_Gamma e^{(t/tau)z} (y0 z^{alpha-1} + tau y1 z^{alpha-2})/(z^alpha + 2mu) dz + Integral_0^t E(t-s) f(s) ds.
    Used as the reference solution for the error analysis (Lemma 3.5, quoted from Jin et al. [12]).
  • standard math Discrete Laplace transform representation of b_k: bhat(z) = sum_{k=-infinity}^{infinity} (z + 2kpi i)^{alpha-3} for |Im z| < 2pi.
    Used to analyze the symbol of the L1 and modified L1 schemes (Equation (8), cited from Wood [35]).
  • ad hoc to paper Uniform temporal grid and bounded ratio tau^alpha/h^2_min (i.e., mu = lambda tau^alpha/2 <= mu0 for eigenvalues lambda of -Delta_h).
    The error estimates in Theorems 3.1, 3.2, 3.3, 4.2, and 4.3 depend on constants C_{alpha,mu0} that blow up as mu0 -> infinity; the paper states this requirement in the introduction and Remark 4.4.
  • domain assumption Data regularity: u0, u1 in L2(Omega), f in L1(0,infinity;L2) intersect 0H^{(1-alpha)/2}(0,infinity;H^{-1}); for inhomogeneous cases, f' in L1.
    The stability and convergence estimates are stated under these regularity conditions in Sections 3 and 4.
  • domain assumption Quasi-uniform and shape-regular triangulation K_h of the convex polytope Omega, with P1 finite element space S_h.
    Spatial discretization assumptions used in Section 4; the error bounds depend on the shape regularity and h_min.
  • standard math Properties of the discrete Laplacian -Delta_h eigenpairs (phi_i, lambda_i) used for eigen-expansion of the full discretization error.
    Used in Section 4 to reduce the PDE error analysis to the scalar ODE estimates of Section 3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Analysis of the L1 scheme for fractional wave equations with nonsmooth data." pith.science (2026). https://pith.science/paper/QPLR5MAX

@misc{pith2026190809145,
  author       = {Pith},
  title        = {Pith review of: Analysis of the L1 scheme for fractional wave equations with nonsmooth data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QPLR5MAX}},
  note         = {Machine review of arXiv:1908.09145}
}
abstract

This paper analyzes the well-known L1 scheme for fractional wave equations with nonsmooth data. A new stability estimate is obtained, and the temporal accuracy $ \mathcal O(\tau^{3-\alpha}) $ is derived for the nonsmooth initial data. In addition, a modified L1 scheme is proposed, and stability and temporal accuracy $ \mathcal O(\tau^2) $ are derived for this scheme with nonsmooth initial data. The convergence of the two schemes in the inhomogeneous case is also established. Finally, numerical experiments are performed to verify the theoretical results.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Numerical analysis of a semilinear fractional diffusion equation

    math.NA 2019-08 conditional novelty 6.0 of 10

    For nonsmooth initial data, the authors derive optimal spatial and temporal error bounds for a semilinear fractional diffusion equation using a new Gronwall inequality.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages · cited by 1 Pith paper

  1. [22]

    H. Luo, B. Li, and X. Xie. Convergence analysis of a Petrov-Gale rkin method for fractional wave problems with nonsmooth data. Journal of Scientific Computing , https://doi.org/10.1007/s10915-019-00962-x, 2019. 27

  2. [1]

    Adolfsson, M

    K. Adolfsson, M. Enelund, and S. Larsson. Adaptive discretizat ion of an integro-differential equation with a weakly singular convolution ke rnel. Computer Methods in Applied Mechanics and Engineering , 192(51):5285 – 5304, 2003

  3. [2]

    Arendt, C

    W. Arendt, C. Batty, M. Hieber, and F. Neubrander. Vector-valued Laplace transforms and Cauchy problems . Birkh¨ auser Basel, 2011

  4. [3]

    Bouchaud and A

    J. Bouchaud and A. Georges. Anomalous diffusion in disordered me dia: Statistical mechanisms, models and physical applications. Physics Reports, 195(45):127–293, 1990

  5. [4]

    Carpinteri, P

    A. Carpinteri, P. Cornetti, and A. G. Sapora. A fractional calcu lus ap- proach to nonlocal elasticity. European Physical Journal-special Topics , 193(1):193–204, 2011

  6. [5]

    F. Chen, Q. Xu, and J. Hesthaven. A multi-domain spectral meth od for time-fractional differential equations. Journal of Computational Physics , 293(1):157–172, 2015

  7. [6]

    Cuesta, C

    E. Cuesta, C. Lubich, and C. Palencia. Convolution quadrature t ime dis- cretization of fractional diffusion-wave equations. Mathematics of Compu- tation, 65(213):1–17, 1996

  8. [7]

    Cuesta and C

    E. Cuesta and C. Palencia. A fractional trapezoidal rule for inte gro- differential equations of fractional order in banach spaces. Applied Nu- merical Mathematics, 45(2):139–159, 2003

Show all 37 references
  1. [8]

    Ervin and J

    V. Ervin and J. Roop, Variational formulation for the stationary fractional advection dispersion equation. Numerical Methods for Partial Differential Equations, 22(3):558–576, 2006. 26

  2. [9]

    G. Gao, Z. Sun, and H. Zhang. A new fractional numerical differe ntiation formula to approximate the Caputo fractional derivative and its ap plica- tions. Journal of Computational Physics , 259(2):33–50, 2014

  3. [10]

    Gerolymatou, I

    E. Gerolymatou, I. Vardoulakis, and R. Hilfer. Modelling infiltratio n by means of a nonlinear fractional diffusion model. Journal of Physics D , 39(18):4104–4110, 2006

  4. [11]

    B. Jin, R. Lazarov, and Z. Zhou. An analysis of the L1 scheme fo r the subdiffusion equation with nonsmooth data. IMA Journal of Numerical Analysis, 36(1):197–221, 2016

  5. [12]

    B. Jin, R. Lazarov, and Z. Zhou. Two fully discrete schemes for fractional diffusion and diffusion-wave equations with nonsmooth data. SIAM journal on scientific computing , 38(1):A146–A170, 2016

  6. [13]

    A. A. Kilbas, H. M. Srivastava, and J. J. Trujillo. Theory and app lications of fractional differential equations. North-Holland Mathematics Studies, 2006

  7. [14]

    T. A. M. Langlands and B. I. Henry. The accuracy and stability o f an implicit solution method for the fractional diffusion equation. Journal of Computational Physics , 205(2):719–736, 2005

  8. [15]

    B. Li, H. Luo, and X. Xie. A time-spectral algorithm for fraction al wave problems. Journal of Scientific Computing , 7(2):1164–1184, 2018

  9. [16]

    B. Li, H. Luo, and X. Xie. Analysis of a time-stepping scheme for t ime fractional diffusion problems with nonsmooth data. SIAM Journal on Nu- merical Analysis (accepted), arXiv:1804.10552, 2019

  10. [17]

    Li and C

    X. Li and C. Xu. A space-time spectral method for the time fra ctional diffusion equation. SIAM Journal on Numerical Analysis , 47(3):2108–2131, 2009

  11. [18]

    H. Liao, D. Li, and J. Zhang. Sharp error estimate of the nonun iform L1 formula for linear reaction-subdiffusion equations. SIAM Journal on Numerical Analysis, 56(2):1112–1133, 2018

  12. [19]

    Lin and C

    Y. Lin and C. Xu. Finite difference/spectral approximations for the time-fractional diffusion equation. Journal of Computational Physics , 225(2):1533–1552, 2007

  13. [20]

    C. Lubich. Convolution quadrature and discretized operationa l calculus. Numerische Mathematik , 52(4):129–145, 1988

  14. [21]

    Lubich, I

    C. Lubich, I. Sloan, and V. Thom´ ee. Nonsmooth data error es timates for approximations of an evolution equation with a positive-type memory term. Mathematics of Computation , 65(213):1–17, 1996

  15. [23]

    McLean, V

    W. McLean, V. Thom´ ee, and L. B. Wahlbin. Numerical solution of an evolution equation with a positive type memory term. The Journal of the Australian Mathematical Society , B(35):23–70, 1993

  16. [24]

    McLean, V

    W. McLean, V. Thom´ ee, and L. B. Wahlbin. Discretization with va riable time steps of an evolution equation with a positive-type memory term . Journal of Computational and Applied Mathematics , 69(1):49 – 69, 1996

  17. [25]

    McLean, K

    W. McLean, K. Mustapha Time-stepping error bounds for frac tional dif- fusion problems with non-smooth initial data. Journal of Computational Physics, 293(1):201-217, 2015

  18. [26]

    McLean and K

    W. McLean and K. Mustapha. A second-order accurate numer ical method for a fractional wave equation. Numerische Mathematik , 105(3):481–510, 2007

  19. [27]

    Mustapha

    K. Mustapha. Time-stepping discontinuous Galerkin methods for fractio nal diffusion problems . Springer-Verlag New York, 2015

  20. [28]

    Mustapha and W

    K. Mustapha and W. McLean. Discontinuous Galerkin method for an evolution equation with a memory term of positive type. Mathematics of Computation, 78(268):1975–1995, 2009

  21. [29]

    Mustapha and W

    K. Mustapha and W. McLean. Piecewise-linear, discontinuous Ga lerkin method for a fractional diffusion equation. Numerical Algorithms , 56(2):159–184, 2011

  22. [30]

    Mustapha and W

    K. Mustapha and W. McLean. Superconvergence of a discontin uous Galerkin method for fractional diffusion and wave equations. SIAM Journal on Numerical Analysis , 51(1):491–515, 2012

  23. [31]

    R. R. Nigmatullin. The realization of the generalized transfer equ ation in a medium with fractal geometry. Physica Status Solidi B-basic Solid State Physics, 133(1):425–430, 1986

  24. [32]

    J. Ren, X. Long, S. Mao, and J. Zhang. Superconvergence of finite ele- ment approximations for the fractional diffusion-wave equation. Journal of Scientific Computing , 72(3): 917–935, 2017

  25. [33]

    Sun and X

    Z. Sun and X. Wu. A fully discrete difference scheme for a diffusion -wave system. Applied Numerical Mathematics , 56(2):193–209, 2006

  26. [34]

    L. Tartar. An Introduction to Sobolev Spaces and Interpolation Spaces . Springer Berlin, 2007

  27. [35]

    D. Wood. The computation of polylogarithms, technical report , 15-29. 1992

  28. [36]

    Y. Yan, M. Khan, and N. J. Ford. An analysis of the modified L1 sc heme for time-fractional partial differential equations with nonsmooth dat a. SIAM Journal on Numerical Analysis , 56(1):210–227, 2018

  29. [37]

    S. B. Yuste and L. Acedo. An explicit finite difference method and a new von neumann-type stability analysis for fractional diffusion equ ations. SIAM Journal on Numerical Analysis , 42(5):1862–1874, 2005. 28

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.