Pith. sign in

REVIEW 1 major objections 5 minor 31 references

Identification problems for anisotropic time-fractional subdiffusion equations

T0 review · 1 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that n energy measurements of a fractional-in-time subdiffusion equation at one fixed time uniquely determine the n unknown positive coefficients, with a conditioned existence statement and continuity as the fractional…

desk verdict The abstract uniqueness theorem for fractional subdiffusion is solid and genuinely extends Mola's alpha=1 result, but the advertised elasticity application does not meet the theorem's hypotheses as written. read the letter →

arxiv 2507.10315 v1 pith:SUNYBSVB submitted 2025-07-14 math.AP

Simone Creo , Maria Rosaria Lancia , Andrea Mola , Gianluca Mola , Silvia Romanelli This is my paper · ORCID
classification math.AP MSC 35R3035R1135K2047D0647B25
keywords identificationproblemsfractionaltimederivativeslinearevolutionequationsinHilbertspacesanisotropicdiffusionwell-posednessresultsinversecoefficientenergy-typemeasurementsGale–Nikaidounivalence
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 addresses a concrete identification question: can the unknown constant coefficients \(\lambda_1,\dots,\lambda_n\) in the anisotropic subdiffusion equation \(\partial_t^\$\alpha$ u(t)+\lambda_1A_1u(t)+\cdots+\lambda_nA_nu(t)=0\) be recovered from \(n\) energy measurements \(\langle A_1u(\bar T),u(\bar T)\rangle,\dots,\langle A_nu(\bar T),u(\bar T)\rangle\)? Its answer is affirmative for every fractional order \(0<\$\alpha$\le1\), provided the initial profile is admissible and the measured vector lies in the range of the solutions' input-output map. The proof localizes the whole difficulty in one nonlinear map: its Jacobian is always negative definite, so a classical global-univalence theorem makes the map one-to-one and the coefficients unique. A reader interested in applications would care because the theorem says that, in principle, a single set of \(n\) energy-type readings already carries enough information to identify anisotropic diffusivities or Lamé parameters in fractional models.

What carries the argument

The load-bearing machinery is the input–output map \[F_{\$\alpha$,i}(\$\lambda$)=\int_0^\infty\int_0^\infty \Phi_\$\alpha$(\tau)\Phi_\$\alpha$(\$\sigma$)\left\langle A_i $e^{{-\frac{\tau+\sigma}}${2}\bar T^\$\alpha$ A(\$\lambda$)}u_0, $e^{{-\frac{\tau+\sigma}}${2}\bar T^\$\alpha$ A(\$\lambda$)}u_0\right\rangle d\tau d\$\sigma$,\] where \(\Phi_\$\alpha$\) is the Wright probability density entering the fractional semigroup representation of the solution. The Jacobian of \(F_\$\alpha$\) is \(-2\bar T^\$\alpha$\) times an integrated Gram matrix of the transported vectors \(A_i $e^{{-\frac{\tau+\sigma}}${2}\bar T^\$\alpha$ A(\$\lambda$)}u_0\). Admissibility of \(u_0\) keeps those vectors linearly independent for all positive times, so the Gram matrix, and hence the Jacobian, is negative definite; the Gale–Nikaido global-univalence theorem then upgrades that pointwise definiteness to global injectivity of \(F_\$\alpha$\) on \(\mathbb{R}^n_+\).

What would settle it

Take \(A_1=A_2=I\) on \(H\), so the operators commute but any \(u_0\) is a common eigenfunction: admissibility fails, every pair \((\lambda_1,\lambda_2)\) with the same sum gives identical energy measurements, and \(F_\$\alpha$\) is plainly not injective. For a test inside the theorem's hypotheses, one can numerically evaluate the Gram determinant of \(A_i $e^{{-\frac{\tau+\sigma}}${2}\bar T^\$\alpha$ A(\$\lambda$)}u_0\); if it ever vanished for admissible \(u_0\), Lemma 3.4 and Proposition 4.1 would fail, while the proof predicts strict positivity at all positive times.

Watch

Extended reading notes

Core claim

The central claim is Theorem 3.5: for any admissible initial datum \(u_0\in\mathcal A\) and any compatible measurement vector \(\varphi\in\operatorname{Im}F_\$\alpha$\), the inverse problem \(P_\$\alpha$\) has exactly one solution \((u_\$\alpha$,\lambda_\$\alpha$)\) for \(0<\$\alpha$\le1\). The object that makes this work is the map \(F_\$\alpha$\) whose components are double integrals against the Wright density of the energy inner products of the semigroup-transported state. Because the Jacobian \(F'_\$\alpha$(\$\lambda$)\) is negative definite at every \(\$\lambda$\in\mathbb R^n_+\), the Gale–Nikaido theorem forces \(F_\$\alpha$\) to be injective on the whole positive orthant. Uniqueness of \(\lambda_\$\alpha$\) follows immediately, and existence is exactly the compatibility condition \(\varphi\in\operatorname{Im}F_\$\alpha$\), with the image open, bounded, and path-connected. In the limit \(\$\alpha$\to1^-\), the reconstructed coefficients and the solution converge to the classical diffusion inverse problem, so the fractional results form a continuous generalization of the \(\$\alpha$=1\) case.

Load-bearing premise

The result rests on the initial profile being admissible—the vectors \(A_1u_0,\dots,A_nu_0\) linearly independent in \(H\)—and on the operators commuting; if either fails, the Gram matrix behind the Jacobian is no longer definite and the injectivity proof collapses.

Editorial extensions

If this is right

  • Arbitrarily many constant coefficients can be recovered from the same number of energy measurements at a single time, for every fractional order \(0<\alpha\le1\).
  • The inverse problem has a solution exactly when the measured vector lies in \(\operatorname{Im}F_\alpha\), an open, path-connected, bounded region; this gives a precise compatibility criterion for experimental data.
  • As \(\alpha\to1^-\), both the reconstructed coefficients and the state converge to the classical parabolic solution, so fractional-order identification is continuously connected to the \(\alpha=1\) case.
  • In the heat-diffusion realization, the \(n\) gradient norms \(\|\nabla_{x_i}u(\bar T)\|_{L^2}^2\) determine the anisotropic diffusivities \(\lambda_i\), and in the elasticity application the two energy functionals determine the Lamé coefficients \(\lambda+\mu\) and \(\mu\).

Reading between the lines

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

  • A practical preprocessing test follows directly: compute the Gram matrix of \(A_1u_0,\dots,A_nu_0\); if its determinant is zero or very small, the measurements cannot be expected to separate the coefficients, so the initial datum should be changed before running an experiment.
  • Since \(F_\alpha\) varies continuously in \(\alpha\), a measurement vector that lies safely inside \(\operatorname{Im}F_\alpha\) for one order should remain compatible for nearby fractional orders; this could guide how experiments are designed when \(\alpha\) itself is uncertain.
  • A natural next question the paper does not answer is whether adding one more energy measurement at a second time would also identify the fractional order \(\alpha\), because the map is continuous in \(\alpha\) but continuity alone does not imply injectivity.
  • The commutativity requirement is structural: the paper recalls a counterexample at \(\alpha=1\) showing noncommuting operators break uniqueness, so any extension to noncommuting anisotropic models would need a genuinely new idea rather than a technical patch.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 5 minor

Summary. This paper studies the inverse problem of identifying the positive coefficients λ1,...,λn in the abstract fractional evolution equation ∂t^α u(t) + Σ_{i=1}^n λ_i A_i u(t) = 0 on a real Hilbert space H from n energy measurements ⟨A_i u(Tbar), u(Tbar)⟩ = φ_i. For 0 < α ≤ 1, the authors introduce the map Fα whose equation Fα(λ)=φ encodes the overdetermination, prove that its Jacobian is negative definite for admissible initial data (Definition 3.2), and use a Gale-Nikaido-type argument to conclude that Fα is injective. Theorem 3.5 consequently gives a unique solution for every compatible measurement φ∈Im Fα, with existence following from compatibility. The paper also proves that as α→1− the solutions of the fractional inverse problems converge to the classical α=1 solution (Theorem 3.6), and it presents applications to a product-domain heat equation and to a plane-elasticity Lamé-coefficient identification problem. Numerical simulations for n=2 illustrate the structure of the map Fα.

Significance. The central abstract uniqueness result is a genuine extension of the α=1 theory in [23] to Caputo-fractional evolution, and it is proved by a compact, parameter-free argument: negative definiteness of the Jacobian follows from the Gram matrix of the vectors A_i e^{-...}u0, so injectivity requires no quantitative hypotheses beyond admissibility and commutativity. The convergence theorem and the numerical study are useful complements, and Remark 4.3 correctly records that commutativity is necessary already at α=1. The main weakness is that the advertised elasticity application in §7.2 does not, as written, satisfy the hypotheses of Theorem 3.5 because the operator domains omit the stress-free boundary condition; the application is therefore not established. In addition, the existence part of Theorem 3.5 is conditional on a compatibility condition that is literally membership in Im Fα, so the substantive content is uniqueness rather than a constructive existence statement.

major comments (1)
  1. [§7.2, operator domains after Eq. (7.2)] The domains D(A) and D(B) are defined as {u∈L^2(Ω): Au∈L^2(Ω), u=0 on Γ_D} and {u∈L^2(Ω): Bu∈L^2(Ω), u=0 on Γ_D}, respectively, with no condition on Γ_N. If Γ_N is nonempty, integration by parts gives ⟨Bu,v⟩−⟨u,Bv⟩ = ∫_{Γ_N}(∂νu·v − u·∂νv)dS, which need not vanish for pairs in the stated domains; hence B is not symmetric, and the same applies to A. Consequently the displayed identities ∫Bu·u = ∥∇u∥^2 and D(B^{1/2}) = H^1_D(Ω) are false as stated, and Theorem 7.2 is not a valid application of Theorem 3.5. The authors must either include ∂νu=0 on Γ_N in both domains and verify self-adjointness (and commutativity of the resulting realizations) or restrict to the pure Dirichlet case Γ_N=∅.
minor comments (5)
  1. [Proposition 4.1(a)] The Jacobian formula in Proposition 4.1(a) contains an erroneous factor of 2. Direct differentiation of ⟨A_i e^{-tA(λ)}u0, e^{-tA(λ)}u0⟩ with t = (τ+σ)Tbar^α/2 gives −(τ+σ)Tbar^α⟨A_i e^{-...}u0, A_j e^{-...}u0⟩, not −2(τ+σ)Tbar^α times the same quantity. The sign, and hence the negative-definiteness conclusion, is unaffected, but the displayed formula should be corrected.
  2. [Section 5 and Figures 1, 3-7] The numerical examples use an L-shaped domain and a circle-with-hole domain, which are not product domains Ω1×...×Ωn as assumed in §7.1, where the operators A_i = −Δ_{x_i} are constructed on a product. The authors should state explicitly that these simulations are heuristic illustrations of the abstract map Fα rather than instances of the §7.1 framework, or adapt the framework to cover the geometries used.
  3. [Theorem 3.5] The statement of Theorem 3.5 fixes T>0 and later uses Tbar∈[T1,T2]; the relationship between the final time T and the measuring time Tbar should be clarified, since Fα depends on Tbar and T otherwise does not appear in the statement.
  4. [§7.1, domain of A_i] In §7.1 the domain D(A_i) is written only as {u∈L^2(Ω): Δ_{x_i}u∈L^2(Ω)}; the Dirichlet condition u=0 on ∂Ω should be included explicitly, since without it the operator is not the self-adjoint Dirichlet realization used in the subsequent identities and in Theorem 7.1.
  5. [Definition 3.3] Definition 3.3 defines compatibility as φ∈Im Fα, which makes the existence assertion in Theorem 3.5 true by definition. The text should state more explicitly that the substantive content of Theorem 3.5 is the injectivity/uniqueness result and that Im Fα is only partially described by the topological properties in Proposition 4.4.

Circularity Check

2 steps flagged · score 2.0 of 10

Central uniqueness proof is self-contained; only the conditional existence statement is definitional (compatibility = membership in Im F_alpha).

  1. self definitional [Section 3.1, Definition 3.3 and Theorem 3.5; Section 4.2]
    "Definition 3.3. For any fixed admissible u0 ∈ A, we define an additional measurement φ ∈ Rn+ to be compatible if and only if φ ∈ Im F α. ... Theorem 3.5. Fixed T > 0, for any admissible initial datum u0 ∈ A and all compatible additional measurements φ ∈ Im F α, there exists a unique solution (uα, λα) to Problem Pα. ... As previously mentioned, the existence of a solution to the nonlinear equation (3.3) is, in fact, equivalent to the compatibility condition displayed in Definition 3.3."

    The existence half of Theorem 3.5 is true by construction: 'compatible' is defined as φ belonging to the range of the forward map Fα, so choosing λ with Fα(λ)=φ and setting uα(t)=Sα,λ(t)u0 yields a solution immediately. The paper itself states that existence is equivalent to this compatibility condition. The uniqueness claim is not circular: it is proved independently via the negative definiteness of Fα' in Proposition 4.1 and the Gale-Nikaido theorem. Thus this is a minor definitional caveat about the conditional existence result, not a circular derivation of the central uniqueness theorem.

  2. other [Section 7.2, domains D(A) and D(B)]
    "Au = −∇(∇·u) ... on the domain D(A) = {u ∈ L2(Ω); Au ∈ L2(Ω) and u ≡ 0 on ΓD} and Bu = −∆u ... on the domain D(B) = {u ∈ L2(Ω); Bu ∈ L2(Ω) and u ≡ 0 on ΓD}."

    This is a correctness gap in an advertised application rather than a circular step: the stated operator domains omit the stress-free condition ∂νu = 0 on ΓN that was imposed in the Inverse Lamé Problem, so the claimed self-adjointness of A and B and the identities ∫Ω Bu·u dxdy = ∫Ω |∇u|² dxdy are not justified when ΓN is nonempty. This affects Theorem 7.2 as an application, but it does not make the abstract derivation of Theorem 3.5 circular.

full rationale

The paper's central abstract results are not circular. The map F_alpha in (3.3) is derived directly from the semigroup representation of the direct problem, and the uniqueness proof uses only that F_alpha' is negative definite on R^n_+ (Proposition 4.1) followed by the Gale-Nikaido theorem (Theorem 4.2). The admissibility and compatibility assumptions in Definitions 3.2 and 3.3 are explicit qualifications on the data, not hidden ways of assuming the conclusion. The existence statement is tautological (compatible means φ ∈ Im F_alpha), which reduces the conditional existence to a definition; however, the uniqueness claim, which is the main advertised theorem, is proved from stated assumptions and does not rely on fitting or on self-citation. The numerical simulations are illustrative and do not feed back into the theorem. The elasticity application in §7.2 has a domain-definition gap that makes the application unproven, but that is a correctness issue, not circularity. Hence the score is 2, reflecting the minor definitional nature of the existence statement and the separate application gap.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The central result relies on standard tools (semigroup theory, Wright-function properties, Gale-Nikaido univalence) plus the explicit hypotheses that the operators commute, the initial datum is admissible, and the measurement vector is compatible. There are no free parameters fitted to data, and no new theoretical entities are introduced.

assumptions (6)
  • standard math Gale-Nikaido univalence theorem: if F in C^1(C; R^n) has positive or negative definite Jacobian on a convex open set C, then F is injective.
    Used in Theorem 4.2 to conclude injectivity of F_alpha from negative definiteness of F'_alpha; cited from [15] without proof.
  • standard math Properties of the Wright-type function Phi_alpha: positivity, integral equal to 1, finite moments integral tau^p Phi_alpha(tau) dtau = Gamma(1+p)/Gamma(1+alpha p).
    Used to define F_alpha and to justify integrability and boundedness of the map; cited from [3] and [16].
  • standard math Semigroup theory for nonnegative self-adjoint operators: -A generates an analytic contraction semigroup e^{-tA}; e^{-tA} is injective, self-adjoint, and A commutes with e^{-tA}.
    Used throughout; Proposition 2.2, standard from [4] and [21].
  • domain assumption The operators A_i are closed, self-adjoint, strictly positive (possibly unbounded), and mutually commuting on H.
    Section 3, start of the abstract framework of Problem P_alpha: these properties are assumed for the whole setting.
  • domain assumption Initial datum u0 is admissible: the vectors A1u0, ..., Anu0 are linearly independent in H.
    Definition 3.2; essential for the positive definiteness of the Gram matrix in Lemma 3.4 and Proposition 4.1(b).
  • domain assumption The measurement vector phi is compatible, i.e., phi belongs to Im F_alpha.
    Definition 3.3; the existence part of the theorem relies on this condition.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Identification problems for anisotropic time-fractional subdiffusion equations." pith.science (2026). https://pith.science/paper/SUNYBSVB

@misc{pith2026250710315,
  author       = {Pith},
  title        = {Pith review of: Identification problems for anisotropic time-fractional subdiffusion equations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SUNYBSVB}},
  note         = {Machine review of arXiv:2507.10315}
}
abstract

We investigate the inverse problem consisting in the identification of constant coefficients for a fractional-in-time partial differential equation governed by a finite sum of positive self-adjoint operators on a Hilbert space under energy-type overdeterminating conditions. We prove the uniqueness of the solution to the inverse problem when the fractional order $\alpha$ of the derivative is in $(0,1)$. A conditioned existence result is also provided, complemented with a suitable selection of numerical calculations. In addition, we prove that, as $\alpha\to 1^{-}$, the solution corresponding to $\alpha$ tends to the classical one ($\alpha=1$). Applications to examples of heat diffusion and elasticity are presented.

Figures

Figures reproduced from arXiv: 2507.10315 by the authors.

Figure 1
Figure 1. A sketch of the computational domains used for the forward problem simulations. The top row plots depict the L-shaped domain, while the bottom plots are referred to the circle with eccentric hole. Each plot also features the contour plot of the initial condition function pre￾scribed: a) symmetrically centered C∞ cutoff function; b) two asymmet￾rically centered C∞ cutoff functions of different radii; c) symmetrically… view at source ↗
Figure 2
Figure 2. Diffusion parameters. All of the 1200 initial-boundary value problems considered have been solved in the time interval t ∈ [0, 0.04], and at the end of such interval, the L 2 norms of the solution derivatives ||ux||2 and ||uy||2 have been computed within the framework of a post pro￾cessing step. In the diagrams presented in Figs. 4, 5, 6 and 7 (see Appendix A), for each configuration of Ωi and ϕ Ωi j , i, j = 1, 2 c… view at source ↗
Figure 3
Figure 3. Solution derivatives norms ||ux||2 vs ||uy||2 obtained using the diffusion parameters in [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Solution derivatives norms ||ux||2 vs ||uy||2 obtained using the diffusion parameters in [PITH_FULL_IMAGE:figures/full_fig_p020_4.png]
Figure 5
Figure 5. Figure 5: Solution derivatives norms ||ux||2 vs ||uy||2 obtained using the diffusion parameters in [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]
Figure 6
Figure 6. Figure 6: Solution derivatives norms ||ux||2 vs ||uy||2 obtained using the diffusion parameters in [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: Solution derivatives norms ||ux||2 vs ||uy||2 obtained using the diffusion parameters in [PITH_FULL_IMAGE:figures/full_fig_p022_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 30 canonical work pages

  1. [23]

    Imaging 12 527–543

    Mola G 2018 Recovering a large number of diffusion constants in a parabolic equation from energy measurements Inverse Probl. Imaging 12 527–543

    work page 2018

  2. [1]

    Africa P, Arndt D, Bangerth W, Blais B, Fehling M, Gassm¨ oller R, Heister T, Heltai L, Kinnewig S, Kronbichler M, Maier M, Munch P, Schreter-Fleischhacker M, Thiele J, Turcksin B, Wells D and Yushutin V 2024 Volume 32 7 369–380 (doi: 10.1515/jnma-2024-0137)

    doi:10.1515/jnma-2024-0137 2024

  3. [2]

    Akamatsu M, Nakamura G and Steinberg S 1991 Identification of Lam´ e coefficients from boundary observations Inverse Problems 7 335–354

    work page 1991

  4. [3]

    Bazhlekova E G 2001 Fractional evolution equations in Banach spaces PhD Thesis Technische Uni- versiteit Eindhoven, Eindhoven

    work page 2001

  5. [4]

    Brezis H 1983 Analyse Fonctionnelle, Th´ eorie et Applications(Masson, Paris); English translation: 2011 Functional Analysis, Sobolev Spaces and Partial Differential Equations (Universitext, Springer, New York)

    work page 1983

  6. [5]

    II Fract

    Caputo M 2008 Linear models of dissipation whose Q is almost frequency independent. II Fract. Calc. Appl. Anal. 11 4–14. Reprinted from 1967 Geophys. J. R. Astr. Soc. 13 529–539

    work page 2008

  7. [6]

    Cheng J, Nakagawa J, Yamamoto M and Yamazaki T 2009 Uniqueness in an inverse problem for a one-dimensional fractional diffusion equation Inverse Problems 25 115002

    work page 2009

  8. [7]

    2nd Edition (Oxford University Press, London)

    Crank J 1975 The Mathematics of Diffusion. 2nd Edition (Oxford University Press, London)

    work page 1975

Show all 31 references
  1. [8]

    Creo S, Lancia M R, Mola G and Romanelli S 2025 Inverse problems in irregular domains: Approx- imation via Mosco convergence Contemp. Math. accepted

    2025

  2. [9]

    de Carvalho Neto P M 2013 Fractional differential equations: a novel study of local and global solutions in Banach spaces PhD Thesis Universidade de Sao Paulo, Sao Paulo

    2013

  3. [10]

    An Application-Oriented Expo- sition Using Differential Operators of Caputo Type (Springer-Verlag, Berlin)

    Diethelm K 2010 The Analysis of Fractional Differential Equations. An Application-Oriented Expo- sition Using Differential Operators of Caputo Type (Springer-Verlag, Berlin)

    2010

  4. [11]

    Durdiev D K, Jumaev J J and Turdiev H H 2025 Inverse problem for determining time dependent coefficient and source functions in a time-fractional diffusion equation J. Math. Sci. 289 475–486 (doi: 10.1007/s10958-024-07204-y)

    2025 doi

  5. [12]

    Favini A, Mola G and Romanelli S 2022 Recovering time-dependent diffusion coefficients in a nonau- tonomous parabolic equation from energy measurements Discrete Cont. Dyn. Syst. Ser. S 6 1439– 1454

    2022

  6. [13]

    Floridia G, Golgeleyen F and Yamamoto M 2025 Initial boundary value problems for time-fractional evolution equations in Banach spaces arXiv:2502.06554

    2025 arXiv

  7. [14]

    Gal C G and Warma M 2020 Fractional-in-Time Semilinear Parabolic Equations and Applications (Math´ ematiques et Applications, Springer)

    2020

  8. [15]

    Annalen 159 81–93

    Gale D and Nikaido H 1965 The Jacobian matrix and global univalence of mappings Math. Annalen 159 81–93

    1965

  9. [16]

    Gorenflo R, Luchko Y and Mainardi F 1999 Analytical properties and applications of the Wright function Fract. Calc. Appl. Anal. 2 383–414

    1999

  10. [17]

    Second edition (Dover Books in Chemistry, Dover)

    Ghez R 2018 Diffusion Phenomena: Cases and Studies. Second edition (Dover Books in Chemistry, Dover)

    2018

  11. [18]

    Porous Med

    Hamada M and de Anna P 2023 A method to measure the diffusion coefficient in liquids Transp. Porous Med. 146 463—474

    2023

  12. [19]

    Jin B, Lu X, Quan Q and Zhou Z 2024 Numerical recovery of the diffusion coefficient in diffusion equations from terminal measurement arXiv:2405.10708

    2024 arXiv

  13. [20]

    Jing X, Jia J and Song X 2025 Simultaneous uniqueness identification of the fractional order and diffusion coefficient in a time–fractional diffusion equation Appl. Math. Lett. 162 109386

    2025

  14. [21]

    Kato T 1966 Perturbation Theory for Linear Operators (Springer–Verlag, Berlin and New York)

    1966

  15. [22]

    2020 Time-Fractional Differential Equations – A Theo- retical Introduction (Springer, Singapore)

    Kubica A, Ryszewska K and Yamamoto M. 2020 Time-Fractional Differential Equations – A Theo- retical Introduction (Springer, Singapore)

    2020

  16. [24]

    Mola G, Okazawa N and Yokota T 2016 Reconstruction of two constant coefficients in linear anisotropic diffusion model Inverse Problems 32 115016

    2016

  17. [25]

    Nakamura G and Uhlmann G 1993 Identification of Lam´ e parameters by boundary measurements Amer. J. Math. 5 1161–1187

    1993

  18. [26]

    Oldham K and Spanier J 1974 The Fractional Calculus : Theory and Applications of Differentiation and Integration to Arbitrary Order (Academic Press, New York)

    1974

  19. [27]

    Qiao Y, Xiong X and Han J 2025 A variational approach to recover the unknown source and initial condition for a time–space fractional diffusion equation J. Appl. Math. Comput. 71 3445–3476 20

    2025

  20. [28]

    Quarteroni A and Valli A 1994 Numerical Approximation of Partial Differential Equations (Springer- Verlag, Berlin)

    1994

  21. [29]

    Serikbaev D, Ruzhansky M and Tokmagambetov N 2023 Inverse problem of determining time– dependent leading coefficient in the time–fractional heat equation arXiv:2306.03545

    2023 arXiv

  22. [30]

    Tanabe H 1979 Equations of Evolution (Monographs and Studies in Math., Pitman, London)

    1979

  23. [31]

    Wright E M 1940 The generalized Bessel function of order greater than one Q. J. Math. 11 36–48 Appendix A. Numerical simulations: additional discussion In the present section, we provide more specific comments to the plots presented in Fig. 3 (see Section 5). In all plots prod...

    1940

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.