Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Structural identifiability of linear-in-parameter parabolic PDEs through auxiliary elliptic operators

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper proves that for fully observed parabolic PDEs linear in their parameters, structural identifiability reduces to the kernel of an auxiliary elliptic operator, and that logistic-type nonlinearities make the model globally…

desk verdict A useful reduction of parabolic identifiability to elliptic spectral theory, with a genuine but localized gap in the flagship logistic example. read the letter →

arxiv 2411.17553 v2 pith:BBVAJPC3 submitted 2024-11-26 math.AP q-bio.QM

Yurij Salmaniw , Alexander P Browning This is my paper · ORCID
classification math.APq-bio.QM MSC 35K5535J2535R3093B30
keywords structuralidentifiabilitypartialdifferentialequationslinear-in-parametermodelsauxiliaryellipticoperatorFredholmalternativereaction-diffusionpracticallogisticgrowth
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

The paper tries to establish that for parabolic PDEs that are linear in their parameters, the question of structural identifiability—whether two different parameter sets can produce the same observed solution—is decided by an auxiliary elliptic operator obtained by subtracting the two equations. For linear equations, identifiability from a given initial condition holds unless the evolving solution lies in the kernel of that operator at some time; the Fredholm alternative then makes the set of problematic initial conditions finite-dimensional and rare. For nonlinear reactions such as logistic growth, uniqueness of solutions to the auxiliary elliptic equation implies unconditional global identifiability as long as the observed solution changes in time. This matters because spatial biological data are increasingly common, while identifiability theory for PDEs is underdeveloped; the paper's reduction turns an infinite-dimensional inverse problem into classical spectral theory and gives explicit parameter combinations that can never be separated.

What carries the argument

The load-bearing object is the auxiliary elliptic operator $L[A] = d\Delta + b\cdot\nabla + c$ for the linear case, and the auxiliary elliptic equation $-d\Delta\psi = f(x,\psi;B)$ for the nonlinear case. Subtracting the two PDEs that the same solution would satisfy eliminates the time derivative and leaves a time-independent equation that the solution must solve at every instant; therefore, if the solution ever leaves the kernel (or, in the nonlinear case, if the elliptic problem has a unique or discrete solution set), the two parameter points are forced to coincide. The Fredholm alternative is used to show that for elliptic $L[A]$ the kernel is finite-dimensional, so non-identifiability is confined to a thin set of initial conditions, while uniqueness theorems for the logistic elliptic problem carry the unconditional identifiability result.

What would settle it

Take the logistic model with $b_1 = b_2$, choose any Dirichlet eigenpair $(\lambda_1, \sin(\pi x))$ on $(0,1)$, and pick $(a_1,d_1) \neq (a_2,d_2)$ with $a_1 - d_1\lambda_1 = a_2 - d_2\lambda_1$; starting from $u_0 = \sin(\pi x)$, both parameter sets produce the identical non-stationary solution $u(x,t) = e^{(a_1 - d_1\lambda_1)t}\sin(\pi x)$, which would contradict the paper's unconditional identifiability claim for the logistic model unless $b=0$ is excluded.

Watch

Extended reading notes

Core claim

The central claim is Theorem 3.7: for the homogeneous fully observed model $u_t = d\Delta u + b\cdot\nabla u + cu$ with Dirichlet, Neumann, Robin, or periodic boundary conditions, two parameter points $A_1$ and $A_2$ are distinguishable unless the difference $A = A_1 - A_2$ belongs to the set $\mathcal{A}$ where the auxiliary elliptic operator $L[A] = d\Delta + b\cdot\nabla + c$ has a nontrivial kernel. The model is unconditionally identifiable on the quotient set $\mathcal{R}$, identifiable from an initial condition $u_0$ whenever $u_0$ avoids $\ker(L[A])$ for every $A \in \mathcal{A}$, and never unconditionally identifiable over the whole parameter space because some initial condition always lands in a kernel. For the logistic reaction-diffusion model $u_t - d\Delta u = au - bu^2$, Theorem 4.1 establishes global unconditional identifiability whenever the solution is nontrivial and $u_t \not\equiv 0$, because the auxiliary elliptic equation $-d\Delta\psi = a\psi - b\psi^2$ has only discrete (in fact unique) nontrivial solutions for the parameter differences considered. The paper works out the precise non-identifiable parameter sets for Dirichlet, Neumann, Robin, and periodic conditions, and shows numerically that initial conditions close to the dominant eigenfunction make parameters practically unidentifiable even when structural identifiability holds.

Load-bearing premise

The strongest results rely on the observed solution being regular enough that the auxiliary elliptic operator can be applied at $t=0$, and on the auxiliary elliptic equation having at most finitely many solutions for every parameter difference; in the logistic case the unexamined $b=0$ limit makes the equation linear, where solutions can form a continuum, and that single case would break the claim of global unconditional identifiability.

Editorial extensions

If this is right

  • For the linear reaction-diffusion equation with Dirichlet conditions, non-identifiable parameter pairs are exactly the curves $c = d(n\pi/\ell)^2$ (and their drift-generalized surfaces $c = \lambda_n(d,b)$); the only structurally non-identifiable initial condition is an eigenfunction, and the parameter combination $c - d\pi^2$ is identifiable even when $c$ and $d$ individually are not.
  • Any initial condition that is not an element of the kernel of the auxiliary elliptic operator restores structural identifiability for the homogeneous linear model, so non-identifiability is a finite-dimensional, measure-zero phenomenon.
  • For the logistic model, no nontrivial time-varying solution can be generated by two distinct parameter sets, so the model is globally unconditionally identifiable under any of the four boundary conditions considered.
  • In full parameter space the linear model is always identifiable in the weak sense that some initial condition separates any two parameters, yet always fails unconditional identifiability because some initial condition falls in the kernel.
  • Experiments initialized near the dominant eigenfunction can be practically non-identifiable: the 95% confidence region for $(c,d)$ follows the theoretically indistinguishable set and can span several orders of magnitude in $d$.

Reading between the lines

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

  • Beyond the paper, the same subtraction argument gives a template for any linear-in-parameter parabolic model: write the parameter difference as an auxiliary operator and compute its kernel; this could be automated for polynomial reaction terms with symbolic linear algebra and spectral solvers.
  • The paper's Example 4.1 silently passes over $b = 0$ in the logistic auxiliary equation; since the equation then becomes linear and its solution set is a vector space whenever $a/d$ is an eigenvalue, the claimed global unconditional identifiability needs either an explicit exclusion of $b=0$ or a separate argument.
  • The practical-identifiability analysis suggests a design rule the authors do not state: the closer an experiment's initial condition is to the dominant eigenfunction of the diffusion operator, the wider the confidence region in the diffusion-growth plane, so spatial initial conditions should be chosen to maximize projection on higher modes.
  • For boundary conditions with repeated eigenvalues, such as periodic conditions on a symmetric domain, the kernel is multidimensional, so more than one initial condition is needed to break non-identifiability; a similar counting rule should apply to any PDE model whose elliptic operator has eigenvalues of multiplicity greater than one.
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

3 major / 5 minor

Summary. The paper develops a structural identifiability framework for fully observed parabolic PDEs that are linear in their parameters. The central idea is to reduce indistinguishability of two parameter points to the existence of nontrivial solutions of an auxiliary elliptic equation obtained by subtracting the two PDEs. For linear homogeneous equations, the kernel of the operator L[A1-A2] characterizes potentially indistinguishable initial conditions, and Fredholm theory is used to show that this kernel is finite-dimensional for nondegenerate elliptic operators. Explicit non-identifiable separable solutions are constructed for Dirichlet, Neumann, Robin, and periodic boundary conditions, and the authors show that initial conditions close to a dominant eigenfunction lead to practical non-identifiability in a profile-likelihood study. For nonlinear reaction terms that are linear in parameters, Theorem 4.1 claims unconditional identifiability when the auxiliary elliptic equation has a unique or discrete solution set, and Example 4.1 applies this to logistic growth to conclude global unconditional identifiability. Section 5 presents numerical experiments with reproducible code illustrating how structural non-identifiability affects parameter inference.

Significance. The linear theory in Sections 2 and 3 is a genuine and useful contribution: it connects PDE identifiability to spectral properties of an auxiliary elliptic operator, yields explicit non-identifiable solution constructions, and gives a clean sufficient condition for identifiability from a given initial condition. The numerical demonstration in Section 5 is also valuable and is supported by publicly available code. However, the flagship nonlinear result--global unconditional identifiability of the logistic model--is not established by the arguments as written. The gap is localized and likely fixable, but it currently affects the paper's main advertised nonlinear claim. If repaired, the paper would be a strong contribution to the emerging theory of PDE structural identifiability.

major comments (3)
  1. [Section 4.1, Example 4.1] The claim that for a in R and b > 0 'any nontrivial solution is unique' is not supported by the cited references, which concern positive solutions. For the one-dimensional Dirichlet problem -psi'' = a psi - psi^2, sign-changing solutions bifurcate from each eigenvalue of -d^2/dx^2; for instance, for a > 4 pi^2 on (0, ell) there exist both a positive solution and sign-changing solutions. Thus the set of all nontrivial solutions is neither a singleton nor discrete, and the conclusion ANI = empty is not established. The symmetry argument for b < 0 merely maps sign-changing solutions to sign-changing solutions and does not remove them. This is load-bearing because Theorem 4.1's 'global unconditional identifiability' for the logistic model depends on ANI being empty.
  2. [Example 4.1 and Eq. (4.3)] The case b = 0 is never treated. When b_1 = b_2 (so b = 0), the auxiliary equation is the linear equation -d Delta psi = a psi; if a/d is an eigenvalue of the Laplacian with the chosen boundary conditions, the solution set is a vector space, i.e., a continuum. This falls squarely within the paper's own definition of ANI in (4.3). This parameter difference is admissible even when both quadratic coefficients are positive. The authors must either prove that solutions from this continuum cannot satisfy the original parabolic problem (e.g., by showing that the quadratic term prevents the time-dependent separation-of-variables ansatz when b_1 = b_2 > 0) or restrict the statement of Theorem 4.1 and Example 4.1 accordingly.
  3. [Theorem 4.1 and Definition 3.5] Theorem 4.1 assumes u_t not identically zero but concludes 'unconditionally identifiable,' which is defined in Definition 3.5 as a property for all nontrivial initial conditions. Initial conditions that are steady states give u_t identically zero and are not covered by the theorem's hypothesis; moreover, such initial conditions can genuinely be indistinguishable. For example, with Neumann boundary conditions and logistic nonlinearity, every constant K > 0 is a solution for any parameter triple (d, a, b) with a/b = K, so the model is not unconditionally identifiable in the sense of Definition 3.5. The theorem and the concluding sentence of Example 4.1 should be restated as identifiability from initial conditions whose solution has nonconstant time dependence, or a separate definition of identifiability within the class of nonstationary solutions should be introduced.
minor comments (5)
  1. [Definitions (4.2)-(4.3)] Because f(x, 0; B) = 0, the zero function is always a solution of the auxiliary elliptic equation; the definition of R as 'has a unique solution' should specify whether the trivial solution is included, since otherwise R excludes every case with a nonzero solution and the intended meaning of 'unique nontrivial solution' is lost.
  2. [Proof of Theorem 4.1] The inference from 'the solution set of the auxiliary elliptic equation is discrete' to 'u(x, t) = psi(x) for some fixed psi for all t' should be justified explicitly: it uses continuity of t maps to u(., t), the connectedness of (0, T), and the fact that a connected subset of a discrete set is a singleton.
  3. [Example 4.1 and Section 4] Example 4.1 refers to 'model (3.7)' where model (4.1) is meant, and its concluding sentence attributes 'unconditionally identifiable' to Definition 3.4, although Definition 3.4 is identifiability from a fixed initial condition and Definition 3.5 is unconditional identifiability.
  4. [Theorem 3.7(b)] The 'in particular' statement that identifiability follows from u0 not in ker(L[A]) for all A in A requires passing to the limit L[A]u(., t) -> L[A]u0 as t -> 0; this needs u0 to belong to the domain of L[A] (e.g., C^2 or H^2 satisfying the boundary condition), not merely the stated compatibility condition Bu0 = 0.
  5. [Section 5 and Figure 4] For the bivariate profile likelihood over (c, d), the 95% likelihood-ratio threshold should be approximately 5.991 (chi-square with 2 degrees of freedom), not 2.997, which is the one-degree-of-freedom threshold; please correct the threshold and check whether the displayed confidence regions change materially.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: identifiability is reduced to independent elliptic theory; the flagged gap in Example 4.1 is a proof gap, not a circular step.

full rationale

The derivation chain is self-contained. The paper reduces parameter distinguishability to the auxiliary elliptic equation by subtracting the two PDEs solved by a common solution, which is a genuine reduction rather than a definitional identification; the subsequent identifiability results rest on the Fredholm alternative, Sturm–Liouville spectral theory, and external elliptic uniqueness results (Pao, Zhao, Ni) rather than on fitted parameters or the authors' own prior conclusions. The citation to the authors' earlier work [20] is contextual and not load-bearing. The definitions of R, AI, and ANI do encode the classification, but Theorem 3.7 and Theorem 4.1 prove the contrapositive from the auxiliary equation, so this is a legitimate framework rather than circularity. One genuine limitation appears in Example 4.1 (Section 4.1): for d > 0 and b = 0, the auxiliary equation is linear, -d Delta psi = a psi, and if a/d is an eigenvalue the solution set is a continuum, placing (d, a, 0) in the paper's own ANI set (4.3); the paper treats only b > 0 and b < 0 before asserting 'In particular, ANI = empty', so the claim of global unconditional identifiability for the logistic model is not fully established. This is a correctness or completeness gap in an example, not a circular step, because the conclusion would not be true merely by definition and the missing case could still be identifiable if the linear continuum fails to generate time-dependent indistinguishable solutions.

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

No free parameters are introduced; the analysis is purely structural. The axioms are standard spectral-theoretic facts, stated regularity assumptions on the domain and solutions, and literature results on uniqueness of positive solutions to the logistic elliptic equation. The auxiliary elliptic operator is a mathematical framing device, not a new physical entity.

assumptions (6)
  • domain assumption The spatial domain Ω is bounded and has a C^2 boundary.
    Stated in Section 3.1 to guarantee classical regularity and standard spectral theory; the authors note it can be weakened.
  • standard math For d>0, the elliptic operator L[A] is a Fredholm operator with discrete spectrum, so the Fredholm alternative applies and dim ker(L[A]) is finite.
    Invoked in Proposition 3.9 and Remark 3.10 to characterize A via eigenvalues of the associated eigenvalue problem (3.8).
  • domain assumption The solution u is smooth enough that t → L[A]u(·,t) is continuous up to t=0, so that u0 ∈ ker(L[A]) follows from u(·,t) ∈ ker(L[A]) for t>0.
    Used in Theorem 3.7(b); requires more initial-data regularity than the stated compatibility condition Bu0=0.
  • domain assumption For the logistic elliptic equation -dΔψ = aψ - bψ² with b>0, any nontrivial solution is unique (and by symmetry the same holds for b<0).
    Used in Example 4.1 to conclude R ∪ AI covers b≠0 and to claim ANI=∅; cited to Pao [33] and Zhao [34, Ch. 2].
  • domain assumption The initial condition is 'compatible', meaning it satisfies the boundary condition and any physical constraints such as nonnegativity.
    The notion of 'compatible' initial conditions in Proposition 3.6(iii) is left intentionally vague; it restricts the set of admissible u0.
  • domain assumption The boundary operator B is parameter-independent and does not contain unknown parameters.
    Stated in Section 1 as a simplifying assumption; parameter-dependent boundary conditions are left to future work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Structural identifiability of linear-in-parameter parabolic PDEs through auxiliary elliptic operators." pith.science (2026). https://pith.science/paper/BBVAJPC3

@misc{pith2026241117553,
  author       = {Pith},
  title        = {Pith review of: Structural identifiability of linear-in-parameter parabolic PDEs through auxiliary elliptic operators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BBVAJPC3}},
  note         = {Machine review of arXiv:2411.17553}
}
read the original abstract

Parameter identifiability is often requisite to the effective application of mathematical models in the interpretation of biological data, however theory applicable to the study of partial differential equations remains limited. We present a new approach to structural identifiability analysis of fully observed parabolic equations that are linear in their parameters. Our approach frames identifiability as an existence and uniqueness problem in a closely related elliptic equation and draws, for homogeneous equations, on the well-known Fredholm alternative to establish unconditional identifiability, and cases where specific choices of initial and boundary conditions lead to non-identifiability. While in some sense pathological, we demonstrate that this loss of structural identifiability has ramifications for practical identifiability; important particularly for spatial problems, where the initial condition is often limited by experimental constraints. For cases with nonlinear reaction terms, uniqueness of solutions to the auxiliary elliptic equation corresponds to identifiability, often leading to unconditional global identifiability under mild assumptions. We present analysis for a suite of simple scalar models with various boundary conditions that include linear (exponential) and nonlinear (logistic) source terms, and a special case of a two-species cell motility model. We conclude by discussing how this new perspective enables well-developed analysis tools to advance the developing theory underlying structural identifiability of partial differential equations.

Figures

Figures reproduced from arXiv: 2411.17553 by the authors.

Figure 1
Figure 1. Structural non-identifiability in a (a) ordinary- and (b) partial differential equation. Shown are indistinguishable solutions from two distinct parameter sets. In (a) we show the first component of X subject to X˙ = M1X (red solid) and X˙ = M2X (black dashed) where M1 = ( 1 0 0 1 ), M2 = ( 2 −1 1 0 ), and X(0) = (1, 1)⊺ . In (b), we show the solution a linear reaction-diffusion equation (given by Eq. (1.2)) at t ∈ … view at source ↗
Figure 2
Figure 2. A flow chart diagram outlining Steps 1-3 described in Section 3.1. 3.2 Linear, homogeneous parabolic equations For convenience, we will denote by L = L[A] the elliptic operator L[A] := d∆ + b · ∇ + c (3.5) for A ∈ [0, ∞) × R n+1. For all d > 0, L[A] is a uniformly elliptic operator; when d = 0, the operator L degenerates and we write L0[A] := b · ∇ + c. (3.6) We also write L1[A] := d∆ + b · ∇ to denote the operator … view at source ↗
Figure 3
Figure 3. A depiction of the set A for the Dirichlet problem without drift (a) and with drift (b). note that for any d > 0 the problem    −d ∂ 2ϕ ∂x2 − b ∂ϕ ∂x = λϕ, for x ∈ (0, ℓ) ϕ = 0, for x = 0, ℓ also enjoys a discrete set of eigenvalue/eigenfunction pairs of the form (ϕn, λn(d, b)) =  e −bx/2d sinp λnx  , b 2 4d + d n 2π 2 ℓ 2  , from which one obtains the no-drift case by taking b = 0. Unlike the b = 0 case, how… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Practical identifiability of parameters in the linear reaction-diffusion equation subject to Dirichlet boundary conditions. (a–c) Initial conditions used to produce synthetic data. Initial conditions are based on a Gaussian centred at x = 1/2 with standard deviations ω…

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. Structural Identifiability of Compartmental Models: Recent Progress and Future Directions

    stat.ME 2025-07 accept

    A survey of recent theory and applications of structural identifiability in compartmental models, including identifiable reparametrizations and graph-based criteria.

Reference graph

Works this paper leans on

41 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [1]

    Liepe J, Kirk P, Filippi S, Toni T, Barnes CP, Stumpf MPH. 2014. A framework for parameter estimation and model selection from experimental data in systems biology using approximate Bayesian computation. Nature Protocols 9:439 456. (doi:10.1038/nprot.2014.025)

    doi:10.1038/nprot.2014.025 2014

  2. [2]

    G´ abor A, Banga JR. 2015. Robust and efficient parameter estimation in dynamic models of biological systems. BMC Systems Biology 9:74. (doi:10.1186/s12918-015-0219-2)

    work page doi:10.1186/s12918-015-0219-2 2015

  3. [3]

    Cobelli C, DiStefano JJ. 1980. Parameter and structural identifiability concepts and ambiguities: a critical review and analysis. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 239:7—24. (doi:10.1152/ajpregu.1980.239.1.r7)

    doi:10.1152/ajpregu.1980.239.1.r7 1980

  4. [4]

    Raue A, Kreutz C, Maiwald T, Bachmann J, Schilling M, Klingm¨ uller U, Timmer J. 2009. Structural and practical identifiability analysis of partially observed dynamical models by exploiting the profile likelihood. Bioinformatics 25:1923–1929. (doi:10.1093/bioinformatics/btp358)

    doi:10.1093/bioinformatics/btp358 2009

  5. [5]

    Chi¸ s OT, Banga JR, Balsa-Canto E. 2011. Structural identifiability of systems biology models: a critical comparison of methods. PLOS One 6:e27755. (doi:10.1371/journal.pone.0027755)

    doi:10.1371/journal.pone.0027755 2011

  6. [6]

    Bellman R, ˚Astr¨ om KJ. 1970. On structural identifiability. Mathematical Biosciences 7:329–339. (doi:10.1016/0025-5564(70)90132-x)

    doi:10.1016/0025-5564(70)90132-x 1970

  7. [7]

    Walter E, Lecourtier Y. 1981. Unidentifiable compartmental models: what to do? Mathematical Biosciences 56:1–25. (doi:10.1016/0025-5564(81)90025-0)

    work page doi:10.1016/0025-5564(81)90025-0 1981

  8. [8]

    Walter E. 1987. Identifiability of Parametric Models . Elsevier Science & Technology, London, United Kingdom. (doi:10.1016/C2013-0-03836-4)

    work page doi:10.1016/c2013-0-03836-4 1987

Show all 41 references
  1. [9]

    Renardy M, Kirschner D, Eisenberg M. 2022. Structural identifiability analysis of age-structured PDE epidemic models. Journal of Mathematical Biology 84:9. (doi:10.1007/s00285-021-01711-1)

    2022 doi

  2. [10]

    Xia X, Moog C. 2003. Identifiability of nonlinear systems with application to HIV/AIDS models. IEEE Transactions on Automatic Control 48:330–336. (doi:10.1109/tac.2002.808494)

    2003

  3. [11]

    Bellu G, Saccomani MP, Audoly S, D’Angi` o L. 2007. DAISY: A new software tool to test global identifia- bility of biological and physiological systems. Computer Methods and Programs in Biomedicine 88:52–61. (doi:10.1016/j.cmpb.2007.07.002)

    2007 doi

  4. [12]

    Barreiro XR, Villaverde AF. 2023. Benchmarking tools for a priori identifiability analysis. Bioinformatics 39:btad065. (doi:10.1093/bioinformatics/btad065)

    2023 doi

  5. [13]

    Ljung L, Glad T. 1994. On global identifiability for arbitrary model parametrizations. Automatica 30:265–

    1994

  6. [14]

    Margaria G, Riccomagno E, Chappell MJ, Wynn HP. 2001. Differential algebra methods for the study of the structural identifiability of rational function state-space models in the biosciences. Mathematical Biosciences 174:1–26. (doi:10.1016/s0025-5564(01)00079-7)

    2001 doi

  7. [15]

    Meshkat N, Kuo CEz, DiStefano J. 2014. On finding and using identifiable parameter combinations in non- linear dynamic systems biology models and COMBOS: a novel web implementation. PLOS One 9:e110261. (doi:10.1371/journal.pone.0110261)

    2014 doi

  8. [16]

    Ligon TS, Fr¨ ohlich F, Chi¸ s OT, Banga JR, Balsa-Canto E, Hasenauer J. 2017. GenSSI 2.0: multi-experiment structural identifiability analysis of SBML models. Bioinformatics 34:1421–1423. (doi:10.1093/bioinformatics/btx735). 28

    2017 doi

  9. [17]

    Vajda S, Godfrey KR, Rabitz H. 1989. Similarity transformation approach to identifiability analysis of nonlinear compartmental models. Mathematical Biosciences 93:217–248. (doi:10.1016/0025-5564(89)90024- 2)

    1989 doi

  10. [18]

    Saccomani MP, Audoly S, D’Angi` o L. 2003. Parameter identifiability of nonlinear systems: the role of initial conditions. Automatica 39:619–632. (doi:10.1016/s0005-1098(02)00302-3)

    2003 doi

  11. [19]

    Ovchinnikov A, Pillay A, Pogudin G, Scanlon T. 2022. Multi-experiment parameter identifiability of ODEs and model theory. SIAM Journal on Applied Algebra and Geometry 6:339–367. (doi:10.1137/21m1389845)

    2022 doi

  12. [20]

    Browning AP, Tac M, Falc C, Baker RE. 2024. Structural identifiability analysis of linear reaction- advection-diffusion processes in mathematical biology. Proceedings of the Royal Society A 480:20230911. (doi:10.1098/rspa.2023.0911)

    2024

  13. [21]

    Ciocanel MV, Ding L, Mastromatteo L, Reichheld S, Cabral S, Mowry K, Sandstede B. 2024. parameter identifiability in PDE models of fluorescence recovery after photobleaching.Bulletin of Mathematical Biology 86:36. (doi:10.1007/s11538-024-01266-4)

    2024 doi

  14. [22]

    Byrne HM, Harrington HA, Ovchinnikov A, Pogudin G, Rahkooy H, Soto P. 2025. Algebraic identifiability of partial differential equation models. Nonlinearity 38:025022. (doi:10.1088/1361-6544/ada510)

    2025 doi

  15. [23]

    McInerney S, Carr EJ, Simpson MJ. 2019. Parameterising continuum models of heat transfer in heteroge- neous living skin using experimental data. International Journal of Heat and Mass Transfer 128:964–975. (doi:10.1016/j.ijheatmasstransfer.2018.09.054)

    2019 doi

  16. [24]

    Simpson MJ, Baker RE, Vittadello ST, Maclaren OJ. 2020. Practical parameter identifiability for spatio-temporal models of cell invasion. Journal of The Royal Society Interface 17:20200055. (doi:10.1098/rsif.2020.0055)

    2020

  17. [25]

    Murray JD. 2002. Mathematical Biology. Springer-Verlag, Berlin. 3 edition. (doi:10.1007/b98868)

    2002 doi

  18. [26]

    Maini PK, McElwain DS, Leavesley DI. 2004. Traveling wave model to interpret a wound- healing cell migration assay for human peritoneal mesothelial cells. Tissue Engineering 10:475–482. (doi:10.1089/107632704323061834)

    2004 doi

  19. [27]

    Swanson KR, Rockne RC, Claridge J, Chaplain MA, Alvord EC, Anderson AR. 2011. Quantifying the role of angiogenesis in malignant progression of gliomas: in silico modeling integrates imaging and histology. Cancer Research 71:7366–7375. (doi:10.1158/0008-5472.can-11-1399)

    2011 doi

  20. [28]

    Stanhope S, Rubin JE, Swigon D. 2014. Identifiability of linear and linear-in-parameters dynamical systems from a single trajectory. SIAM Journal on Applied Dynamical Systems 13:1792–1815

    2014

  21. [29]

    Jin W, Shah ET, Penington CJ, McCue SW, Chopin LK, Simpson MJ. 2016. Reproducibility of scratch assays is affected by the initial degree of confluence: Experiments, modelling and model selection. Journal of Theoretical Biology 390:136 145. (doi:10.1016/j.jtbi.2015.10.040)

    2016 doi

  22. [30]

    Brezis H. 2011. Functional analysis, Sobolev spaces and partial differential equations. Universitext. Springer, New York. ISBN 978-0-387-70913-0

    2011

  23. [31]

    Kato T. 1995. Perturbation theory for linear operators . Classics in Mathematics. Springer-Verlag, Berlin. ISBN 3-540-58661-X. Reprint of the 1980 edition

    1995

  24. [32]

    Deimling K. 1985. Nonlinear functional analysis . Springer-Verlag, Berlin. ISBN 3-540-13928-1. (doi:10.1007/978-3-662-00547-7). URL https://doi.org/10.1007/978-3-662-00547-7

    1985 doi

  25. [33]

    Pao CV. 1992. Nonlinear Parabolic and Elliptic Equations . Springer New York, NY. 1 edition. ISBN 978-1-4613-6323-1. (doi:10.1007/978-1-4615-3034-3). URL https://doi.org/10.1007/978-1-4615-3034-3 . springer Science+Business Media New York, Springer Book Archive

    1992 doi

  26. [34]

    Zhao XQ. 2017. Dynamical systems in population biology . CMS Books in Mathematics/Ouvrages de Math´ ematiques de la SMC. Springer, Cham. second edition. ISBN 978-3-319-56432-6; 978-3-319-56433-

    2017

  27. [35]

    Baym M, Lieberman TD, Kelsic ED, Chait R, Gross R, Yelin I, Kishony R. 2016. Spatiotemporal microbial evolution on antibiotic landscapes. Science 353:1147–1151. (doi:10.1126/science.aag0822). 29

    2016 doi

  28. [36]

    (doi:10.1007/978-3-319-56433-3)

    doi

  29. [37]

    Ni WM. 2011. The mathematics of diffusion. volume 82 of CBMS-NSF Regional Conference Series in Applied Mathematics. Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA

    2011

  30. [38]

    Jacobs J, Salmaniw Y, Lam KY, Zhai L, Wang H, Zhang B. 2024. Fundamental principles on the effect of habitat fragmentation on species with different movement rates. Conservation Biology

    2024

  31. [39]

    Pawitan Y. 2013. In all likelihood: statistical modelling and inference using likelihood . Oxford University Press, Oxford. 30

    2013

  32. [40]

    Vittadello ST, McCue SW, Gunasingh G, Haass NK, Simpson MJ. 2018. Mathematical models for cell migra- tion with real-time cell cycle dynamics. Biophysical Journal 114:1241–1253. (doi:10.1016/j.bpj.2017.12.041)

    2018 doi

  33. [276]

    (doi:10.1016/0005-1098(94)90029-9)

    doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.