Pith. sign in

REVIEW 2 major objections 6 minor 27 references

Overcoming logarithmic singularities in the Cahn-Hilliard equation with Flory-Huggins potential: An unconditionally convergent ADMM approach

T0 review · 2 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper introduces an ADMM iterative solver for the Cahn-Hilliard equation with Flory-Huggins potential and proves it converges unconditionally for any time step and mesh while keeping iterates inside $(-1,1)$.

desk verdict Solid first-order ADMM convergence proof for the logarithmic Cahn-Hilliard equation, but the second-order claims are asserted, not proved, and the numerics lack baselines. read the letter →

arxiv 2506.09361 v1 pith:FFAGPCN3 submitted 2025-06-11 math.NA cs.NA

Ruo Li , Shengtong Liang , Zhonghua Qiao This is my paper
classification math.NAcs.NA MSC 65H1065M22
keywords Cahn-HilliardequationFlory-Hugginspotentialalternatingdirectionmethodofmultipliersconvexsplittingiterativesolverlogarithmicsingularityunconditionalconvergenceboundpreservation
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 claims that the nonlinear systems produced by convex-splitting time-stepping for the Cahn-Hilliard equation with Flory-Huggins logarithmic potential can be solved by a specially designed ADMM iteration that converges no matter how large the time step or mesh is, and whose intermediate iterates never leave the physical interval $(-1,1)$. If true, this removes the two restrictions that have made existing solvers brittle: reliance on the strict separation property and on small step sizes. The authors prove the claim for the first-order convex-splitting scheme by showing that a Lyapunov functional decreases monotonically to zero, and they state the analogous theorem for the second-order scheme, with the proof deferred as similar. Numerical experiments in two- and three-dimensional coarsening simulations, including a three-dimensional run with time step $\tau=0.01$, are consistent with the unconditional-convergence claim.

What carries the argument

The load-bearing object is the ADMM iteration for the equivalent constrained minimax problem, together with the Lyapunov functional $\Psi^{(k)}=\frac{1}{\rho_u}\|e_{u_3}^{(k)}\|_2^2+\frac{1}{\rho_w}\|e_{w_3}^{(k)}\|_2^2+\rho_u\|e_{u_2}^{(k)}\|_2^2+\rho_w\|e_{w_2}^{(k)}\|_2^2$, which measures weighted squared distances to a stationary point. Lemma 4.2 shows that one ADMM sweep strictly decreases this functional by the residuals of the splitting, and Proposition 4.3 combines this monotone decrease with subsequential convergence to obtain full convergence to the unique solution. The same machinery gives bound preservation through the singular logarithm and the mass-conservation bound of Proposition 4.5.

What would settle it

Run the second-order Algorithm 2 on a small periodic two-dimensional domain (for instance $N=32$, $L=1$) with $\theta_0=3$, $\epsilon=0.01$, a deliberately large time step such as $\tau=0.05$, and random initial data in $(-0.05,0.05)$, tracking whether the natural analogue of the Lyapunov functional is nonincreasing and whether $\|u_2^{(k)}-u^{n+1}\|_2$ decreases to the set tolerance. If the monotonicity fails at any positive $\tau$ and $h$, or if the iterates stop decreasing before tolerance, the unconditional-convergence claim for the second-order scheme is refuted; the first-order claim can be tested the same way at extreme $\tau$.

Watch

Extended reading notes

Core claim

The central discovery is that the nonlinear system from the convex-splitting scheme can be rewritten as a saddle-point minimax problem, min over $u$ and max over $w$ of an energy $Z(u,w)$, and then solved by a variant of the alternating direction method of multipliers (ADMM). The splitting separates a linear component, which becomes a constant-coefficient system solvable by fast Fourier transform, from the logarithmic nonlinearity, which reduces to independent scalar equations with monotone derivatives solvable by Newton's method. Theorem 4.4 establishes $\lim_{k\to\infty}\|u_2^{(k)}-u^{n+1}\|_2=0$ for the first-order scheme with no condition on the time step $\tau$ or mesh size $h$; Theorem 4.7 states the same for the second-order scheme. Because the nonlinear subproblem retains the full logarithm $\log(1+u)-\log(1-u)$, every iterate lies in $(-1,1)$ without regularization, and the change in total mass is bounded by a quantity that vanishes with the stopping tolerance.

Load-bearing premise

The second-order convergence theorem and its mass bound are asserted with the proof omitted as similar to the first-order argument, so the claim for the second-order scheme stands on the unproven assumption that the same Lyapunov monotonicity holds for the modified energy of the second-order scheme.

Editorial extensions

If this is right

  • The nonlinear solve inside each convex-splitting step can be driven to arbitrary accuracy, so the unconditional solvability of convex splitting is no longer undermined by solver restrictions.
  • Users can take large time steps without a convergence condition from the solver, as demonstrated by the three-dimensional run with $\tau=0.01$.
  • The method is stated for second-order temporal discretizations and for three-dimensional problems, extending beyond solvers that were limited to first-order schemes.
  • All iterates stay inside $(-1,1)$, so the logarithmic singularities are handled directly rather than by truncation or regularization.
  • Mass is preserved up to the stopping tolerance, making the solver compatible with conservative formulations of the phase-field dynamics.

Reading between the lines

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

  • Editorial inference: the same minimax-ADMM splitting should transfer to other phase-field models whose chemical potential contains a convex singular term plus a concave polynomial part, because the proof only uses monotonicity of the scalar subproblem and positivity of the discrete bilaplacian.
  • Editorial inference: with time-step restrictions removed, the practical bottleneck becomes iteration count; tuning $\rho_u$ and $\rho_w$ adaptively could reduce the observed number of iterations near late-stage coarsening.
  • Editorial inference: a direct testable extension is whether the contraction rate of the Lyapunov functional depends on the ratio $\rho_u/\rho_w$; if it does, simple closed-form rate estimates could guide parameter choice without extra computation.
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

2 major / 6 minor

Summary. The paper presents an ADMM-based iterative solver for the nonlinear systems arising from convex-splitting finite-difference schemes for the Cahn-Hilliard equation with the Flory-Huggins logarithmic potential. The solver splits the discrete energy into linear and nonlinear parts, uses FFT-based solves for the linear subproblem and scalar Newton solves for the singular logarithmic subproblem, and updates Lagrange multipliers in the standard ADMM fashion. The authors claim that the solver is unconditionally convergent with no restrictions on the time step tau or mesh size h, that it preserves the pointwise bound (-1,1) throughout the iterations, and that it nearly conserves mass. A detailed convergence proof is given for the first-order-in-time convex-splitting scheme (Lemma 4.2, Proposition 4.3, Theorem 4.4). The corresponding theorems for the second-order-in-time scheme (Theorems 4.7 and 4.8) are stated without proof, with the comment that the proofs are 'similar' and omitted. Numerical experiments include a convergence test and 2D/3D coarsening simulations, which use the second-order scheme.

Significance. The first-order convergence proof is a genuine contribution: it avoids the strict separation property, holds for arbitrary positive step sizes, and keeps the output iterates in the physical interval (-1,1). The algorithm is simple and practical, and the numerical results are consistent with the claimed robustness. However, the central 'unconditional convergence' claim is rigorously established only for the first-order scheme; the second-order scheme, which is the one used in the numerical sections, lacks proof. This gap is substantial because the second-order algorithm contains additional terms that are not covered by the first-order argument. With a complete second-order analysis, this would be a strong paper; without it, the main advertised claim is not fully proven. The mass-conservation statement for the second-order scheme also appears to be incorrect as stated.

major comments (2)
  1. [Section 4, Theorems 4.7 and 4.8] The convergence and mass-conservation theorems for the second-order scheme are asserted without proof. The text in Section 4 states 'the proofs for the second-order scheme are similar, and thus we omit it here.' This is not sufficient because Algorithm 2, equations (3.13)-(3.15), differs from Algorithm 1 in several load-bearing ways: the BDF2 combination (4/3)u^n - (1/3)u^{n-1}, the factor 2/3 in front of tau Delta_h w1, and the extra stabilization term A tau theta0^2 Delta_h(u^{n+1}-u^n) in (2.12b). These changes alter the energy (3.12) and the subproblems, and it is not demonstrated that the analogue of Lemma 4.2 holds. Since the numerical experiments in Section 5 use the second-order scheme, the central claim of unconditional convergence for the scheme actually used in the experiments is not supported by the analysis. Please provide a complete proof of Theorems 4.7 and 4.8, or explicitly limit the claim to the first-order scheme.
  2. [Section 4, Proposition 4.8] The stated mass bound (4.25) does not follow from Algorithm 2. Summing (3.13b) and (3.14b) gives u2^(k+1) - (4/3)u^n + (1/3)u^{n-1} = (2 tau/3) Delta_h w1^(k+1) + rho_w (w2^(k) - w2^(k+1)). Taking the mass M defined in (4.20) yields M(u2^(k+1)) - (4/3)M(u^n) + (1/3)M(u^{n-1}) = rho_w M(w2^(k)-w2^(k+1)), which is not the bound (4.25) involving M(u^n). The proposition should be corrected to reflect the BDF2 mass relation, or the statement of the claim should be revised.
minor comments (6)
  1. [Equation (1.1) and (5.1)] The gradient term in the energy appears as (epsilon^2/2) |nabla u|, which is missing the square; it should be (epsilon^2/2) |nabla u|^2.
  2. [Equation (4.2b)] The first equality in (4.2b) should have theta0 u^n, not theta0 u*, to be consistent with (2.11b) and the verification of (4.3c).
  3. [Section 2, equations (2.9)] For the 3D periodic grid, the inner products should include h^3 (or h^d in d dimensions) rather than h^2 as written; with N^3 summation points, h^2 is dimensionally inconsistent for a 3D domain.
  4. [Section 5, stopping criteria (5.5), (5.9), (5.13)] The stopping criteria include an explicit h^2 factor in front of the squared discrete norms. Since the discrete L2 norm defined in (2.9) already includes the mesh-size factor, the criterion should be clarified or written in terms of the actual vector norms used in the implementation.
  5. [Section 4, Proposition 4.3] The uniqueness of the stationary point is invoked in (4.15) but not proved. A short argument that any stationary point of L must satisfy (2.11) via (4.3), and hence uniqueness follows from the known unique solvability of the convex-splitting scheme, would make the proof complete.
  6. [Section 5.1, Table 1] The convergence rate for the second-order scheme at the first refinement is 0.630, which is far from the expected second order; a brief comment on this transient behavior would be helpful.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular reduction found; the convergence proof is self-contained relative to the external convex-splitting scheme, with only minor non-load-bearing self-citations and an omitted second-order proof that is a completeness gap, not circularity.

full rationale

The convergence target u^{n+1} is the solution of the convex splitting system (2.11)/(2.12), whose existence and uniqueness are cited from the external work [5], not from the present authors. Lemma 4.1 constructs a stationary point of the ADMM Lagrangian from that solution, and Lemma 4.2 derives the descent inequality for Ψ^(k) algebraically from the algorithm equations (3.9)-(3.11) using only positivity of -Δ_h and monotonicity of the logarithm; no fitted parameter or posterior data enters. The limit is identified with u^{n+1} through the KKT equations, which are equivalent to the original scheme by construction, but that equivalence is the intended target of the solver being proved, not a circular input. The only self-citations ([12], [24], involving Z. Qiao) appear in introductory survey lists of discretization methods and do not carry the convergence claim. One genuine caveat is not circularity but an omitted proof: Section 4 states 'Here, we only consider the first-order scheme; the proofs for the second-order scheme are similar, and thus we omit it here,' yet Theorem 4.7 and Section 5's experiments rely on the second-order Algorithm 2, whose additional BDF2 terms and Aτθ0^2Δ_h(u^{n+1}-u^n) stabilization are not analyzed. This is a completeness/correctness risk, not a circular reduction. Score 2 reflects the minor self-citation/content gap rather than any equivalence between inputs and outputs.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central convergence claim rests on three supports: the known existence theorem from [5] for the discrete scheme, the monotonicity of the logarithmic scalar equation which guarantees the subproblem is well-posed, and the asserted but unproved extension to the second-order scheme. No new physical entities are introduced.

free parameters (3)
  • ρ_u = user-selected (e.g., τ^{-1/2} or 1)
    ADMM penalty parameter for u-coupling; convergence holds for any positive value, but practical rate depends on the choice.
  • ρ_w = user-selected (e.g., τ^{1/2} or 1)
    ADMM penalty parameter for w-coupling; convergence holds for any positive value.
  • α = 1/2 in all numerical tests
    Splitting weight in Z1/Z2 decomposition; must lie in (0,1), convergence unconditional but rate depends on it.
assumptions (3)
  • domain assumption The first-order convex splitting scheme (2.11) has a unique solution u* in (-1,1) for all τ and h (cited from [5]).
    Used in Lemma 4.1 to construct the stationary point that the ADMM iterates converge to. The paper does not reprove this result.
  • standard math The scalar equations (3.10) have a unique solution in (-1,1) at each grid point because log(1+u)-log(1-u) is strictly increasing from -∞ to +∞ and the linear terms have nonnegative coefficients.
    Stated only in Remark 3.1 as 'monotonic derivative'; existence is not proven in the text, but follows from the monotone function argument.
  • ad hoc to paper The convergence analysis for the second-order scheme (Theorems 4.7, 4.8) is analogous to the first-order analysis.
    Section 4 asserts 'the proofs for the second-order scheme are similar' without giving them. This is an unverified premise for the second-order convergence claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Overcoming logarithmic singularities in the Cahn-Hilliard equation with Flory-Huggins potential: An unconditionally convergent ADMM approach." pith.science (2026). https://pith.science/paper/FFAGPCN3

@misc{pith2026250609361,
  author       = {Pith},
  title        = {Pith review of: Overcoming logarithmic singularities in the Cahn-Hilliard equation with Flory-Huggins potential: An unconditionally convergent ADMM approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FFAGPCN3}},
  note         = {Machine review of arXiv:2506.09361}
}
abstract

The Cahn-Hilliard equation with Flory-Huggins potential serves as a fundamental phase field model for describing phase separation phenomena. Due to the presence of logarithmic singularities at $u=\pm 1$, the solution $u$ is constrained within the interval $(-1,1)$. While convex splitting schemes are commonly employed to preserve this bound and guarantee unconditional unique solvability, their practical implementation requires solving nonlinear systems containing singular logarithmic terms at each time step. This introduces significant challenges in both ensuring convergence of iterative solvers and maintaining the solution bounds throughout the iterations. Existing solvers often rely on restrictive conditions -- such as the strict separation property or small time step sizes -- to ensure convergence, which can limit their applicability. In this work, we introduce a novel iterative solver that is specifically designed for singular nonlinear systems, with the use of a variant of the alternating direction method of multipliers (ADMM). By developing a tailored variable splitting strategy within the ADMM framework, our method efficiently decouples the challenging logarithmic nonlinearity, enabling effective handling of singularities. Crucially, we rigorously prove the unconditional convergence of our ADMM-based solver, which removes the need for time step constraints or strict separation conditions. This allows us to fully leverage the unconditional solvability offered by convex splitting schemes. Comprehensive numerical experiments demonstrate the superior efficiency and robustness of our ADMM variant, strongly validating both our algorithmic design and theoretical results.

Figures

Figures reproduced from arXiv: 2506.09361 by the authors.

Figure 1
Figure 1. (Convergence test) The number of iterations over time t. Left: first-order scheme, right: second-order scheme. 0 0.1 0.2 0.3 0.4 -5 -4 -3 -2 -1 0 1 10-9 0 0.1 0.2 0.3 0.4 -2 -1 0 1 2 3 10-8 [PITH_FULL_IMAGE:figures/full_fig_p015_1.png] view at source ↗
Figure 2
Figure 2. (Convergence test) The change in total mass over time t. Left: first-order scheme, right: second-order scheme. This example demonstrates the numerical results of the coarsening process starting from smooth initial conditions. The coefficients in the solver are (5.8) ρu = 1, ρw = 1, α = 1/2. The convergence criterion is set as (5.9) q h 2∥u (k) 1 − u (k) 2 ∥ 2 2 + h 2∥w (k) 1 − w (k) 2 ∥ 2 2 ≤ γ = 10−8 . We use the s… view at source ↗
Figure 3
Figure 3. (Convergence test) The energy over time t. Left: first￾order scheme, right: second-order scheme. 0 0.1 0.2 0.3 0.4 0.05 0.1 0.15 0 0.1 0.2 0.3 0.4 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0 0.1 0.2 0.3 0.4 0.04 0.06 0.08 0.1 0 0.1 0.2 0.3 0.4 0.02 0.04 0.06 0.08 0.1 [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: (Convergence test) 1 − max(u) and 1 + min(u) over time t. Left: first-order scheme, right: second-order scheme. The number of iterations, the change in total mass, the energy, 1 − max(u) and 1 + min(u) over time are given in [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]
Figure 5
Figure 5. Figure 5: (2D coarsening process) The coarsening process of the phase field u at different time steps. From top to bottom, left to right, t = 0, 0.2, 0.4, 0.6, 0.8, 1. The numerical solution is close to 1 in the white region and close to −1 in the black region. 0 0.2 0.4 0.6 0.8…
Figure 6
Figure 6. Figure 6 [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: (2D coarsening process) 1 − max(u) and 1 + min(u) over time t [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]
Figure 8
Figure 8. Figure 8: (3D coarsening process) The coarsening process of the phase field u at different time steps. The figure shows the interface of u = 0 by interpolating the numerical solution. 0 0.5 1 1.5 2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 -2 -1.5 -1 -0.5 0 0.5 0.1 0.12 0.14 0.16 0.18 [P…
Figure 9
Figure 9. Figure 9: (3D coarsening process) The energy over time t. and constructing effective initial guesses. Finally, inspired by the ideas presented in this work, we plan to investigate directly linearized discrete schemes to avoid the high computational cost of repeatedly solving non…
Figure 10
Figure 10. Figure 10: (3D coarsening process) Left: the number of iterations over time t, middle: the variation in total mass over time t. 0.5 1 1.5 2 10-2 10-1 100 0.5 1 1.5 2 10-2 10-1 100 [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: (3D coarsening process) 1 − max(u) and 1 + min(u) over time t. [3] J. W. Cahn and J. E. Hilliard, Free energy of a nonuniform system. i. interfacial free energy, The Journal of Chemical Physics 28 (1958), 258–267. [4] L. Q. Chen and J. Shen, Applications of semi-impli…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [5]

    W. Chen, C. Wang, X. Wang, and S. M. Wise,Positivity-preserving, energy stable numerical schemes for the Cahn-Hilliard equation with logarithmic potential, Journal of Computational Physics: X3(2019), 100031

    work page 2019

  2. [9]

    A. E. Diegel, C. Wang, and S. M. Wise,Convergence analysis of a preconditioned steepest descent solver for the Cahn-Hilliard equation with logarithmic potential, International Journal of Numerical Analysis and Modeling22(2025), 483–509

    work page 2025

  3. [1]

    Abels and M

    H. Abels and M. Wilke,Convergence to equilibrium for the Cahn-Hilliard equation with a logarithmic free energy, Nonlinear Analysis: Theory, Methods & Applications67(2007), 3176–3193

    work page 2007

  4. [2]

    D. P. Bertsekas and J. N. Tsitsiklis,Parallel and distributed computation: Numerical methods, Athena Scientific, 1997. 20 RUO LI, SHENGTONG LIANG, AND ZHONGHUA QIAO 0 0.5 1 1.5 220 40 60 80 100 0.5 1 1.5 210-10 10-5 Figure 10.(3D coarsening process) Left: the number of iterations over timet, middle: the variation in total mass over timet. 0.5 1 1.5 210-2 ...

    work page 1997

  5. [3]

    J. W. Cahn and J. E. Hilliard,Free energy of a nonuniform system. i. interfacial free energy, The Journal of Chemical Physics28(1958), 258–267

    work page 1958

  6. [4]

    L. Q. Chen and J. Shen,Applications of semi-implicit Fourier-spectral method to phase field equations, Computer Physics Communications108(1998), 147–158

    work page 1998

  7. [6]

    Cherfils, A

    L. Cherfils, A. Miranville, and S. Zelik,The Cahn-Hilliard equation with logarithmic poten- tials, Milan Journal of Mathematics79(2011), 561–596

    work page 2011

  8. [7]

    M. I. M. Copetti and C. M. Elliott,Numerical analysis of the Cahn-Hilliard equation with a logarithmic free energy, Numerische Mathematik63(1992), 39–65

    work page 1992

Show all 27 references
  1. [8]

    Debussche and L

    A. Debussche and L. Dettori,On the Cahn-Hilliard equation with a logarithmic free energy, Nonlinear Analysis: Theory, Methods & Applications24(1995), 1491–1514

    1995

  2. [10]

    Doi,Soft matter physics, Oxford University Press, 2013

    M. Doi,Soft matter physics, Oxford University Press, 2013

    2013

  3. [11]

    Du and X

    Q. Du and X. Feng,Chapter 5 - the phase field method for geometric moving interfaces and their numerical approximations, Geometric partial differential equations - part i, 2020, pp. 425–508

    2020

  4. [12]

    Q. Du, L. Ju, X. Li, and Z. Qiao,Maximum bound principles for a class of semilinear parabolic equations and exponential time-differencing schemes, SIAM Review63(2021), no. 2, 317– 359. AN ITERATIVE SOL VER FOR CH EQUATION WITH LOG-SINGULARITIES 21

    2021

  5. [13]

    C. M. Elliott and S. Larsson,Error estimates with smooth and nonsmooth data for a finite element method for the Cahn-Hilliard equation, Mathematics of Computation58(1992), 603–630

    1992

  6. [14]

    Fazel, K

    M. Fazel, K. T. Pong, D. Sun, and P. Tseng,Hankel matrix rank minimization with ap- plications to system identification and realization, SIAM Journal on Matrix Analysis and Applications34(2013), 946–977

    2013

  7. [15]

    Furihata,A stable and conservative finite difference scheme for the Cahn-Hilliard equa- tion, Numerische Mathematik87(2001), 675–699

    D. Furihata,A stable and conservative finite difference scheme for the Cahn-Hilliard equa- tion, Numerische Mathematik87(2001), 675–699

    2001

  8. [16]

    Hochbruck and A

    M. Hochbruck and A. Ostermann,Exponential integrators, Acta Numerica19(2010), 209–286

    2010

  9. [17]

    M. O. Karabag, D. Fridovich-Keil, and U. Topcu,Alternating direction method of multipliers for decomposable saddle-point problems, 2022 58th Annual Allerton Conference on Commu- nication, Control, and Computing (Allerton), 2022, pp. 1–8

    2022

  10. [18]

    M. O. Karabag, M. Ornik, and U. Topcu,Deception in supervisory control, IEEE Transactions on Automatic Control67(2021), 738–753

    2021

  11. [19]

    Li and T

    D. Li and T. Tang,Stability of the semi-implicit method for the Cahn-Hilliard equation with logarithmic potentials, Annals of Applied Mathematics37(2021), 31–60

    2021

  12. [20]

    Miranville,On a phase-field model with a logarithmic nonlinearity, Applications of Math- ematics57(2012), 215–229

    A. Miranville,On a phase-field model with a logarithmic nonlinearity, Applications of Math- ematics57(2012), 215–229

    2012

  13. [21]

    ,The Cahn-Hilliard equation: Recent advances and applications, Society for Industrial and Applied Mathematics, 2019

    2019

  14. [22]

    Miranville and S

    A. Miranville and S. Zelik,Robust exponential attractors for the Cahn-Hilliard type equations with singular potentials, Mathematical Methods in the Applied Sciences27(2004), 545–582

    2004

  15. [23]

    Tang,On effective numerical methods for phase-field models, Proceedings of the interna- tional congress of mathematicians: Rio de janeiro 2018, 2018, pp

    T. Tang,On effective numerical methods for phase-field models, Proceedings of the interna- tional congress of mathematicians: Rio de janeiro 2018, 2018, pp. 3669–3690

    2018

  16. [24]

    Tang and Z

    T. Tang and Z. Qiao,Efficient numerical methods for phase-field equations (in Chinese), Scientia Sinica Mathematica50(2020), 775–794

    2020

  17. [25]

    S. M. Wise, C. Wang, and J. S. Lowengrub,An energy-stable and convergent finite-difference scheme for the phase field crystal equation, SIAM Journal on Numerical Analysis47(2009), 2269–2288

    2009

  18. [26]

    Wodo and B

    O. Wodo and B. Ganapathysubramanian,Modeling morphology evolution during solvent- based fabrication of organic solar cells, Computation Materials Science55(2012), 113–126

    2012

  19. [27]

    M. Yuan, W. Chen, C. Wang, S. Wise, and Z. Zhang,A second order accurate in time, energy stable finite element scheme for the Flory-Huggins-Cahn-Hilliard equation, Advances in Applied Mathematics and Mechanics14(2022), 1477–1508. CAPT, LMAM and School of Mathematical Sciences,...

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.