Finite-element discretization of the smectic density equation

Patrick E. Farrell, Abdalaziz Hamdan, Scott P. MacLachlanarXiv 2022 · DOI · arXiv

Introduction

Smectic A liquid crystals represent a unique phase of matter that exhibits properties intermediate between those of conventional liquids and crystalline solids. They are characterized by a periodic, one-dimensional layered structure. Modeling the density variations within these layers requires solving a fourth-order partial differential equation (PDE). While fourth-order operators, such as the biharmonic operator used in thin-plate elasticity, are well-understood in numerical analysis, the "smectic density equation" introduces specific mathematical challenges that necessitate a specialized approach.

This paper provides a rigorous numerical foundation for this equation, evaluating three distinct finite-element discretization strategies: conforming, interior penalty, and mixed methods. The authors offer a comprehensive guide to their well-posedness, error bounds, and practical trade-offs in both two and three dimensions.

The Mathematical Challenge: The "Wrong-Sign" Helmholtz Operator

The primary difficulty in simulating smectic density lies in the nature of the fourth-order operator. Unlike the standard biharmonic equation, the smectic density equation features a "wrong-sign" second-order shift. This shift causes the operator to behave similarly to an indefinite Helmholtz equation, where the lowest-energy states are plane waves rather than smooth functions.

This characteristic leads to several numerical hurdles:

  1. Pollution Effects: As the wavenumber $q$ (representing the density of layers) increases, standard numerical error estimates degrade, requiring significant mesh refinement to maintain accuracy.
  2. Boundary Conditions: Natural boundary conditions for liquid crystals are often more complex than the "clamped" conditions typically addressed in mathematical literature.
  3. Dimensionality: Methods that perform well in 2D often become computationally prohibitive or mathematically complex when extended to 3D.

Finite Element Formulations

The authors analyze and compare three primary finite-element (FE) strategies to address these challenges:

1. $H^2$-Conforming Elements (Argyris)

This approach utilizes high-order (5th-degree) polynomials to ensure that both the solution and its derivatives remain continuous across element boundaries. While this method provides exceptional accuracy in 2D, it is notoriously difficult to implement. In 3D, the complexity becomes "explosive," requiring 9th-degree polynomials, making it practically unusable for large-scale simulations.

2. $C^0$ Interior Penalty Method (C0IP)

The C0IP method employs standard continuous elements but enforces higher-order smoothness weakly. This is achieved by adding "penalty" terms at the edges of elements to discourage discontinuities in the derivatives. While flexible and applicable to 3D geometries, C0IP systems can be difficult to solve efficiently using standard iterative methods.

3. Mixed Finite-Element Formulation

The core innovation of this research is the development of a mixed-method approach. This strategy decomposes the fourth-order PDE into a system of lower-order equations by introducing the gradient of the solution as an independent variable, constrained by a Lagrange multiplier. This technique, reminiscent of methods used in fluid dynamics (such as the Stokes problem), provides a balance of stability and solvability.

Numerical Experiments and Results

To validate these methods, the authors employed the Method of Manufactured Solutions, defining analytical sine-wave solutions and testing the algorithms on 2D unit squares and 3D unit cubes.

Key Findings:

  • Convergence Verification: All three methods achieved their theoretical convergence rates. The mixed method, in particular, demonstrated $O(h^{k+1})$ convergence in the $L^2$ norm, confirming the predicted error bounds.
  • The 3D Bottleneck: The study confirmed that while Argyris elements are superior in 2D, they are non-viable in 3D. The C0IP and Mixed methods emerged as the only practical choices for three-dimensional modeling.
  • Robustness and Scalability: The mixed formulation proved to be the most robust against variations in material parameters. Crucially, it is compatible with monolithic multigrid preconditioning, a high-performance computing technique essential for solving the massive systems of equations required for realistic 3D liquid crystal simulations.
  • Boundary Handling: The researchers successfully demonstrated the handling of full Neumann boundary conditions, which are physically relevant for soft matter physics but often omitted in simplified models.

Conclusion and Future Directions

This work establishes a clear roadmap for the numerical simulation of smectic liquid crystals. By identifying the mixed finite-element method as the most scalable and robust approach, the authors pave the way for more complex simulations.

Future work will focus on two main areas: the full implementation and optimization of multigrid solvers for the mixed formulation, and the extension of this framework to coupled models. In physical systems, density is intrinsically linked to the orientation of molecules (the director field); extending this rigorous numerical analysis to these non-linear, coupled systems is the next logical step in modeling the complex behavior of liquid crystals.

Summary updated

Abstract

The fourth-order PDE that models the density variation of smectic A liquid crystals presents unique challenges in its (numerical) analysis beyond more common fourth-order operators, such as the classical biharmonic. While the operator is positive definite, the equation has a "wrong-sign" shift, making it somewhat more akin to an indefinite Helmholtz operator, with lowest-energy modes consisting of plane waves. As a result, for large shifts, the natural continuity, coercivity, and inf-sup constants degrade considerably, impacting standard error estimates. In this paper, we analyze and compare three finite-element formulations for such PDEs, based on $H^2$-conforming elements, the $C^0$ interior penalty method, and a mixed finite-element formulation that explicitly introduces approximations to the gradient of the solution and a Lagrange multiplier. The conforming method is simple but is impractical to apply in three dimensions; the interior-penalty method works well in two and three dimensions but has lower-order convergence and (in preliminary experiments) seems difficult to precondition; the mixed method uses more degrees of freedom, but works well in both two and three dimensions, and is amenable to monolithic multigrid preconditioning. Our analysis reveals different behaviours of the error bounds with the shift parameter and mesh size for the different schemes. Numerical results verify the finite-element convergence for all discretizations, and illustrate the trade-offs between the three schemes.
BibTeX
@misc{farrell2022finite,
  author = {Patrick E. Farrell and Abdalaziz Hamdan and Scott P. MacLachlan},
  title = {{Finite-element discretization of the smectic density equation}},
  year = {2022},
  eprint = {2207.12916},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2207.12916},
}

Loading related content...

Figure