Abstract
Multiple equilibrium states arise in many physical systems, including various types of liquid crystal structures. Having the ability to reliably compute such states enables more accurate physical analysis and understanding of experimental behavior. This paper adapts and extends a deflation technique for the computation of multiple distinct solutions arising in the context of modeling equilibrium configurations of nematic and cholesteric liquid crystals. The deflation method is applied as part of an overall free-energy variational approach and is modified to fit the framework of optimization of a functional with pointwise constraints. It is shown that multigrid methods designed for the undeflated systems may be applied to efficiently solve the linear systems arising in the application of deflation. For the numerical algorithm, the deflation approach is interwoven with nested iteration, creating a dynamic and efficient method that further enables the discovery of distinct solutions. Finally, four numerical experiments are performed demonstrating the efficacy and accuracy of the algorithm in detecting important physical phenomena, including bifurcation and disclination behaviors.
Introduction
In soft matter physics, liquid crystals are characterized by their ability to exist in a variety of stable or meta-stable equilibrium configurations. These states are highly sensitive to external influences such as temperature, electric fields, and boundary conditions. For researchers using numerical simulations, capturing the full range of these configurations is a significant challenge. Standard solvers, such as Newton’s method, are inherently local; they typically converge to the solution nearest to the initial guess and fail to explore the broader energy landscape.
Traditionally, scientists have relied on "multistart" methods—applying random initial guesses in the hope of landing in different basins of attraction. However, this approach is unsystematic, computationally expensive, and provides no guarantee of discovering all physically relevant states, such as topological defects or bifurcated structures. This paper presents a more rigorous and efficient alternative: a systematic numerical deflation technique designed to discover multiple equilibrium states in liquid crystal models.
Methodology
The Deflation Operator
The core of the proposed method is the adaptation of a deflation technique to the Frank-Oseen free-energy model. When a solution $r$ is identified, the algorithm applies a transformation to the nonlinear residual. This transformation effectively multiplies the original problem by a factor that approaches infinity as the solver nears the known solution $r$. By "canceling out" known solutions, the algorithm forces the Newton iteration to seek out new, distinct configurations that would otherwise be ignored.
Pointwise Constraints and Optimization
Liquid crystal models require the "director" (representing the average molecular orientation) to maintain a unit length at every point in the domain. The authors successfully integrated the deflation operator into a framework that manages these nonlinear pointwise constraints using Lagrange multipliers. This ensures that every discovered state remains physically valid.
Computational Efficiency
A common drawback of deflation is that it can transform a sparse, easily solvable system into a dense one. To maintain efficiency, the authors utilized the Sherman-Morrison formula, demonstrating that the deflated system is essentially a rank-one update of the original system. This allows the continued use of highly optimized, pre-existing multigrid solvers. Furthermore, the algorithm employs Nested Iteration (NI), starting on a coarse grid and progressively refining the mesh. Solutions found at coarser levels serve as high-quality initial guesses for finer levels, significantly reducing the total computational "Work Units" (WUs) required.
Numerical Experiments and Results
The efficacy of the algorithm was evaluated through four benchmark problems in liquid crystal physics, with simulations performed on meshes up to $256 \times 256$ (over 1.1 million degrees of freedom).
- Tilt-Twist Configuration: The method successfully mapped out "pitchfork bifurcations" in a nematic liquid crystal between two substrates, identifying the exact point where a simple twist becomes energetically unfavorable compared to a complex tilt-twist state.
- Freedericksz Transition: This standard test case involved an external electric field. The numerical results for critical voltages ($V_c$) matched analytical predictions with high precision.
- Escape and Disclination: The algorithm demonstrated the ability to distinguish between "escape" solutions—where the director tilts into the third dimension to avoid a singularity—and true topological disclinations.
- Cholesteric Complexity: In the study of chiral (helical) liquid crystals, the deflation process discovered six distinct equilibrium states. In contrast, a standard solver without deflation was only able to identify three, illustrating the superior discovery power of the proposed method.
Discussion and Conclusion
The integration of deflation with nested iteration and multigrid solvers provides a deterministic and efficient path to exploring the energy landscapes of complex physical systems. The method not only captures stable states but also identifies "saddle point" solutions that are mathematically elusive yet physically significant.
While the approach is highly effective, the authors note that the choice of the initial set of guesses still plays a role in the order of solution discovery. Future work may involve the "dynamic generation" of initial guesses and the investigation of "generalized tunneling methods" to overcome high energy barriers between states. Additionally, extending the algorithm to handle the complex, non-rectangular geometries found in industrial applications like LCD sensors remains a key objective. Ultimately, this deflationary approach offers a robust framework for the systematic discovery of hidden states in soft matter physics.
Summary updated