🔢 mathematics

A locking free mixed FEM based on a pure pseudostress based formulation for the elasticity eigenproblem

This paper presents and analyzes a novel locking-free mixed finite element method for the elasticity eigenvalue problem in two and three dimensions that utilizes a pure pseudostress formulation without enforcing symmetry, while providing rigorous convergence proofs, a priori and a posteriori error estimates, and numerical validation.

Original authors: Arbaz Khan, Felipe Lepe, Jesus Vellojin

Published 2026-07-09

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Arbaz Khan, Felipe Lepe, Jesus Vellojin

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are an engineer trying to predict how a bridge, a car chassis, or a building will vibrate when the wind blows or an earthquake hits. To do this, you need to solve a complex mathematical puzzle called the "elasticity eigenvalue problem." In simple terms, this puzzle tells you the natural "notes" (frequencies) a structure sings when it shakes. If you get the math wrong, your simulation might say the bridge is safe when it's actually about to collapse, or it might get stuck on a calculation error known as "locking."

This paper introduces a new, smarter way to solve this puzzle using a method called the Mixed Finite Element Method (FEM). Here is how the authors' approach works, explained through everyday analogies:

1. The Problem with the Old Way: "The Tightrope Walker"

Traditionally, simulating how materials like rubber or soft tissues (which are almost impossible to squish) vibrate is like asking a tightrope walker to balance on a wire that keeps getting thinner. As the material gets "stiffer" or "incompressible" (like rubber), the math becomes unstable. The computer gets confused, the calculations "lock up," and the results become garbage. This is called numerical locking.

2. The New Solution: "The Pure Pseudostress"

The authors propose a new way to look at the problem. Instead of trying to track the "stress" (the internal forces) and the "displacement" (how much the material moves) separately and forcing them to fit together perfectly, they decided to focus on a single, clever variable they call pseudostress.

The Analogy: Imagine you are trying to figure out how a crowd of people is moving in a stadium. Instead of trying to count every single person's face (displacement) and every single push they make (stress) simultaneously, you just track the "flow" of the crowd (pseudostress).
The Magic Trick: Once the computer calculates this "flow," the authors show you can easily reconstruct the actual movement of the people (displacement) afterward using a simple recipe (post-processing). You don't need to force the math to be perfect during the calculation; you just calculate the flow and fix the details later.

3. No "Symmetry" Required: "The Flexible Puzzle"

Most traditional methods require the math to be perfectly symmetrical (like a mirror image) to work. This is like trying to solve a puzzle where every piece must be a perfect square. The authors' new method is like a puzzle where the pieces can be any shape, as long as they fit together loosely. They don't need to enforce strict symmetry rules, which makes the math much more flexible and less likely to break when dealing with "squishy" materials.

4. The "Locking-Free" Promise

The paper proves that this new method works perfectly whether the material is a hard steel beam or a soft, almost-incompressible rubber.

The Result: Even when the material is so stiff that it refuses to change volume (the "nearly incompressible" limit), the math doesn't crash. It continues to give accurate answers without "locking up."

5. The "Smart Ruler": Error Estimation

One of the coolest features of this paper is a new tool they built called an a posteriori error estimator.

The Analogy: Imagine you are painting a wall. A normal method just paints the whole wall and hopes for the best. This new method is like a smart ruler that walks around after you paint a section and says, "Hey, this corner is messy, you need to paint it again," while saying, "This flat part is perfect, leave it alone."
Why it matters: This allows the computer to automatically focus its power only on the tricky, messy parts of the structure (like sharp corners or cracks) and ignore the smooth, easy parts. This saves massive amounts of computing time and money.

6. The Proof: "The Test Drive"

The authors didn't just talk about theory; they ran the numbers.

They tested their method on 2D shapes (like a square) and 3D shapes (like a cube with a notch cut out of it).
They tested it with materials ranging from normal to extremely "rubbery" (Poisson ratio close to 0.5).
The Outcome: The method worked perfectly every time. It found the correct vibration frequencies, it didn't get stuck (no locking), and the "smart ruler" correctly identified where the math needed more attention.

Summary

In short, this paper presents a robust, flexible, and efficient new recipe for simulating how structures vibrate. By focusing on a single "flow" variable (pseudostress) and using a smart way to check for errors, the authors have created a tool that works reliably even for the most difficult, "squishy" materials, without the computer getting stuck or needing to be told exactly how to behave at every step.

Technical Summary: A Locking-Free Mixed FEM Based on a Pure Pseudostress Formulation for the Elasticity Eigenproblem

Problem Statement
The paper addresses the numerical approximation of the linear elasticity eigenvalue problem, which governs the vibration of elastic structures. The primary challenge addressed is the "locking" phenomenon that occurs in standard numerical methods when the Poisson ratio ( $\nu$ ) approaches $1/2$ (the nearly incompressible or perfectly incompressible regime). In this regime, the Lamé coefficient $\lambda$ tends to infinity, causing standard formulations to fail or lose accuracy. Furthermore, the authors aim to develop a formulation that avoids the enforcement of stress symmetry (either weakly or strongly), a common requirement in mixed elasticity methods that complicates implementation and analysis.

Methodology
The authors propose a novel mixed finite element method (FEM) based exclusively on the pseudostress tensor ( $\rho$ ). The displacement field ( $u$ ) is not an independent unknown in the primary system but is recovered via a post-processing step.

Pure Pseudostress Formulation:
The continuous problem is reformulated to depend only on the pseudostress tensor $\rho$ and the displacement $u$ . By utilizing the identity relating $\rho$ to $\nabla u$ , the system is reduced to a pure pseudostress system of partial differential equations. The variational formulation is defined on the space $X = H(\text{div}, \Omega)$ , specifically on a subspace $X_0$ where the trace of the tensor integrates to zero.
Mathematical Analysis:
- Non-Compact Operator Theory: Unlike standard elasticity formulations where the solution operator is compact, the solution operator $T^\lambda$ associated with this pseudostress formulation is non-compact. Consequently, the authors analyze convergence and error estimates using the theory of non-compact operators (referencing works [7, 8]).
- Spectral Characterization: The spectrum of the solution operator is decomposed into three parts: an infinite-multiplicity eigenvalue at 1 (associated with divergence-free tensors), an eigenvalue at 0, and a sequence of finite-multiplicity eigenvalues converging to 0.
- Limit Analysis: The paper rigorously proves that as $\lambda \to \infty$ (i.e., $\nu \to 1/2$ ), the spectrum of the nearly incompressible problem converges to the spectrum of the perfectly incompressible case, which coincides with the Stokes eigenvalue system. This convergence is established without numerical locking.
Discretization:
The method employs Raviart–Thomas (RT) finite elements, which are naturally suited for $H(\text{div})$ -conforming spaces. The discrete spaces are constructed to satisfy the trace-free condition required for $X_0$ .
Error Analysis:
- A Priori Estimates: The authors derive optimal order error estimates for both eigenvalues and eigenfunctions. Notably, the eigenvalue error converges with order $O(h^{2\min\{r,k\}})$ , where $r$ is the regularity of the eigenfunction and $k$ is the polynomial degree.
- A Posteriori Estimates: A residual-based a posteriori error estimator is introduced. The authors prove its reliability (upper bound) and efficiency (lower bound) for both the nearly incompressible and perfectly incompressible regimes. The estimator is designed to be robust with respect to the Poisson ratio.

Key Contributions

Locking-Free Formulation: The proposed method avoids numerical locking as $\nu \to 1/2$ , maintaining optimal convergence rates in both nearly and perfectly incompressible regimes.
No Symmetry Enforcement: The formulation does not require the explicit enforcement of stress symmetry, simplifying the finite element implementation.
Pure Pseudostress Approach: The work extends the pure pseudostress concept (previously applied to Stokes problems) to the elasticity eigenvalue problem, allowing for the recovery of displacement and stress via post-processing.
Non-Compact Analysis: The paper successfully applies the theory of non-compact operators to the elasticity eigenproblem, a necessary step due to the specific structure of the pseudostress formulation.
Robust A Posteriori Estimator: A novel error indicator is proposed that remains efficient and reliable even in the limit of incompressibility, a feature not commonly found in existing estimators for this specific formulation.

Numerical Results
The theoretical findings are validated through numerical experiments in 2D and 3D using the FEniCSx environment.

Convergence: Tests on a square domain and a unit cube demonstrate that the method achieves second-order convergence ( $O(h^2)$ ) for eigenvalues across various Poisson ratios, including $\nu = 0.5$ .
Adaptive Refinement: Experiments on non-convex domains (a 2D sector with a re-entrant corner and a 3D notched cube) show that the adaptive strategy driven by the proposed estimator successfully recovers optimal convergence rates ( $O(\text{dof}^{-1})$ in 2D and $O(\text{dof}^{-2/3})$ in 3D) despite the reduced regularity of the eigenfunctions near singularities.
Robustness: The efficiency indices remain bounded and stable as $\nu$ approaches $0.5$, confirming the locking-free nature of the method and the robustness of the error estimator.

Significance
The paper claims that this work provides a robust, locking-free numerical framework for analyzing the vibration of elastic structures, particularly in regimes where materials are nearly or perfectly incompressible. By eliminating the need for symmetry enforcement and utilizing a pure pseudostress formulation, the method offers a simplified yet mathematically rigorous alternative to traditional mixed methods. The ability to accurately capture the spectrum in the incompressible limit without spurious eigenvalues or loss of convergence order represents a significant advancement in the computational mechanics of elastic eigenproblems.