Exponential Rosenbrock-Euler Integrators for Elastodynamic Simulation

Area of Science:

• Computational physics and computer graphics.
• Numerical methods for dynamic simulations.

Background:

• Simulating soft flexible objects is computationally expensive due to nonlinear stiffness calculations.
• Standard implicit integrators often introduce artificial damping, affecting simulation accuracy.
• Existing methods struggle with non-convex energies and non-positive definite stiffness matrices.

Purpose of the Study:

• To develop a novel numerical method for efficient and accurate simulation of soft material dynamics.
• To avoid discretization-dependent artificial damping in simulations.
• To enable stable simulations with large time steps, suitable for computer graphics applications.

Main Methods:

• Proposed and implemented an exponential Rosenbrock-Euler (ERE) integration method.
• The ERE method is designed to handle nonlinear stiffness and non-convex elastic energies.
• The integrator is specifically engineered for cases with non-positive definite symmetric stiffness matrices.

Main Results:

• The ERE method effectively avoids discretization-dependent artificial damping.
• The method performs well with large time steps, reducing computational cost.
• Demonstrated accurate qualitative behavior even with challenging non-convex energies and non-positive definite stiffness.

Conclusions:

• The exponential Rosenbrock-Euler method offers an efficient and stable approach for simulating soft materials.
• This method broadens the applicability of numerical solvers to a wider range of practical scenarios.
• The system shows efficient performance across diverse soft material simulations.