Array-representation Integration Factor Method for High-dimensional Systems.

Area of Science:

• Numerical analysis
• Computational science
• Scientific computing

Background:

• Implicit integration methods (IIF) offer stability for stiff reactions and high-order derivatives in numerical simulations.
• A key challenge for IIF is managing dense exponential matrices arising from sparse discretization of differential operators.
• Existing compact IIF (cIIF) methods are effective for Laplacian operators but limited in scope.

Purpose of the Study:

• To introduce an array-representation technique for efficient handling of exponential matrices from general linear differential operators.
• To develop an array-representation compact implicit integration method (AcIIF) applicable to high-dimensional systems.
• To demonstrate the efficiency, accuracy, and robustness of AcIIF in simulating complex scientific models.

Main Methods:

• Developed an array-representation technique to compute exponential matrices for general linear differential operators.
• Incorporated this technique into the implicit integration method (IIF) to create AcIIF.
• Applied AcIIF to simulate reaction-diffusion equations, Fokker-Planck equations (3D/4D), and chemical master equations.

Main Results:

• The array-representation technique requires computing exponentials only for small, problem-independent matrix sizes.
• AcIIF demonstrated significant efficiency, accuracy, and robustness across various high-dimensional simulations.
• The method effectively handles cross-derivatives and non-constant diffusion coefficients.

Conclusions:

• The array-representation technique provides an efficient solution for handling exponential matrices in IIF.
• AcIIF preserves the excellent stability of IIF while extending its applicability to complex, high-dimensional problems.
• This method has broad potential for simulating diverse complex systems with high-dimensional data.