Core Concepts
Proposing a new regularization method using rank-1 coefficient mixed matrices for nonlinear DAEs.
Abstract
The study introduces a novel approach to address the singularity of system Jacobians in differential-algebraic equations (DAEs). Existing methods rely on symbolic computations, leading to high computational costs. The proposed method approximates the system Jacobian with more expressive symbolic matrices, called rank-1 coefficient mixed matrices. This approach aims to capture detailed algebraic relationships efficiently and effectively. By utilizing linear symbolic matrices, the method offers a faster combinatorial algorithm for singularity testing, ensuring computational tractability. Through numerical experiments, the method demonstrates applicability and efficiency for large-scale DAEs.
Stats
Iwata–Oki–Takamatsu proposed an IOT-method to find a certificate without symbolic computations.
The IOT method globally preserves the solutions of the DAE.
The proposed method extends the idea of the IOT method by using rank-1 coefficient mixed matrices.
The algorithm for finding a vanishing pair of a 1CM-matrix runs in O((n + m)3 log(n + m)) time.
Quotes
"The proposed method approximates the system Jacobian by more expressive symbolic matrices."
"Our algorithm is faster than existing algorithms and possesses global equivalence properties."
"The study confirms that our method runs fast for large-scale DAEs from real instances."