Machine-Learning-Driven Matrix Ordering for Power Grid Analysis
Ganqu Cui1,a, Wenjian Yu1,b, Xin Li2, Zhiyu Zeng3,c and Ben Gu3,d
1BNRist, Dept. Computer Science & Tech., Tsinghua Univ., Beijing, China
acgq15@mails.tsinghua.edu.cn
byu-wj@tsinghua.edu.cn
2iAPSE, Duke Kunshan Univ., Kunshan, China
xinli.ece@duke.edu
3Cadence Design Systems, Inc., Austin, USA
czzeng@cadence.com
dgxin@cadence.com
ABSTRACT
A machine-learning-driven approach for matrix ordering is proposed for power grid analysis based on domain decomposition. It utilizes support vector machine or artificial neural network to learn a classifier to automatically choose the optimal ordering algorithm, thereby reducing the expense of solving the subdomain equations. Based on the feature selection considering sparse matrix properties, the proposed method achieves superior efficiency in runtime and memory usage over conventional methods, as demonstrated by industrial test cases.
Keywords: Classification, Direct sparse solver, Fill-reducing ordering, Machine learning, Power grid analysis.
