Identification of Nonzero Elements in the Polynomial Matrices of Mixed Varma Processes
针对VARMA时间序列模型参数矩阵稀疏的特点，提出识别VARMA(1,1)模型中非零元素的方法，并展示如何将其推广到高阶模型构建，帮助提高计算效率。
SUMMARY The parameter density is the polynomial matrices of vector autoregressive moving average (VARMA) time series models is generally very low. Thus, estimation of fully parameterized high order models or of models with many variables is an ineffective, if not impractical, way of utilizing computing power. In the article we present methods for identifying the nonzero elements in VARMA(1,1) models, and show how these procedures can be used to construct models of higher order.