Efficient and effective multivariate time series (MTS) forecasting is critical for real-world applications, such as traffic management and energy dispatching. Most of the current deep learning studies (e.g., Spatio-Temporal Graph Neural Networks and …