Tridiagonal matrices arrises under Markovian dynamics. The computational cost for doing inference with a (general) tridiagonal matrix is $O(n)$, while a general matrix is $O(n^3)$ in the number of datapoints $n$. Exploiting this is critical for scalability.
Tridiagonal matrices arrises under Markovian dynamics. The computational cost for doing inference with a (general) tridiagonal matrix is $O(n)$, while a general matrix is $O(n^3)$ in the number of datapoints $n$. Exploiting this is critical for scalability.