generate AR covariance matrix of an Autoregressive model.

[nitaifingerhut]

Reference issue

What does this implement/fix?

Additional information

[rkern]

Can you fill out the information requested in the PR template? What is this function for? How does it fit in with the rest of scipy, which has nothing else for autoregressive models? This would probably fit better in statsmodels, which does, in case it doesn't already have an equivalent.

[josef-pkt]

isn't this just linalg toeplitz

[rkern]

Yes, the most efficient and idiomatic way to implement this would be to use scipy.linalg.toeplitz(). But my first question is "why should we add this in the absence of any other tools for autoregressive models?" It seems like statsmodels already handles this capably (keeping the efficient vector form until the last moment before using toeplitz() to expand it when needed).

[nitaifingerhut]

Wasn’t familiar with statsmodel Will take a look Thanks!

…

Sent from my iPhone

On 10 Jul 2021, at 23:22, Robert Kern ***@***.***> wrote:  Yes, the most efficient and idiomatic way to implement this would be to use scipy.linalg.toeplitz(). But my first question is "why should we add this in the absence of any other tools for autoregressive models?" It seems like statsmodels already handles this capably (keeping the efficient vector form until the last moment before using toeplitz() to expand it when needed). — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.