LongMemorySVResult.at_boundary#

property LongMemorySVResult.at_boundary: bool[source]#

Whether d sits at the stationarity boundary of the search.

The Whittle criterion is maximized over |d| < 0.5; an estimate within 0.01 of that edge means the data prefer a nonstationary log variance, and d should be read as a lower bound rather than a point.

Example

A random-walk log variance pushes d to the edge:

>>> import numpy as np
>>> rng = np.random.default_rng(0)
>>> h = np.cumsum(0.1 * rng.standard_normal(1500))
>>> y = np.exp(h / 2) * rng.standard_normal(1500)
>>> res = LongMemorySV(y, mean="zero").fit()
>>> round(res.d, 3), res.at_boundary
(0.499, True)