For stationary AR(1) segments of length 2, 3 and 4, the variance of the number of sign changes is computed exactly; the independent interval approximation matches exactly for n=2,3 but deviates slightly for n=4, especially for negative correlation.
Zero-Crossing Statistics for Non-Markovian Time Series
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
In applications spaning from image analysis and speech recognition, to energy dissipation in turbulence and time-to failure of fatigued materials, researchers and engineers want to calculate how often a stochastic observable crosses a specific level, such as zero. At first glance this problem looks simple, but it is in fact theoretically very challenging. And therefore, few exact results exist. One exception is the celebrated Rice formula that gives the mean number of zero-crossings in a fixed time interval of a zero-mean Gaussian stationary processes. In this study we use the so-called Independent Interval Approximation to go beyond Rice's result and derive analytic expressions for all higher-order zero-crossing cumulants and moments. Our results agrees well with simulations for the non-Markovian autoregressive model.
citation-role summary
citation-polarity summary
fields
math.HO 1years
2019 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Number of Sign Changes: Segment of AR(1)
For stationary AR(1) segments of length 2, 3 and 4, the variance of the number of sign changes is computed exactly; the independent interval approximation matches exactly for n=2,3 but deviates slightly for n=4, especially for negative correlation.