Department of Mathematical Science¹
Computational Science Program²
The University of Texas at El Paso
El Paso, Texas
Department of Data Science³
Ramapo College of New Jersey
Mahwah, New Jersey

Analysis of Simulated Stationary Processes Using the CDFA

The Hurst exponent measures a time series’ tendency to regress substantially to the mean or cluster in a certain direction also known as long-memory behavior. The Detrended Fluctuation Analysis (DFA) method sometimes overestimates the Hurst exponent. To address this issue, the Cantor DFA (CDFA) was developed which is based on Cantor set theory. In this paper, we implement the CDFA method to simulated stationary processes to analyze their long-memory behavior. Computation time is also examined from a parallel optimization scheme for the CDFA. Experimental results show that computation time is cut in half when we use two processors instead of one. However, as the number of processors increases, computation time reduces at a decreasing rate until the law of diminishing returns set in.

View Paper