. 2019 Mar 1;10:115. doi: 10.3389/fphys.2019.00115

Table 1.

Largest relative error in F_q(n) estimates induced by calculating the sum of n products with the fast Equation (15).

	Time series
	{wn_i}; {Bm_i}; {wb_i};			{bcn_i};
Length	DFA₁	DFA₂	Length	DFA₁	DFA₂
N = 10²	1.4E-11	1.9E-11	N = 2⁸	3.1E-11	2.1E-11
N = 10³	1.0E-09	2.3E-08	N = 2¹⁰	4.8E-10	1.2E-10
N = 10⁴	1.0E-07	5.0E-07	N = 2¹²	2.6E-09	3.9E-09
N = 10⁵	1.9E-05	6.1E-06	N = 2¹⁴	1.1E-06	4.0E-07
N = 10⁶	6.0E-03	4.4E-04	N = 2¹⁶	4.6E-06	2.2E-06

{wn_i}, white noise; {Bm_i}, Brownian motion; {wb_i}, sum of {wn_i} and {Bm_i}; {bcn_i}, stochastic binomial cascade embedded in noise (see text for details).