Algorithm 2: Embedding dimension determination
	Input: one-dimensional time series S = {xi,…, xn}, time delay τ;    max embedding dimension M, threshold ε.
	Output: embedding dimension m*
1	$\mathrm{for} k=1: n-m\tau$
2	for m = 1: M
3	E(m) = 0; E*(m) = 0; m* = 0;
4	for i = 1: n − mτ
5	$n\left(i,m\right)=\underset{j}{arg}\min X_i^{\left(m\right)}-X_j^{\left(m\right)}{}_2$
6	$a\left(i,m\right)=\frac{X_i^{\left(m+1\right)}-X_{n\left(i,m\right)}^{\left(m+1\right)}{}_{\infty }}{X_i^{\left(m\right)}-X_{n\left(i,m\right)}^{\left(m\right)}{}_{\infty }}$
7	E(m) = E(m) + α(i, m)/(n − mτ)
8	E* (m) = E* (m) + |x(i + mτ) − x(n(i,m) + mτ)|/(n − mτ)
9	end for
10	end for
11	E1(1) = E(2)/E(1)
12	E2(1) = E*(2)/E*(1)
13	for m = 2: M − 1
14	E1(m) = E(m + 1)/E(m)
15	E2(m) = E*(m + 1)/E*(m)
16	if (|E1(m) − E1(m − 1)| < ε) and (|E2(m) − 1| > ε)
17	m* = m
18	end if
19	end for