Graphical illustration of the sequence identification procedure that is used to calculate the sample entropy for a simulated light curve. We set r = 0.1 × σ_ligi,_{t curve} and m = 2 in this example. The dashed horizontal lines represent x₀ ± r, x\ ± r and x₂ ± r. The points that fall within this boundary, namely d[x_i,x_j] < r, are color-coded in, respectively, purple, blue and green. The first unique sequence or template vector of two components (m = 2) is u_1,m = (x₀,x₁). Two other sequences match this template vector in the light curve, namely the sequences (x₃₀,x₃₁) and (x₆₀,x₆₁). Extending u₁ to three components (m + 1) then gives (x₀,x₁,x₂), occurring only once more in the light curve at (x₃₀, x₃₁, x₃₂), as x₆₂ does not fall within the tolerance margin r for the extension of (x₆₀, x₆₁ ) to m + 1. There are thus three matches of length ra for the template vector u₁ and two matches for its extension to length m + 1. This procedure is then repeated for all other two component template vectors (u_i,m+1 = (x₁,x₂),…,(x_N-2, x_N-1)) and their three component extensions (u_i,m+1 = (x₁,x₂,x_з),…, (x_n-2, x_n-1, x_n))