Regular paperfolding sequence

In mathematics the regular paperfolding sequence, also known as the dragon curve sequence, is an infinite automatic sequence of 0s and 1s defined as the limit of the following process:

1

1 1 0

1 1 0 1 1 0 0

1 1 0 1 1 0 0 1 1 1 0 0 1 0 0

At each stage an alternating sequence of 1s and 0s is inserted between the terms of the previous sequence. The sequence takes its name from the fact that it represents the sequence of left and right folds along a strip of paper that is folded repeatedly in half in the same direction. If each fold is then opened out to create a right-angled corner, the resulting shape approaches the dragon curve fractal.^[1] For instance the following curve is given by folding a strip four times to the right and then unfolding to give right angles, this gives the first 15 terms of the sequence when 1 represents a right turn and 0 represents a left turn.

Starting at n = 1, the first few terms of the regular paperfolding sequence are:

1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, ... (sequence A014577 in the OEIS)

Properties[]

The value of any given term t_n in the regular paperfolding sequence can be found recursively as follows. If n = m·2^k where m is odd then

${\displaystyle t_{n}={\begin{cases}1&{\text{if }}m\equiv 1\mod 4\\0&{\text{if }}m\equiv 3\mod 4\end{cases}}}$

Thus t₁₂ = t₃ = 0 but t₁₃ = 1.

The paperfolding word 1101100111001001..., which is created by concatenating the terms of the regular paperfolding sequence, is a fixed point of the morphism or string substitution rules

11 → 1101

01 → 1001

10 → 1100

00 → 1000

as follows:

11 → 1101 → 11011001 → 1101100111001001 → 11011001110010011101100011001001 ...

It can be seen from the morphism rules that the paperfolding word contains at most three consecutive 0s and at most three consecutive 1s.

The paperfolding sequence also satisfies the symmetry relation:

${\displaystyle t_{n}={\begin{cases}1&{\text{if }}n=2^{k}\\1-t_{2^{k}-n}&{\text{if }}2^{k-1}<n<2^{k}\end{cases}}}$

which shows that the paperfolding word can be constructed as the limit of another iterated process as follows:

1

1 1 0

110 1 100

1101100 1 1100100

110110011100100 1 110110001100100

In each iteration of this process, a 1 is placed at the end of the previous iteration's string, then this string is repeated in reverse order, replacing 0 by 1 and vice versa.

Generating function[]

The generating function of the paperfolding sequence is given by

${\displaystyle G(t_{n};x)=\sum _{n=1}^{\infty }t_{n}x^{n}\,.}$

From the construction of the paperfolding sequence, it can be seen that G satisfies the functional relation

${\displaystyle G(t_{n};x)=G(t_{n};x^{2})+\sum _{n=0}^{\infty }x^{4n+1}=G(t_{n};x^{2})+{\frac {x}{1-x^{4}}}\,.}$

Paperfolding constant[]

Substituting x = 0.5 into the generating function gives a real number between 0 and 1 whose binary expansion is the paperfolding word

${\displaystyle G(t_{n};{\frac {1}{2}})=\sum _{n=1}^{\infty }{\frac {t_{n}}{2^{n}}}}$

This number is known as the paperfolding constant^[2] and has the value

${\displaystyle \sum _{k=0}^{\infty }{\frac {8^{2^{k}}}{2^{2^{k+2}}-1}}=0.85073618820186...}$ (sequence A143347 in the OEIS)

General paperfolding sequence[]

The regular paperfolding sequence corresponds to folding a strip of paper consistently in the same direction. If we allow the direction of the fold to vary at each step we obtain a more general class of sequences. Given a binary sequence (f_i), we can define a general paperfolding sequence with folding instructions (f_i).

For a binary word w, let w^‡ denote the reverse of the complement of w. Define an operator F_a as

${\displaystyle F_{a}:w\mapsto waw^{\ddagger }\ }$

and then define a sequence of words depending on the (f_i) by w₀ = ε,

${\displaystyle w_{n}=F_{f_{1}}(F_{f_{2}}(\cdots F_{f_{n}}(\varepsilon )\cdots ))\ .}$

The limit w of the sequence w_n is a paperfolding sequence. The regular paperfolding sequence corresponds to the folding sequence f_i = 1 for all i.

If n = m·2^k where m is odd then

${\displaystyle t_{n}={\begin{cases}f_{j}&{\text{if }}m\equiv 1\mod 4\\1-f_{j}&{\text{if }}m\equiv 3\mod 4\end{cases}}}$

which may be used as a definition of a paperfolding sequence.^[3]

Properties[]

• A paperfolding sequence is not ultimately periodic.^[3]
• A paperfolding sequence is 2-automatic if and only if the folding sequence is ultimately periodic (1-automatic).

References[]