This paper explores the intersection of combinatorics on words, number theory, and formal power series. The authors bridge the gap between finite fields—where certain sequences are algebraic—and the field of rational numbers, where those same sequences generate complex, transcendental objects. Specifically, they describe the continued fraction expansion of a transcendental power series over $\mathbb{Q}$ whose coefficients are derived from a sequence of 1s and 2s.
The Problem: Irrationality in Power Series Fields
In number theory, continued fractions represent numbers or power series as a sequence of quotients. For real numbers, Roth’s Theorem establishes that any algebraic irrational number has an irrationality measure of 2, meaning it cannot be "too well" approximated by rational numbers.
However, in the field of formal power series, the rules change based on the characteristic of the field. In fields with positive characteristic (like $\mathbb{F}_p$), algebraic elements can have irrationality measures much larger than 2. This research investigates what happens when a sequence known to be algebraic in a finite field is "lifted" to the field of rational numbers $\mathbb{Q}$. The authors focus on a sequence introduced by David Robbins thirty years ago, which satisfies a quartic equation in $\mathbb{F}_3$, and analyze its structure and approximation properties over $\mathbb{Q}$.
Methodology and Construction
The research employs theoretical techniques from combinatorics and analysis:
- Recursive Word Construction: The authors define an infinite word $W$ over the alphabet $\{1, 2\}$ using the recursive rule: $W_n = W_{n-1}, 2, W_{n-2}, 2, W_{n-1}$.
- Power Series Mapping: They define a series $\theta = \sum w(n)T^{-n}$, treating the sequence as coefficients in the field $\mathbb{Q}((T^{-1}))$.
- Rational Approximation: To determine the continued fraction, the authors construct a sequence of rational approximations ($R_n/S_n$) and analyze the combinatorial structure of the word $W$ to prove their accuracy.
- Degree Analysis: By calculating the degrees of the partial quotients ($a_n$) in the continued fraction expansion, they demonstrate that the errors in their rational approximations match the expected behavior of a continued fraction.
Key Insights
The $\mathbb{F}_3$ Connection
The sequence of 1s and 2s is not arbitrary; it represents the coefficients of a root of the quartic equation $x^4 + x^2 - Tx + 1 = 0$ in the finite field $\mathbb{F}_3$. While the sequence is algebraic in the finite field world, the authors prove it becomes transcendental when viewed over the rational numbers.
Predictable Irregularity
The degrees of the partial quotients in the continued fraction follow a specific, repeating pattern of length four: $(d_{4n+1}, 1, d_{4n+3}, 1)$. This structured growth allows for the precise calculation of the irrationality measure.
Irrationality Measure of 3
A significant result of this work is the proof that the irrationality measure $\nu(\theta)$ is exactly 3. This indicates that the series is significantly more "approximable" by rational functions than a typical algebraic power series over a field of characteristic zero.
Results and Future Directions
- Theorem 3: Provides an explicit formula for the degrees of the partial quotients $a_n$ of the series $\theta$, showing they follow a recursive growth pattern linked to the lengths of the words $W_n$.
- Conjecture 1: While the degrees are proven, the exact polynomial forms (coefficients) of the quotients remain a conjecture, suggesting a deep underlying symmetry yet to be fully mapped.
The authors note that these results depend heavily on the specific choice of the pair $(1, 2)$. If other symbols were used, the "coprimality" of the rational approximations might fail, breaking the regularity of the continued fraction. Future work may involve determining if other "automatic" sequences from finite fields yield similarly structured continued fractions when mapped to $\mathbb{Q}$.
Summary updated