>>575 補足

原本PDFを見て貰った方が視認性は良いが、後の検索性のためにコピペする（＾＾
http://www.unirioja.es/cu/jvarona/downloads/Differentiability-DA-Roth.pdf
DIFFERENTIABILITY OF A PATHOLOGICAL FUNCTION, DIOPHANTINE APPROXIMATION, AND A REFORMULATION OF THE THUE-SIEGEL-ROTH THEOREM JUAN LUIS VARONA 2009
(抜粋)
P7
4. The theorem of Thue-Siegel-Roth revisited

Or, equivalently, if x is an irrational algebraic number, there exists a positive constant C(x, α) such
that ｜x - p/q ｜< C(x, α)/q^(2+α) (10)
has no rational solution.

P8
Remark 3. We have proved Theorem 3 by using the Thue-Siegel-Roth theorem.
But we have said that it is a reformulation. So, let us see how to
deduce the Thue-Siegel-Roth theorem from Theorem 3.
Given x algebraic and irrational, and ν > 2, Theorem 3 ensures that fν
is differentiable at x, so there exists
lim y→x {fν(y) - fν(x)}/(y - x) = f’ν (x).
By approximating y → x by irrationals y, it follows that f’ν (x) = 0.
Consequently, by approximating y → x by rationals, i.e., y = p/q, we also must have
lim p/q→x ｛fν(p/q) - fν(x)｝/(p/q - x ) = lim p/q→x (1/qν)/(p/q - x) = 0.
Then, for every ε > 0, there exists δ > 0 such that
1/(q^ν) <= ε｜p/q - x｜
when p/q ∈ (x - δ, x + δ). From here, it is easy to check that the same
happens for every p/q ∈ Q, perhaps with a greather constant ε' in the place
of ε. Thus, (10) with α = ν-2 and some positive constant C(x, α) = 1/ε' has
no rational solution, and we have obtained the Thue-Siegel-Roth theorem.
(引用終り)

つづく