News for hackers. Stuff that matters.
Re:The Mystery of 355/113 by cperciva on Tuesday July 24, 2012 @10:26AM (#4285801)
FWIW, all of the convergents from the continued fraction expansion of a number have "quality" greater than 2. Conversely, the Thue–Siegel–Roth theorem says that any algebraic number and any epsilon > 0, the algebraic number has only finitely many approximations with quality more than 2 + epsilon; so the "quality" values of successive best approximations are a series which converges to 2 from above.

This was how the first known transcendental numbers -- the Liouville numbers -- were constructed: They have approximations with unbounded quality, thus they cannot be algebraic. In practice, however, this isn't a very useful method: "Almost all" transcendental numbers also obey the Thue–Siegel–Roth theorem.

Re:The Mystery of 355/113 by michaelf on Tuesday July 24, 2012 @11:36PM (#4289239)
Is it also the case that all transcendental numbers have approximations of unbounded quality? Or only that Liouville numbers have approximations of unbounded quality, and therefore cannot be algebraic?
Re:The Mystery of 355/113 by cperciva on Wednesday July 25, 2012 @02:23AM (#4289659)
Liouville numbers have approximations of unbounded quality and thus cannot be algebraic. "Most" transcendental numbers only have finitely many approximations of quality higher than 2 + epsilon for any epsilon.