Theorem 3.34 — ratio test
DisprovedRudin.ch03_ratio_testanalysisseries
The series converges if , and diverges if there is an such that for all the terms are nonzero and .
Preamble
import Mathlib import Definitions.Def_Rudin_ch03_series open Filter Topology
Formal statement
namespace Rudin
/-- Rudin, Theorem 3.34 (ratio test): `∑ aₙ` converges if `limsup ‖aₙ₊₁ / aₙ‖ < 1`, and
diverges if there is an `N` with `‖aₙ₊₁‖ ≥ ‖aₙ‖` for all `n ≥ N` and `aₙ ≠ 0` there. -/
theorem ch03_ratio_test (a : ℕ → ℂ) :
(limsup (fun n => ((‖a (n + 1)‖ / ‖a n‖ : ℝ) : EReal)) atTop < 1 → SeriesConverges a) ∧
((∃ N, ∀ n ≥ N, a n ≠ 0 ∧ ‖a n‖ ≤ ‖a (n + 1)‖) → ¬ SeriesConverges a) := by sorry
end RudinSource
Walter Rudin, Principles of Mathematical Analysis, 3rd edition, McGraw-Hill, 1976, Chapter 3, p. 66, Theorem 3.34
Read-back
What the Lean code literally says, in plain math · Aristotle (Harmonic)
Let . Two independent implications are asserted:
- Convergence. If
(the quotients viewed in the extended reals), then the series converges. Note that division is total: whenever the quotient is taken to be , not undefined.
- Divergence. If there exists such that for every both and , then the series does not converge. The monotonicity here is non-strict, and both conditions are required simultaneously from the index onwards.
The two parts share only the sequence ; neither hypothesis is assumed in the other.
Human review
Confirmed by the mission captain (proposal self-audit).
The convergence half is false as stated because Lean's division by zero gives ratio 0 whenever a n = 0. Take a n = 1 for odd n and 0 for even n: every ratio is 0, so the limsup is 0 < 1, but the terms do not tend to 0 and the series diverges. The row is already marked Disproved on the platform. Please add the hypothesis that a n is nonzero (at least for all large n), which is what Rudin's ratio presupposes.