Σ 1/n² 收敛——p = 2 的 p-级数,和 ≈ 1.6449 p-级数对比值与根值判别法都失效(L ≈ 1、ρ ≈ 1),但比较估计给出 p = −ln|a(10⁹)|/ln(10⁹) = 2 > 1,故级数收敛;部分和(到 10⁵)加 p 律尾部估计出和为 π²/6 ≈ 1.6449。Series convergence analysis — Σ aₙ with aₙ = 1/n^2
nth-term test: |a(10^9)| ≈ 1.00e-18 → 0 — passes
ratio test: L = |a(10^6+1)|/|a(10^6)| ≈ 0.999998 — inconclusive (L ≈ 1)
root test: ρ = |a(10^9)|^(1/10^9) ≈ 1 — inconclusive (ρ ≈ 1)
p-series comparison: p ≈ 2 — converges (p ≈ 2 > 1) (Σ 1/n^p converges for p > 1)
integral test: ∫₂^10^4 |a(x)|dx ≈ 0.4999 (+tail ≈ 0.0001) — decays fast — convergent
estimated sum ≈ 1.6449 (partial sum + p-law tail)
VERDICT: CONVERGES (p-series comparison: p ≈ 2 > 1)
Σ cos(πn)/n 条件收敛——Leibniz 交错判别法 cos(πn) = (−1)ⁿ,这正是交错调和级数:符号被识别为严格交错,且 |aₙ| = 1/n 单调趋于 0,满足 Leibniz 判别。绝对值级数表现为 p = 1(发散),故收敛是条件性的,部分和趋于 −ln 2 ≈ −0.6931。Series convergence analysis — Σ aₙ with aₙ = cos(pi*n)/n
Alternating series detected (signs strictly alternate); |aₙ| decreases toward 0 — Leibniz test applies
nth-term test: |a(10^9)| ≈ 1.00e-9 → 0 — passes
ratio test: L = |a(10^6+1)|/|a(10^6)| ≈ 0.999999 — inconclusive (L ≈ 1)
root test: ρ = |a(10^9)|^(1/10^9) ≈ 1 — inconclusive (ρ ≈ 1)
p-series comparison (on |aₙ|): p ≈ 1 — diverges (p ≈ 1 ≤ 1) (borderline harmonic-like; classified with the harmonic side (p = 1 diverges))
integral test: ∫₂^10^4 |a(x)|dx ≈ 5.3924 — each decade adds a comparable amount — grows without bound — divergent
estimated sum ≈ -0.6931 (Leibniz remainder bound |R| ≤ 1.0e-5)
VERDICT: CONVERGES CONDITIONALLY (p-series comparison: p ≈ 1 ≤ 1; Leibniz test; not absolutely)