绝对误差限 & 导数
1. 结论:
对函数 f ( x ) f(x) f(x)在 x ′ x' x′处使用泰勒展开: f ( x ) = f ( x ′ ) + f ′ ( x ′ ) ( x − x ′ ) + o ( x − x ′ ) f(x) = f(x') + f'(x')(x-x') + o(x-x') f(x)=f(x′)+f′(x′)(x−x′)+o(x−x′)
于是: ∣ f ( x ) − f ( x ′ ) ∣ = ∣ f ′ ( x ′ ) ( x − x ′ ) + o ( x − x ′ ) ∣ ≤ ∣ f ′ ( x ′ ) ∣ ⋅ ϵ ( x ) + o ( ϵ ( x ) ) |f(x) - f(x')| = |f'(x')(x-x') + o(x-x')| \leq |f'(x')|\cdot \epsilon(x) + o(\epsilon(x)) ∣f(x)−f(x′)∣=∣f′(x′)(x−x′)+o(x−x