Differentiability at a point Let X be a subset of Rn, and let x0∈X be an element of X which is also a limit point of X. Let f:X→Rm be a function. Let L be a linear transformation from Rn to Rm. We say that f is differentiable atx0 on X with derivativeL and write f′(x0):=L if
Uniqueness of derivatives Let E be a subset of Rn, f:E→Rm be a function, x0∈E be an interior point of E, and let L1:Rn→Rm and L2:Rn→Rm be linear transformations. Suppose that f is differentiable at x0 with derivative L1, and also differentiable at x0 with derivative L2. Then L1=L2.
proof:
because L1=L2, so there exists v∈Rn∧v=0, s.t. L1(v)=L2(v). R1(t):=f(x0+tv)−f(x0)−L1(tv). R2(t):=f(x0+tv)−f(x0)−L2(tv). t(L1(v)−L2(v))=L1(tv)−L2(tv)=R2(t)−R1(t). ∣∣v∣∣∣∣(L1(v)−L2(v))∣∣=∣∣tv∣∣∣∣t(L1(v)−L2(v))∣∣=∣∣tv∣∣∣∣R2(t)−R1(t)∣∣≤∣∣tv∣∣∣∣R2(t)∣∣+∣∣tv∣∣∣∣R1(t)∣∣.
so L1(v)−L2(v)=0, contradiction.
梯度:设 X⊆Rn,f:X→R 是一个定义在 X 上的实值函数(标量场)。若 f 在内点 x0∈X 处可微,存在唯一的线性变换 L:Rn→R(即 f′(x0))。由于 L 是从 Rn 到 R 的线性映射,根据里斯表示定理 (Riesz Representation Theorem),在给定内积 ⟨⋅,⋅⟩ 的欧几里得空间中,必然存在唯一的向量 v∈Rn,使得对于任意 h∈Rn,都有:L(h)=⟨v,h⟩这个唯一的向量 v 就称为 f 在 x0 处的梯度,记作 ∇f(x0) 或 grad f(x0)。
方向导数:设 u 为单位向量(∥u∥=1),f 在 x0 处沿方向 u 的方向导数定义为:Duf(x0)=limt→0tf(x0+tu)−f(x0)将全微分公式代入上式(令 h=tu):Duf(x0)=limt→0tf(x0)+L(tu)+o(∥tu∥)−f(x0)由于 L 是线性变换,L(tu)=tL(u),且 ∥tu∥=∣t∣∥u∥=∣t∣:Duf(x0)=limt→0ttL(u)+o(t)=L(u)
3. Properties
可微蕴含连续. 设 X⊆Rn,x0∈X 是 X 的极限点,f:X→Rm 在 x0 可微。则 f 在 x0 连续。
Let F be a field, ⟨F,Fn⟩=:Fn, f:Fn→Fm be a linear mapping. Then ∃L∈Fm×n, s.t. f(x)=Lx. Furthermore, L is unique.
设 f,g:[a,b]→R 在 x∈[a,b] 可微。则 f+g 与 fg 在 x 可微。若 g(x)=0,则 f/g 在 x 可微。并且
(f+g)′(x)=f′(x)+g′(x)
(fg)′(x)=f′(x)g(x)+f(x)g′(x)
(gf)′(x)=g(x)2g(x)f′(x)−g′(x)f(x)
Let f:Rn→Rm be a function, and f is differentiable at x0 on X. g:Rm→Rk be a function, and g is differentiable at f(x0) on f(X). Then g∘f is differentiable at x0 on X, and (g∘f)′(x0)=g′(f(x0))∘f′(x0).