好特么多公式啊

最新推荐文章于 2024-11-17 00:29:16 发布

Miykael_xxm

最新推荐文章于 2024-11-17 00:29:16 发布

阅读量218

点赞数

分类专栏：技术

技术专栏收录该内容

38 篇文章

订阅专栏

本文推广了多元函数带Peano余项的Taylor公式，并详细给出了其数学推导过程。适用于研究多元函数微分特性的高级数学场景。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

转载博客
http://blog.csdn.net/phoenix198425/article/details/78659060

多元函数带 Peano余项的Taylor公式的推广

设：
$m, n \in \mathbb N, m, n \ge 1,$
${\operatorname{D}}_j = \frac {\partial }{\partial x_j},$
${{\operatorname{D}}_j}^n = (\frac {\partial }{\partial x_j})^n = \frac {\partial^n }{\partial x_j^n},$
${{\operatorname{D}}_i}^m {{\operatorname{D}}_j}^n = (\frac {\partial }{\partial x_i})^m (\frac {\partial }{\partial x_j})^n= \frac {\partial^{m + n} }{\partial x_i^m \partial x_j^n},$
若 $f ( \mathbf{x} ) : \mathbb R^n \rightarrow \mathbb R, n \in \mathbb N, n \ge 1$ 在点 $\mathbf{x}\ k$ 阶可微， $k \in \mathbb N, k \ge 1,$ 在点 $\mathbf{x}$ 的某一邻域 $U$ 上有定义，则 $\forall (\mathbf{x} + \Delta \mathbf{x}) \in U,$
$f ( \mathbf{x} + \Delta \mathbf{x} ) - f ( \mathbf{x} ) = \sum_{i = 1}^{k} \dfrac{1}{i!} (\sum _{j = 1}^{n} (\Delta x_j {\operatorname{D}}_{j} ))^i f(\mathbf{x}) + o({[r(\Delta \mathbf{x})]}^{k}), r(\Delta \mathbf{x}) = ||\Delta \mathbf{x} ||$

证明：

(1) 令：

R k (Δ x) = f (x + Δ x) - f (x) - \sum i = 1 k 1 i ! (\sum j = 1 n (Δ x j D j)) i f (x),

$R_k(\Delta \mathbf{x}) = f ( \mathbf{x} + \Delta \mathbf{x} ) - f ( \mathbf{x} ) - \sum_{i = 1}^{k} \dfrac{1}{i!} (\sum _{j = 1}^{n} (\Delta x_j {\operatorname{D}}_{j} ))^i f(\mathbf{x}),$

h (t, m, f, Δ x) = 1 m ! (\sum j = 1 n (Δ x j D j)) m f (x)

$h(t, m, f, \Delta \mathbf{x}) = \dfrac{1}{m!} (\sum _{j = 1}^{n} (\Delta x_j {\operatorname{D}}_{j} ))^m f(\mathbf{x})$

= 1 m ! \sum i = 0 m C i m (Δ x t) i (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m - i) f (x),

$= \frac{1}{m!} \sum _{i = 0} ^{m} C_m ^i {(\Delta x_t)} ^ i {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m - i)} f(\mathbf{x}),$

t, m \in ℕ, 1 \leq t \leq n, 1 \leq m \leq k,

$t, m \in \mathbb N, 1 \le t \le n, 1 \le m \le k,$

则：

\partial \partial Δ x t h (t, m + 1, f, Δ x)

$\frac {\partial } {\partial \Delta x_t} h(t, m + 1, f, \Delta \mathbf{x})$

= \partial \partial Δ x t 1 ( m + 1 ) ! \sum i = 0 m + 1 C i m + 1 (Δ x t) i (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m + 1 - i) f (x)

$= \frac {\partial } {\partial \Delta x_t} \dfrac{1}{(m + 1)!} \sum _{i = 0} ^{m + 1} C_{m + 1} ^i{(\Delta x_t)} ^ i {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m + 1 - i)} f(\mathbf{x})$

= 1 ( m + 1 ) ! \sum i = 1 m + 1 C i m + 1 i (Δ x t) (i - 1) (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m + 1 - i) f (x)

$= \dfrac{1}{(m + 1)!} \sum _{i = 1} ^{m + 1} C_{m + 1} ^i i {(\Delta x_t)} ^ {(i - 1)} {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m + 1 - i)} f(\mathbf{x})$

= 1 m ! \sum i = 1 m + 1 1 m + 1 ( m + 1 ) ! i ! ( m + 1 - i ) ! i (Δ x t) (i - 1) (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m + 1 - i) f (x)

$= \dfrac{1}{m!} \sum _{i = 1} ^{m + 1} \frac {1} {m + 1} \frac {(m + 1)!}{i! (m + 1 - i)!} i {(\Delta x_t)} ^ {(i - 1)} {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m + 1 - i)} f(\mathbf{x})$

= 1 m ! \sum i = 1 m + 1 m ! ( i - 1 ) ! ( m + 1 - i ) ! (Δ x t) (i - 1) (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m + 1 - i) f (x)

$= \dfrac{1}{m!} \sum _{i = 1} ^{m + 1} \frac {m!}{(i - 1)! (m + 1 - i)!} {(\Delta x_t)} ^ {(i - 1)} {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m + 1 - i)} f(\mathbf{x})$

= 1 m ! \sum i = 1 m + 1 C i - 1 m (Δ x t) (i - 1) (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m + 1 - i) f (x)

$=\dfrac{1}{m!} \sum _{i = 1} ^{m + 1} C_{m} ^{i - 1} {(\Delta x_t)} ^ {(i - 1)} {({\operatorname{D}}_{t} )} ^ i (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m + 1 - i)} f(\mathbf{x})$

= 1 m ! \sum i = 0 m C i m (Δ x t) i (D t) (i + 1) (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m - i) f (x)

$=\dfrac{1}{m!} \sum _{i = 0} ^{m} C_{m} ^{i} {(\Delta x_t)} ^ {i} {({\operatorname{D}}_{t} )} ^ {(i + 1)} (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m - i)} f(\mathbf{x})$

= 1 m ! \sum i = 0 m C i m (Δ x t) i (D t) i (\sum 1 \leq j \leq n j \neq t (Δ x j D j)) (m - i) D t f (x)

$=\dfrac{1}{m!} \sum _{i = 0} ^{m} C_{m} ^{i} {(\Delta x_t)} ^ {i} {({\operatorname{D}}_{t} )} ^ {i} (\sum _{1 \le j \le n} _{j \neq t} (\Delta x_j {\operatorname{D}}_{j} ))^{(m - i)} {\operatorname{D}}_{t} f(\mathbf{x})$

= h (t, m, D t f, Δ x)

$= h(t, m, {\operatorname{D}}_{t} f, \Delta \mathbf{x})$

且：

R k (Δ x) = f (x + Δ x) - f (x) - \sum i = 1 k h (t, i, f, Δ x)

$R_k(\Delta \mathbf{x}) = f ( \mathbf{x} + \Delta \mathbf{x} ) - f ( \mathbf{x} ) - \sum_{i = 1}^{k} h(t, i, f, \Delta \mathbf{x})$
于是

k≥2 $k \ge 2$ 时：

\partial \partial Δ x t R k (Δ x)

$\frac {\partial } {\partial \Delta x_t} R_k(\Delta \mathbf{x})$

= \partial \partial x t f (x + Δ x) - D t f (x) - \sum i = 2 k h (t, i - 1, D t f, Δ x)

$= \frac {\partial } {\partial x_t} f ( \mathbf{x} + \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 2}^{k} h(t, i - 1, {\operatorname{D}}_{t} f, \Delta \mathbf{x})$

= D t f (x + Δ x) - D t f (x) - \sum i = 1 k - 1 h (t, i, D t f, Δ x)

$= {\operatorname{D}}_{t} f ( \mathbf{x} + \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k - 1} h(t, i, {\operatorname{D}}_{t} f, \Delta \mathbf{x})$

因此 $k \ge 2$ 时：

R' k (θ Δ x) Δ x = \sum t = 1 n [D t f (x + θ Δ x) - D t f (x) - \sum i = 1 k - 1 h (t, i, D t f, θ Δ x)] Δ x t

$R_k'(\theta \Delta \mathbf{x}) \Delta \mathbf{x} = \sum _{t = 1} ^{n} [{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k - 1} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})] \Delta x_t$

(2) $k \ge 2$ 时：

d [r (Δ x)] k = k r k - 1 1 r (Δ x) ⊺ d Δ x

$\operatorname{d} [r(\Delta \mathbf{x})]^{k} = k r^{k - 1} \frac {1}{r} {(\Delta \mathbf{x})}^\intercal \operatorname{d} \Delta \mathbf{x}$

\Rightarrow {[r (θ Δ x)] k}' = k (r (θ Δ x)) k - 1 1 r ( θ Δ x ) (θ Δ x) ⊺)

$\Rightarrow \{[r(\theta \Delta \mathbf{x})]^{k}\}' = k {(r(\theta \Delta \mathbf{x}))}^{k - 1} \frac {1}{r(\theta \Delta \mathbf{x})} {(\theta \Delta \mathbf{x})}^\intercal )$

\Rightarrow {[r (θ Δ x)] k}' Δ x = k (r (θ Δ x)) k - 1 1 r ( θ Δ x ) (θ Δ x) ⊺ Δ x = k (r (θ Δ x)) k - 1 r (Δ x)

$\Rightarrow \{[r(\theta \Delta \mathbf{x})]^{k}\}' \Delta \mathbf{x} = k {(r(\theta \Delta \mathbf{x}))}^{k - 1} \frac {1}{r(\theta \Delta \mathbf{x})} {(\theta \Delta \mathbf{x})}^\intercal \Delta \mathbf{x} = k {(r(\theta \Delta \mathbf{x}))}^{k - 1} r(\Delta \mathbf{x} )$

(3) 易知：

R k (0 ⃗) = 0, r (0 ⃗) = 0

$R_k (\vec{0}) = 0, r(\vec{0}) = 0$
下面用归纳法证明命题。
1)

k=1 $k = 1$ 时显然成立。
2) 设

k $k$ 时成立，即：

limΔx→0⃗ Rk(Δx)[r(Δx)]k $\lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {R_k (\Delta \mathbf{x})} {{[r(\Delta \mathbf{x})]}^k}$

=limΔx→0⃗ f(x+Δx)−f(x)−∑ki=1h(t,i,f,Δx)[r(Δx)]k $= \lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac { f ( \mathbf{x} + \Delta \mathbf{x} ) - f ( \mathbf{x} ) - \sum_{i = 1}^{k} h(t, i, f, \Delta \mathbf{x}) } {{[r(\Delta \mathbf{x})]}^k}$

=0 $= 0$
则

k+1 $k + 1$ 时，由微分中值定理的推广，

$\lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {R_{k + 1} (\Delta \mathbf{x})} {{[r(\Delta \mathbf{x})]}^{k + 1}}$
$= \lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {R_{k + 1} '(\theta \Delta \mathbf{x}) \Delta \mathbf{x}} { \{[r(\theta \Delta \mathbf{x})]^{{k + 1}}\}' \Delta \mathbf{x} }$
$= \lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {\sum _{t = 1} ^{n} [{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})] \Delta x_t} { (k + 1) {(r(\theta \Delta \mathbf{x}))}^{k} r(\Delta \mathbf{x} )}$
$= \sum _{t = 1} ^{n} \lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})}{{(r(\theta \Delta \mathbf{x}))}^{k} } \cdot \frac {1}{k + 1} \cdot \frac {\Delta x_t}{ r(\Delta \mathbf{x} )}$

由： $\lim \limits_{\Delta \mathbf{x} \to \vec{0}} \theta \Delta \mathbf{x} = 0$
以及： $\lim \limits_{\theta \Delta \mathbf{x} \to \vec{0}} \frac {{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})}{{(r(\theta \Delta \mathbf{x}))}^{k} } = 0$
可得： $\lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})}{{(r(\theta \Delta \mathbf{x}))}^{k} } = 0$

又： $\forall t \in \mathbb N, 1 \le t \le n, |\frac {\Delta x_t}{ r(\Delta \mathbf{x} )}| \le 1$
因此：
$\sum _{t = 1} ^{n} \lim \limits_{\Delta \mathbf{x} \to \vec{0}} \frac {{\operatorname{D}}_{t} f ( \mathbf{x} + \theta \Delta \mathbf{x} ) - {\operatorname{D}}_{t} f(\mathbf{x}) - \sum_{i = 1}^{k} h(t, i, {\operatorname{D}}_{t} f, \theta \Delta \mathbf{x})}{{(r(\theta \Delta \mathbf{x}))}^{k} } \cdot \frac {1}{k + 1} \cdot \frac {\Delta x_t}{ r(\Delta \mathbf{x} )} = 0$