广东高数专插本笔记--第八章 多元函数微分法及其应用

2023-07-13学习21 分钟331 次阅读
AI 摘要鱼粮不足,AI 今天先歇着 · 正文在下面

多元函数的基本概念

1、多元函数的概念

1.1、二元函数的定义

定义1 设有三个变量xx,yy和zz,如果对于变量xx,yy在非空平面点集DD内任意取定的每一对数值,变量zz按一定规律都有唯一确定的值与之对应,则称变量zz是变量xx,yy的二次函数,记为

z=f(x,y)或z=z(x,y),(x,y)∈D.\boxed{z=f(x,y)\text{或}z=z(x,y),(x,y)\in D}.

其中xx,yy称为自变量,zz称为因变量,自变量xx,yy在变化过程中的取值范围DD叫作函数zz的定义域。

对于确定的(x0,y0)(x_0,y_0),函数zz有唯一确定的值z0z_0与之对应,则称z0z_0为函数f(x,y)f(x,y)在点(x0,y0)(x_0,y_0)处的函数值,记为f(x0,y0)f(x_0,y_0)或z∣(x0,y0)z\bigg|_{(x_0,y_0)}。

例子1

求二元函数z=ln⁡(x+y)z=\ln(x+y)的定义域.

解:

由对数函数的定义域可知,(x+y)>0(x+y)>0,即二元函数的定义域为D={(x,y)∣x+y>0}D=\{(x,y)|x+y>0\}。

例子2

设f(x+y,x−y)=x2−y22xyf(x+y,x-y)=\dfrac{x^2-y^2}{2xy},则f(x,y)=___f(x,y)=\_\_\_.

方法一

f(x+y,x−y)=(x+y)(x−y)2[(x+y)2−(x−y2)]4=2(x+y)(x−y)(x+y)2−(x−y)2f(x+y,x-y)=\dfrac{(x+y)(x-y)}{\dfrac{2[(x+y)^2-(x-y^2)]}{4}}=\dfrac{2(x+y)(x-y)}{(x+y)^2-(x-y)^2},

所以f(x,y)=2xyx2−y2f(x,y)=\dfrac{2xy}{x^2-y^2}.

方法二

令u=x+y,v=x−yu=x+y,v=x-y,则x=u+v2,y=u−v2x=\dfrac{u+v}{2},y=\dfrac{u-v}{2},

所以f(u,v)=(u+v2)2−(u−v2)22⋅u+v2⋅u−v2=uvu2−v22=2uvu2−v2f(u,v)=\dfrac{(\dfrac{u+v}{2})^2-(\dfrac{u-v}{2})^2}{2\cdot \dfrac{u+v}{2} \cdot \dfrac{u-v}{2}}=\dfrac{uv}{\dfrac{u^2-v^2}{2}}=\dfrac{2uv}{u^2-v^2}.

故f(x,y)=2xyx2−y2f(x,y)=\dfrac{2xy}{x^2-y^2}.

⭐️⭐️⭐️ 注意:xy=14[(x+y)2−(x−y)2],x2+y2=12[(x+y)2+(x−y)2]xy=\dfrac{1}{4}[(x+y)^2-(x-y)^2],x^2+y^2=\dfrac{1}{2}[(x+y)^2+(x-y)^2]有助于快速求解二元函数表达式的问题.

1.2、多元函数的定义

定义2 若对于变量x1x_1,x2x_2,......,xnx_n在nn维空间点集DD内的每一组值,变量zz依照某一法则有唯一确定的值与之对应,则称zz为x1x_1,x2x_2,......,xnx_n的nn元函数,常记为z=f(x1,x2,...,xn)z=f(x_1,x_2,...,x_n)。

当n=1n=1时,nn元函数就是一元函数;当n≥2n\geq 2时,nn元函数统称为多元函数。

1.3、二元函数的几何意义

2、二元函数的极限与连续性

2.1、二元函数的极限

定义3 设函数z=f(x,y)z=f(x,y)在点P0(x0,y0)P_0(x_0,y_0)的某领域内有定义

注意:如果点P(x,y)P(x,y)沿不同路径趋于点P0(x0,y0)P_0(x_0,y_0)时,函数值趋于不同的值,就可以断定极限不存在。

例子1

求极限lim⁡x→0y→0xy3−xy+9\lim\limits_{x\to 0\atop y\to 0}\dfrac{xy}{3-\sqrt{xy+9}}.

解:

lim⁡x→0y→0xy3−xy+9=lim⁡x→0y→0xy(3+xy+9)9−xy−9=lim⁡x→0y→0(3+xy+9)=3+3=6.\lim\limits_{x\to 0\atop y\to 0}\dfrac{xy}{3-\sqrt{xy+9}}=\lim\limits_{x\to 0\atop y\to 0}\dfrac{xy(3+\sqrt{xy+9})}{9-xy-9}=\lim\limits_{x\to 0\atop y\to 0}(3+\sqrt{xy+9})=3+3=6.

例子2

求极限lim⁡x→0y→10sin⁡xyx.\lim\limits_{x\to 0\atop y\to 10}\dfrac{\sin xy}{x}.

解:

lim⁡x→0y→10sin⁡xyx=lim⁡x→0y→10sin⁡xyxy⋅y=lim⁡x→0y→10sin⁡xyxy⋅lim⁡x→0y→10y=1⋅10=10.\lim\limits_{x\to 0\atop y\to 10}\dfrac{\sin xy}{x}=\lim\limits_{x\to 0\atop y\to 10}\dfrac{\sin xy}{xy}\cdot y=\lim\limits_{x\to 0\atop y\to 10}\dfrac{\sin xy}{xy}\cdot\lim\limits_{x\to 0\atop y\to 10}y=1\cdot 10=10.(凑成重要极限的形式)

2.2、二元函数的连续性

定义4 设函数z=f(x,y)z=f(x,y)在点P0(x0,y0)P_0(x_0,y_0)的某领域内有定义,如果

lim⁡x→0y→0f(x,y)=f(x0,y0),\lim\limits_{x\to 0\atop y\to 0}f(x,y)=f(x_0,y_0),

则称二元函数z=f(x,y)z=f(x,y)在点P0(x0,y0)P_0(x_0,y_0)处连续。

例子1

极限lim⁡x→1y→0ln⁡(x+ey)x2+y2=___\lim\limits_{x\to 1\atop y\to 0}\dfrac{\ln(x+e^y)}{\sqrt{x^2+y^2}}=\_\_\_.

解:

原式=ln⁡(x+ey)x2+y2∣(1,0)=ln⁡21=ln⁡2\text{原式}=\dfrac{\ln(x+e^y)}{\sqrt{x^2+y^2}}\bigg|_{(1,0)}=\dfrac{\ln2}{1}=\ln2.

多元函数的偏导数与全微分

1、偏导数

1.1、一阶偏导数

1.2、高阶偏导数

2、全微分

2.1、全微分的概念

定义3 设函数z=f(x,y)z=f(x,y)在点(x,y)(x,y)的某领域内有定义,若函数在点(x,y)(x,y)处的全增量可

Δz=f(x+Δx,y+Δy)−f(x,y)\boxed{\Delta z=f(x+\Delta x,y+\Delta y)-f(x,y)}

表示为

Δz=AΔx+BΔy+o(ρ),\boxed{\Delta z=A\Delta x+B\Delta y+o(\rho)},

其中AA,BB不依赖于Δx\Delta x,Δy\Delta y,仅与xx,yy有关,ρ=(Δx)2+(Δy)2\rho =\sqrt{(\Delta x)^2+(\Delta y)^2},则称函数z=f(x,y)z=f(x,y)在点(x,y)(x,y)处可微分,而AΔx+BΔyA\Delta x+B \Delta y称为函数z=f(x,y)z=f(x,y)在点(x,y)(x,y)处的全微分,记作dzdz,即

dz=AΔx+BΔy.\boxed{dz=A\Delta x+B\Delta y}.

一般地,我们记自变量Δx=dx\Delta x=dx,Δy=dy\Delta y=dy,所以dz=AΔx+BΔy=Adx+Bdydz=A\Delta x+B\Delta y=Adx+Bdy。

如果函数z=f(x,y)z=f(x,y)在区域DD内各点处都可微分,则称该函数在DD内可微分。

定理2 (全微分存在的必要条件) 如果函数z=f(x,y)z=f(x,y)在点(x,y)(x,y)处可微分,则z=f(x,y)z=f(x,y)在点(x,y)(x,y)处的偏导数∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}必定存在,且函数z=f(x,y)z=f(x,y)在该点处的全微分为

dz=∂z∂xdx+∂z∂ydy.\boxed{dz=\dfrac{\partial z}{\partial x}dx+\dfrac{\partial z}{\partial y}dy}.

定理3 (全微分存在的充分条件) 如果函数z=f(x,y)z=f(x,y)的偏导数∂z∂x\dfrac{\partial z}{\partial x}、∂z∂y\dfrac{\partial z}{\partial y}在点(x,y)(x,y)处连续,则z=f(x,y)z=f(x,y)在点(x,y)(x,y)处可微分。

二元函数全微分概念可以推广到一般地nn元函数(n≥2)(n\geq 2),例如,设三元函数u=f(x,y,z)u=f(x,y,z)可微,则其全微分

du=fx(x,y,z)dx+fy(x,y,z)dy+fz(x,y,z)dz.\boxed{du=f_x(x,y,z)dx+f_y(x,y,z)dy+f_z(x,y,z)dz}.

例子1

求函数z=x2y2z=x^2y^2在点(2,−1)(2,-1)处,当Δx=0.02\Delta x=0.02,Δy=−0.01\Delta y=-0.01时的全增量与全微分.

解:

有定义可知全增量等于Δz=f(x+Δx,y+Δy)−f(x,y)=f(2+0.02,−1−0.01)−f(2,−1)=4.16282404−4≈0.16\Delta z=f(x+\Delta x,y+\Delta y)-f(x,y)=f(2+0.02,-1-0.01)-f(2,-1)=4.16282404-4\approx0.16.

函数z=x2y2z=x^2y^2的两个偏导数为∂z∂x=2xy2,∂z∂y=2x2y,\dfrac{\partial z}{\partial x}=2xy^2,\dfrac{\partial z}{\partial y}=2x^2y,

所以dz=2xy2∣(2,−1)Δx+2x2y∣(2,−1)Δy=4⋅0.02−8⋅(−0.01)=0.08+0.08=0.16.dz=2xy^2\bigg|_{(2,-1)}\Delta x+2x^2y\bigg|_{(2,-1)}\Delta y=4\cdot 0.02-8\cdot (-0.01)=0.08+0.08=0.16.

例子2

设函数z=ex+y2z=e^x+y^2,求dzdz.

解:

函数的两个偏导数为∂z∂x=ex,∂z∂y=2y\dfrac{\partial z}{\partial x}=e^x,\dfrac{\partial z}{\partial y}=2y,

所以dz=exdx+2ydy.dz=e^xdx+2ydy.

例子3

设u=cos⁡x+sin⁡2z−y2zu=\cos x+\sin^2z-y^2z,求dudu.

解:

函数的三个偏导数为∂u∂x=−sin⁡x,∂u∂y=−2yz,αuαz=2sin⁡zcos⁡z−y2=sin⁡2z−y2\dfrac{\partial u}{\partial x}=-\sin x,\dfrac{\partial u}{\partial y}=-2yz,\dfrac{\alpha u}{\alpha z}=2\sin z\cos z-y^2=\sin2z-y^2,

所以du=∂u∂xdx+∂u∂ydy+αuαzz=−sin⁡xdx−2yzdy+(sin⁡2z−y2)dz.du=\dfrac{\partial u}{\partial x}dx+\dfrac{\partial u}{\partial y}dy+\dfrac{\alpha u}{\alpha z}z=-\sin xdx-2yzdy+(\sin2z-y^2)dz.

2.2、偏导数的存在性与连续性、函数可微性与函数连续性之间的关系

3、复合函数的偏导数与全微分

3.1、链式法则

image-20221117192139775

设函数u=u(x,y)u=u(x,y),v=v(x,y)v=v(x,y)在点(x,y)(x,y)处有连续偏导数,函数z=f(u,v)z=f(u,v)在对应点(u,v)(u,v)处有连续偏导数,则复合函数z=f[u(x,y),v(x,y)]z=f[u(x,y),v(x,y)]在点(x,y)(x,y)处有对xx与yy的偏导数,且有下列链式法则:

∂z∂x=∂z∂u⋅∂u∂x+∂z∂v⋅∂v∂x,\boxed{\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x}},

∂z∂y=∂z∂u⋅∂u∂y+∂z∂v⋅∂v∂y,\boxed{\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial y}},

链式法则求偏导的规律可以通过树形图表示,自上到下表示函数与自变量之间的对应关系,一句**“沿线相乘,分线相加”**的原则。

例子1

设z=u2ln⁡vz=u^2\ln v,而u=xyu=\dfrac{x}{y},v=3x−2yv=3x-2y,求∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}.

解:

∂z∂x=∂z∂u⋅∂u∂x+∂z∂v⋅∂v∂x=2uln⁡v⋅1y+u2v⋅3=2xyln⁡(3x−2y)+3x2y2(3x−2y)\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x}=2u\ln v\cdot\dfrac{1}{y}+\dfrac{u^2}{v}\cdot3=\dfrac{2x}{y}\ln(3x-2y)+\dfrac{3x^2}{y^2(3x-2y)},

∂z∂y=∂z∂u⋅∂u∂y+∂z∂v⋅∂v∂y=2uln⁡v⋅(−xy2)+u2v⋅(−2)=−2x2y3ln⁡(3x−2y)−2x2y2(3x−2y)\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial y}=2u\ln v\cdot(-\dfrac{x}{y^2})+\dfrac{u^2}{v}\cdot(-2)=-\dfrac{2x^2}{y^3}\ln(3x-2y)-\dfrac{2x^2}{y^2(3x-2y)}.

3.2、其他情形

image-20221117194655883

(1)设函数z=f(u,v,w)z=f(u,v,w),u=ϕ(x,y)u=\phi(x,y),v=ψ(x,y)v=\psi(x,y),w=ω(x,y)w=\omega(x,y),如图所示,如果函数u=ϕ(x,y)u=\phi(x,y),v=ψ(x,y)v=\psi(x,y)和w=ω(x,y)w=\omega(x,y)在点(x,y)(x,y)处都有对xx和对yy的偏导数,z=f(u,v,w)z=f(u,v,w)处具有连续偏导数,则z=f[ϕ(x,y),ψ(x,y),ω(x,y)]z=f[\phi(x,y),\psi(x,y),\omega(x,y)]在点(x,y)(x,y)处的两个偏导数存在,且

∂z∂x=∂z∂u⋅∂u∂x+∂z∂v⋅∂v∂x+∂z∂w⋅∂w∂x,\boxed{\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x}+\dfrac{\partial z}{\partial w}\cdot\dfrac{\partial w}{\partial x}},

∂z∂y=∂z∂u⋅∂u∂y+∂z∂v⋅∂v∂y+∂z∂w⋅∂w∂y.\boxed{\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial y}+\dfrac{\partial z}{\partial w}\cdot\dfrac{\partial w}{\partial y}}.

image-20221117195359392

(2)设函数z=f(u,v)z=f(u,v),u=ϕ(x)u=\phi(x),v=ψ(x)v=\psi(x),如图所示,如果函数u=ϕ(x)u=\phi(x)和v=ψ(x)v=\psi(x)在点xx处都可导,z=f(u,v)z=f(u,v)在点(u,v)(u,v)处具有连续偏导数,则z=f[u=ϕ(x),v=ψ(x)]z=f[u=\phi(x),v=\psi(x)]在点xx处可导,且

∂z∂x=∂z∂u⋅dudx+∂z∂v⋅dvdx,\boxed{\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot\dfrac{du}{dx}+\dfrac{\partial z}{\partial v}\cdot\dfrac{dv}{dx}},

这时zz对xx的导数叫作全导数。

image-20221117200713644

(3)设函数z=f(u,x)z=f(u,x),u=ϕ(x,y)u=\phi(x,y),如图所示,如果函数u=ϕ(x,y)u=\phi(x,y)在点(x,y)(x,y)处有对xx和yy的偏导数,z=f(u,x)z=f(u,x)在点(u,x)(u,x)处具有连续偏导数,则函数z=f[ϕ(x,y),x]z=f[\phi(x,y),x]在点(x,y)(x,y)处的两个偏导数存在,且

∂z∂x=∂f∂u⋅∂u∂x+∂f∂x,\boxed{\dfrac{\partial z}{\partial x}=\dfrac{\partial f}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial f}{\partial x}},

∂z∂y=∂f∂u⋅∂u∂y.\boxed{\dfrac{\partial z}{\partial y}=\dfrac{\partial f}{\partial u}\cdot\dfrac{\partial u}{\partial y}}.

⭐️⭐️注意:其中∂z∂x\dfrac{\partial z}{\partial x}与∂f∂x\dfrac{\partial f}{\partial x}是不同的,∂z∂x\dfrac{\partial z}{\partial x}表示在复合函数z=f[ϕ(x,y),x]z=f[\phi(x,y),x]中把yy看成常量,对xx求偏导,而∂f∂x\dfrac{\partial f}{\partial x}表示在函数z=f[u(x,y)]z=f[u(x,y)]中把uu看成常量,对xx求偏导。

image-20221117201353789

(4)设函数z=f(u)z=f(u)有连续的导数,u=u(x,y)u=u(x,y)在点(x,y)(x,y)处有对xx和对yy的偏导数,如图所示,函数z=f[u(x,y)]z=f[u(x,y)]在对应点(x,y)(x,y)处有连续的偏导数,则复合函数z=f[u(x,y)]z=f[u(x,y)]在点(x,y)(x,y)处有对xx与yy的连续偏导数,且

∂z∂x=dfdu⋅∂u∂x,\boxed{\dfrac{\partial z}{\partial x}=\dfrac{df}{du}\cdot\dfrac{\partial u}{\partial x}},

∂z∂y=dfdu⋅∂u∂y.\boxed{\dfrac{\partial z}{\partial y}=\dfrac{df}{du}\cdot\dfrac{\partial u}{\partial y}}.

⭐️⭐️⭐️链式法则求导规律:沿线相乘,分线相加,单线导数,分线偏导。

例子1

设z=yxz=\dfrac{y}{x},而x=etx=e^t,y=1−e2ty=1-e^{2t},求dzdt\dfrac{dz}{dt}.

解:

由题可知变量关系图如下:

image-20221117201808045

所以

dzdt=∂z∂x⋅dxdt+azay⋅dydt=−yx2⋅et+1x⋅(−2e2t)=−1−e2te2t⋅et+1et⋅(−2e2t)=e2t−1et−2e2tet=−e2t−1et=−et−e−t.\begin{aligned}\dfrac{dz}{dt}&=\dfrac{\partial z}{\partial x}\cdot\dfrac{dx}{dt}+\dfrac{az}{ay}\cdot\dfrac{dy}{dt}\\&=-\dfrac{y}{x^2}\cdot e^t+\dfrac{1}{x}\cdot(-2e^{2t})\\&=-\dfrac{1-e^{2t}}{e^{2t}}\cdot e^t+\dfrac{1}{e^t}\cdot(-2e^{2t})\\&=\dfrac{e^{2t}-1}{e^{t}}-\dfrac{2e^{2t}}{e^t}\\&=\dfrac{-e^{2t}-1}{e^t}\\&=-e^t-e^{-t}.\end{aligned}

例子2

求z=f(x,6xy)z=f(x,\dfrac{6x}{y})的偏导数,其中f(u,v)f(u,v)可微分.

解:

令u=6xyu=\dfrac{6x}{y},则z=f(x,u)z=f(x,u),变量间的关系图如下:

image-20221117203513413

∂z∂x=∂f∂x+∂z∂u⋅∂u∂x=fx+6yfu,\dfrac{\partial z}{\partial x}=\dfrac{\partial f}{\partial x}+\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}=f_x+\dfrac{6}{y}f_u,

∂z∂y=∂z∂u⋅∂u∂y=−6xy2fu.\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}=-\dfrac{6x}{y^2}f_u.

例子3

求z=sin⁡(x2y)z=\sin(x^2y)在点(1,2)(1,2)处的一阶偏导数.

解:

令u=x2yu=x^2y,则z=sin⁡uz=\sin u ,变量间的关系如下:

image-20221117205404529

∂z∂x=∂z∂u⋅∂u∂x=cos⁡u⋅2xy=2xycos⁡(x2y),\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}=\cos u\cdot2xy=2xy\cos(x^2y),

∂z∂y=∂z∂u⋅∂u∂y=cos⁡u⋅x2=x2cos⁡(x2y),\dfrac{\partial z}{\partial y}=\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}=\cos u\cdot x^2=x^2\cos(x^2y),

∂z∂x∣(1,2)=4cos⁡2,∂z∂y∣(1,2)=cos⁡2.\dfrac{\partial z}{\partial x}\bigg|_{(1,2)}=4\cos2,\dfrac{\partial z}{\partial y}\bigg|_{(1,2)}=\cos2.

例子4

设z=x3f(xy,tan⁡(xy))z=x^3f(xy,\tan(xy)),ff具有一阶连续的偏导数,求∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}.

解:

令u=xy,v=tan⁡(xy)u=xy,v=\tan(xy),则z=z(x,f(u,v))z=z(x,f(u,v)),变量间的关系如下图所示:

image-20221117212904057

∂z∂x=3x2f(u,v)+x3⋅(∂f∂u⋅∂u∂x+αfαv⋅∂v∂x)=3x2f+x3⋅[fu⋅y+fv⋅(ysec⁡2(xy))]=3x2f+x3yfu+x3ysec⁡2(xy)fv,\begin{aligned}\dfrac{\partial z}{\partial x}&=3x^2f(u,v)+x^3\cdot(\dfrac{\partial f}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\alpha f}{\alpha v}\cdot\dfrac{\partial v}{\partial x})\\&=3x^2f+x^3\cdot[f_u\cdot y+f_v\cdot(y\sec^2(xy))]\\&=3x^2f+x^3yf_u+x^3y\sec^2(xy)f_v,\end{aligned}

∂z∂y=x3⋅(∂f∂u⋅∂u∂y+αfαv⋅∂v∂y)=x3⋅[fu⋅x+fv⋅(xsec⁡2(xy))]=x4fu+x4sec⁡2(xy)fv.\begin{aligned}\dfrac{\partial z}{\partial y}&=x^3\cdot(\dfrac{\partial f}{\partial u}\cdot\dfrac{\partial u}{\partial y}+\dfrac{\alpha f}{\alpha v}\cdot\dfrac{\partial v}{\partial y})\\&=x^3\cdot[f_u\cdot x+f_v\cdot(x\sec^2(xy))]\\&=x^4f_u+x^4\sec^2(xy)f_v.\end{aligned}

例子5

设函数z=f(2x+y)+g(x,xy)z=f(2x+y)+g(x,xy),其中f(t)f(t),g(u,v)g(u,v)皆可微分,求∂z∂x\dfrac{\partial z}{\partial x},∂z∂y.\dfrac{\partial z}{\partial y}.

解:

由函数可知变量关系如下图所示:

image-20221117232256349

∂z∂x=∂z∂f⋅∂f∂x+∂z∂g⋅∂g∂x=∂z∂f⋅∂f∂t⋅∂t∂x+∂z∂g⋅(∂g∂u⋅∂u∂x+∂g∂v⋅∂v∂x)=∂z∂t⋅∂t∂x+∂z∂g⋅∂g∂u⋅∂u∂x+∂z∂g⋅∂g∂v⋅∂v∂x=∂z∂t⋅∂t∂x+∂z∂u⋅∂u∂x+∂z∂v⋅∂v∂x=(f′(2x+y)+0)⋅2+(0+gu(x,xy))⋅1+(0+gv(x,xy))⋅y=f′(2x+y)⋅2+g1′(x,xy)⋅1+g2′(x,xy)⋅y,\begin{aligned}\dfrac{\partial z}{\partial x}&=\dfrac{\partial z}{\partial f}\cdot\dfrac{\partial f}{\partial x}+\dfrac{\partial z}{\partial g}\cdot\dfrac{\partial g}{\partial x}\\&=\dfrac{\partial z}{\partial f}\cdot\dfrac{\partial f}{\partial t}\cdot\dfrac{\partial t}{\partial x}+\dfrac{\partial z}{\partial g}\cdot(\dfrac{\partial g}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial g}{\partial v}\cdot\dfrac{\partial v}{\partial x})\\&=\dfrac{\partial z}{\partial t}\cdot\dfrac{\partial t}{\partial x}+\dfrac{\partial z}{\partial g}\cdot\dfrac{\partial g}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial g}\cdot\dfrac{\partial g}{\partial v}\cdot\dfrac{\partial v}{\partial x}\\&=\dfrac{\partial z}{\partial t}\cdot\dfrac{\partial t}{\partial x}+\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x}\\&=(f'(2x+y)+0)\cdot2+(0+g_u(x,xy))\cdot1+(0+g_v(x,xy))\cdot y\\&=f'(2x+y)\cdot2+g'_1(x,xy)\cdot1+g'_2(x,xy)\cdot y,\end{aligned}

∂z∂y=∂z∂f⋅∂f∂y+∂z∂g⋅∂g∂y=∂z∂f⋅∂f∂t⋅∂t∂y+(∂z∂g⋅∂g∂v⋅∂v∂x)=∂z∂t⋅∂t∂y+∂z∂v⋅∂v∂x=(f′(2x+y)+0)⋅1+(0+gv(x,xy))⋅x=f′(2x+y)+xg2′(x,xy).\begin{aligned}\dfrac{\partial z}{\partial y}&=\dfrac{\partial z}{\partial f}\cdot\dfrac{\partial f}{\partial y}+\dfrac{\partial z}{\partial g}\cdot\dfrac{\partial g}{\partial y}\\&=\dfrac{\partial z}{\partial f}\cdot\dfrac{\partial f}{\partial t}\cdot\dfrac{\partial t}{\partial y}+(\dfrac{\partial z}{\partial g}\cdot\dfrac{\partial g}{\partial v}\cdot\dfrac{\partial v}{\partial x})\\&=\dfrac{\partial z}{\partial t}\cdot\dfrac{\partial t}{\partial y}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x}\\&=(f'(2x+y)+0)\cdot1+(0+g_v(x,xy))\cdot x\\&=f'(2x+y)+xg'_2(x,xy).\end{aligned}

3.3、全微分形式不变性

设u=u(x,y)u=u(x,y),v=(x,y)v=(x,y)在点(x,y)(x,y)处有连续偏导数,函数z=f(u,v)z=f(u,v)在对应点(u,v)(u,v)处有连续偏导数,则z=f(u,v)=f[u(x,y),v(x,y)]z=f(u,v)=f[u(x,y),v(x,y)]的全微分

dz=∂z∂xdx+∂z∂ydy=(∂z∂u⋅∂u∂x+∂z∂v⋅∂v∂x)dx+(∂z∂u⋅∂u∂y+∂z∂v⋅∂v∂y)dy=∂z∂u(∂u∂xdx+∂u∂ydy)+∂z∂v(∂v∂xdx+∂v∂ydy)=∂z∂udu+∂z∂vdv.\begin{aligned}dz&=\dfrac{\partial z}{\partial x}dx+\dfrac{\partial z}{\partial y}dy=(\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial x})dx+(\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y}+\dfrac{\partial z}{\partial v}\cdot\dfrac{\partial v}{\partial y})dy\\&=\dfrac{\partial z}{\partial u}(\dfrac{\partial u}{\partial x}dx+\dfrac{\partial u}{\partial y}dy)+\dfrac{\partial z}{\partial v}(\dfrac{\partial v}{\partial x}dx+\dfrac{\partial v}{\partial y}dy)=\dfrac{\partial z}{\partial u}du+\dfrac{\partial z}{\partial v}dv.\end{aligned}

这表明当ff,uu,vv都是可微函数时,尽管uu,vv不是自变量,但函数z=f(u,v)z=f(u,v)的全微分也具有与uu,vv是自变量时的全微分相同的形式,这种性质称为全微分形式不变性。

对于多元函数的全微分,四则运算法则仍成立,即当uu,vv均可微时,有

d(u±v)=du±dv;d(u\pm v)=du\pm dv;

d(uv)=vdu+udv;d(uv)=vdu+udv;

d(uv)=vdu−udvv2(v≠0).d(\dfrac{u}{v})=\dfrac{vdu-udv}{v^2}(v\neq0).

例子1

设二元函数z=(x+1)ln⁡(1+y)z=(x+1)\ln(1+y),求∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}.

解:

方法一 利用全微分形式不变性直接计算,得

dz=ln⁡(1+y)d(x+1)+(x+1)d[ln⁡(1+y)]=ln⁡(1+y)dx+x+11+ydy\begin{aligned}dz&=\ln(1+y)d(x+1)+(x+1)d[\ln(1+y)]\\&=\ln(1+y)dx+\dfrac{x+1}{1+y}dy\end{aligned}

于是∂z∂x=ln⁡(1+y),∂z∂y=x+11+y\dfrac{\partial z}{\partial x}=\ln(1+y),\dfrac{\partial z}{\partial y}=\dfrac{x+1}{1+y}.

方法二 ∂z∂x=ln⁡(1+y),∂z∂y=(x+1)⋅11+y=x+11+y.\dfrac{\partial z}{\partial x}=\ln(1+y),\dfrac{\partial z}{\partial y}=(x+1)\cdot\dfrac{1}{1+y}=\dfrac{x+1}{1+y}.

3.4 复合函数的全微分

求解二元复合函数的全微分时,可以利用复合函数求偏导法则先求出∂z∂x\dfrac{\partial z}{\partial x}和∂z∂y\dfrac{\partial z}{\partial y},然后代入公式dz=∂z∂xdx+∂z∂ydydz=\dfrac{\partial z}{\partial x}dx+\dfrac{\partial z}{\partial y}dy,或者利用全微分形式不变性,对函数两边同时微分,然后进行整理,求出全微分dzdz。

例子1

设二元函数z=(x+1)ln⁡(1+y)z=(x+1)\ln(1+y),求dz∣(1.0)dz\bigg|_{(1.0)}.

解:

dz=ln⁡(1+y)dx+x+11+ydydz=\ln(1+y)dx+\dfrac{x+1}{1+y}dy,

所以dz∣(1,0)=ln⁡(1+0)dx+1+11+0dy=2dy.dz\bigg|_{(1,0)}=\ln(1+0)dx+\dfrac{1+1}{1+0}dy=2dy.

例子2

设二元函数z=arctan⁡xyz=\arctan xy,求dzdz.

解:

令xy=uxy=u,则z=arctan⁡uz=\arctan u,

dz=(∂z∂u⋅∂u∂x)dx+(∂z∂u⋅∂u∂y)dy=11+u2⋅ydx+11+u2⋅xdy=y1+x2y2dx+x1+x2y2dy.dz=(\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial x})dx+(\dfrac{\partial z}{\partial u}\cdot\dfrac{\partial u}{\partial y})dy=\dfrac{1}{1+u^2}\cdot ydx+\dfrac{1}{1+u^2}\cdot xdy=\dfrac{y}{1+x^2y^2}dx+\dfrac{x}{1+x^2y^2}dy.

3.5、复合函数的二阶偏导数

在求出复合函数的一阶偏导数∂z∂x\dfrac{\partial z}{\partial x}和∂z∂y\dfrac{\partial z}{\partial y}后,按二阶偏导数的定义再对其求偏导数,可得到相应的二阶偏导数.

例子1

设函数z=exeyz=e^{xe^y},则∂2z∂x∂y=___.\dfrac{\partial^2z}{\partial x\partial y}=\_\_\_.

解:

令u=xeyu=xe^y,则,z=eu,z=e^u,

∂z∂x=∂z∂u⋅∂u∂x=eu⋅ey=exey+y,\dfrac{\partial z}{\partial x}=\dfrac{\partial z}{\partial u}\cdot \dfrac{\partial u}{\partial x}=e^u\cdot e^y=e^{xe^y+y},

令v=xey+y,w=evv=xe^y+y,w=e^v,

则∂w∂y=ev⋅(xey+1)=exey+y(xey+1)\dfrac{\partial w}{\partial y}=e^v\cdot(xe^y+1)=e^{xe^y+y}(xe^y+1),即∂2z∂x∂y=exey+y(xey+1)\dfrac{\partial^2z}{\partial x\partial y}=e^{xe^y+y}(xe^y+1).

例子2

设z=f(xy,x2+y2)z=f(\dfrac{x}{y},x^2+y^2),其中ff具有二阶连续偏导数,求∂2z∂x∂y.\dfrac{\partial^2z}{\partial x\partial y}.

解:

令u=xy,v=x2+y2u=\dfrac{x}{y},v=x^2+y^2,则z=f(u,v)z=f(u,v),

∂z∂x=∂f∂u⋅∂u∂x+∂f∂v⋅∂v∂x=fu⋅1y+fv⋅(2x)=1yf1′+2xf2′,\dfrac{\partial z}{\partial x}=\dfrac{\partial f}{\partial u}\cdot\dfrac{\partial u}{\partial x}+\dfrac{\partial f}{\partial v}\cdot\dfrac{\partial v}{\partial x}=f_u\cdot\dfrac{1}{y}+f_v\cdot(2x)=\dfrac{1}{y}f'_1+2xf'_2,

∂2z∂x∂y=−1y2f1′+1y(−xy2f11′′+2yf12′′)+2x(−xy2f21′′+2yf22′′)=−1y2f1′−xy3f′′11+2f12′′−2x2y2f21′′+4xyf22′′=−1y2f1′−xy3f11′′+2(1−x2y2)f12′′+4xyf22′′.\begin{aligned}\dfrac{\partial^2z}{\partial x\partial y}&=-\dfrac{1}{y^2}f'_1+\dfrac{1}{y}(-\dfrac{x}{y^2}f''_{11}+2yf''_{12})+2x(-\dfrac{x}{y^2}f''_{21}+2yf''_{22})\\&=-\dfrac{1}{y^2}f'_1-\dfrac{x}{y^3}f''{11}+2f''_{12}-\dfrac{2x^2}{y^2}f''_{21}+4xyf''_{22}\\&=-\dfrac{1}{y^2}f'_1-\dfrac{x}{y^3}f''_{11}+2(1-\dfrac{x^2}{y^2})f''_{12}+4xyf''_{22}.\end{aligned}

(解释:在求二阶求导时,对f1′f'_1求偏导时需分别对uu和vv求导,f2′f'_2也同理)

4、隐函数的偏导数与全微分

4.1、一元隐函数的导数

设y=y(x)y=y(x)是由方程F(x,y)=0F(x,y)=0在P(x,y)P(x,y)的某一领域内所确定的一个连续且具有连续导数的函数,函数F(x,y)F(x,y)在点P(x,y)P(x,y)的某一领域内有连续偏导数,由多元复合函数的求导法则,对方程F(x,y)=0F(x,y)=0两边求关于xx的导数,其中yy是xx的函数,得

∂F∂x+∂F∂y⋅dydx=0,\boxed{\dfrac{\partial F}{\partial x}+\dfrac{\partial F}{\partial y}\cdot\dfrac{d y}{d x}=0},

即Fx+Fy⋅dydx=0F_x+F_y\cdot\dfrac{dy}{dx}=0。若Fy≠0F_y\neq0,则有

dydx=−FxFy.\boxed{\dfrac{dy}{dx}=-\dfrac{F_x}{F_y}}.

例子1

求由方程x2+xy+2y−2x=0x^2+xy+2^y-2^x=0所确定的隐函数y=y(x)的导数y=y(x)的导数dydx\dfrac{dy}{dx}.

解:

令F(x,y)=x2+xy+2y−2x=0F(x,y)=x^2+xy+2^y-2^x=0,则Fx=2x+y−2xln⁡2,Fy=x+2yln⁡2,F_x=2x+y-2^x\ln2,F_y=x+2^y\ln2,

则dydx=−FxFy=−2x+y−2xln⁡2x+2yln⁡2.\dfrac{dy}{dx}=-\dfrac{F_x}{F_y}=-\dfrac{2x+y-2^x\ln2}{x+2^y\ln2}.

4.2、二元隐函数的偏导数

设z=f(x,y)z=f(x,y)是由方程F(x,y,z)=0F(x,y,z)=0在点P(x,y,z)P(x,y,z)的某一领域内所确定的一个连续且具有连续偏导数的二元函数,函数F(x,y,z)F(x,y,z)在点P(x,y,z)P(x,y,z)的某一领域内具有连续偏导数,对F(x,y,z)=0F(x,y,z)=0两边分别求关于xx,yy的偏导数,得

∂F∂x+∂F∂z⋅∂z∂x=0,∂F∂y+∂F∂z⋅∂z∂y=0.\boxed{\dfrac{\partial F}{\partial x}+\dfrac{\partial F}{\partial z}\cdot\dfrac{\partial z}{\partial x}=0,\dfrac{\partial F}{\partial y}+\dfrac{\partial F}{\partial z}\cdot\dfrac{\partial z}{\partial y}=0}.

若Fx≠0F_x\neq0,则有

∂z∂x=−FxFz,∂z∂y=−FyFz.\boxed{\dfrac{\partial z}{\partial x}=-\dfrac{F_x}{F_z},\dfrac{\partial z}{\partial y}=-\dfrac{F_y}{F_z}}.

例子1

求由方程x2+y2+2x−2yz=ezx^2+y^2+2x-2yz=e^z所确定的隐函数z=z(x,y)z=z(x,y)的偏导数∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}.

解:

方法一

令F(x,y,z)=x2+y2+2x−2yz−ezF(x,y,z)=x^2+y^2+2x-2yz-e^z,则Fx=2x+2,Fy=2y−2z,F_x=2x+2,F_y=2y-2z,

Fz=−2y−ez,F_z=-2y-e^z,若Fz≠0F_z\neq 0则∂z∂x=−FxFz=2(x+1)2y+ez,\dfrac{\partial z}{\partial x}=-\dfrac{F_x}{F_z}=\dfrac{2(x+1)}{2y+e^z},

∂z∂y=−FyFz=2(y−z)2y+ez\dfrac{\partial z}{\partial y}=-\dfrac{F_y}{F_z}=\dfrac{2(y-z)}{2y+e^z}.

方法二

利用全微分形式不变性,方程两边微分可得

2xdx+2ydy+2dx−2zdy−2ydz=ezdz2xdx+2ydy+2dx-2zdy-2ydz=e^zdz,

当ez+2y≠0e^z+2y\neq0,整理得

dz=2(x+1)2y+ezdx+2(y−z)2y+ezdy,dz=\dfrac{2(x+1)}{2y+e^z}dx+\dfrac{2(y-z)}{2y+e^z}dy,

故∂z∂x=2(x+1)2y+ez,∂z∂y=2(y−z)2y+ez.\dfrac{\partial z}{\partial x}=\dfrac{2(x+1)}{2y+e^z},\dfrac{\partial z}{\partial y}=\dfrac{2(y-z)}{2y+e^z}.

例子2

设函数z=z(x,y)z=z(x,y)由方程xz=ln⁡zy\dfrac{x}{z}=\ln\dfrac{z}{y}确定,求∂z∂x+∂z∂y.\dfrac{\partial z}{\partial x}+\dfrac{\partial z}{\partial y}.

解:

令F(x,y,z)=xz−ln⁡zyF(x,y,z)=\dfrac{x}{z}-\ln\dfrac{z}{y},则

Fx=1z,Fy=−1zy⋅(−zy2)=1y,Fz=−xz2−1zy⋅1y=−xz2−1z,F_x=\dfrac{1}{z},F_y=-\dfrac{1}{\frac{z}{y}}\cdot(-\dfrac{z}{y^2})=\dfrac{1}{y},F_z=-\dfrac{x}{z^2}-\dfrac{1}{\frac{z}{y}}\cdot\dfrac{1}{y}=-\dfrac{x}{z^2}-\dfrac{1}{z},

则∂z∂x=−FxFz=−1z−xz2−1z=zx+z,∂z∂y=−FyFz=−1y−xz2−1z=z2y(x+z),\dfrac{\partial z}{\partial x}=-\dfrac{F_x}{F_z}=-\dfrac{\dfrac{1}{z}}{-\dfrac{x}{z^2}-\dfrac{1}{z}}=\dfrac{z}{x+z},\dfrac{\partial z}{\partial y}=-\dfrac{F_y}{F_z}=-\dfrac{\dfrac{1}{y}}{-\dfrac{x}{z^2}-\dfrac{1}{z}}=\dfrac{z^2}{y(x+z)},

所以∂z∂x+∂z∂y.=zx+z+z2y(x+z)=z(z+y)y(x+z).\dfrac{\partial z}{\partial x}+\dfrac{\partial z}{\partial y}.=\dfrac{z}{x+z}+\dfrac{z^2}{y(x+z)}=\dfrac{z(z+y)}{y(x+z)}.

例子3

设函数u=f(x,y,z)=x3y2z2u=f(x,y,z)=x^3y^2z^2,其中z=z(x,y)z=z(x,y)由方程x3+y3+z3−3xyz=0x^3+y^3+z^3-3xyz=0所确定,求∂u∂x\dfrac{\partial u}{\partial x}.

解:

∂u∂x=3x2y2z2+2x3y2z⋅∂z∂x\dfrac{\partial u}{\partial x}=3x^2y^2z^2+2x^3y^2z\cdot\dfrac{\partial z}{\partial x}

令F(x,y,z)=x3+y3+z3−3xyzF(x,y,z)=x^3+y^3+z^3-3xyz,

Fx=3x2−3yz,Fz=3z2−3xyF_x=3x^2-3yz,F_z=3z^2-3xy

则∂z∂x=−x2−yzz2−xy\dfrac{\partial z}{\partial x}=-\dfrac{x^2-yz}{z^2-xy}

所以∂u∂x=3x2y2z2+2x3y2z⋅(yz−x2)z2−xy\dfrac{\partial u}{\partial x}=3x^2y^2z^2+\dfrac{2x^3y^2z\cdot(yz-x^2)}{z^2-xy}

4.3、二元隐函数的全微分

求解由F(x,y,z)=0F(x,y,z)=0所确定的隐函数z=z(x,y)z=z(x,y)的全微分时,可以利用二元隐函数求偏导数的公式或者复合函数求偏导法则先求出∂z∂x\dfrac{\partial z}{\partial x}和∂z∂y\dfrac{\partial z}{\partial y},然后代入公式dz=∂z∂xdx+∂z∂ydydz=\dfrac{\partial z}{\partial x}dx+\dfrac{\partial z}{\partial y}dy,或者利用全微分形式不变性,对F(x,y,z)=0F(x,y,z)=0两边同时微分,然后进行整理,求出全微分dzdz。

例子1

设函数z=z(x,y)z=z(x,y)由方程xyz+x2+y2+z2=2xyz+\sqrt{x^2+y^2+z^2}=\sqrt{2}所确定,求函数z(x,y)z(x,y)在点(1,0,−1)(1,0,-1)处的全微分.

解:

令F(x,y,z)=xyz+x2+y2+z2−2F(x,y,z)=xyz+\sqrt{x^2+y^2+z^2}-\sqrt{2},

则Fx=yz+xx2+y2+z2,Fy=xz+yx2+y2+z2F_x=yz+\dfrac{x}{\sqrt{x^2+y^2+z^2}},F_y=xz+\dfrac{y}{\sqrt{x^2+y^2+z^2}},

Fz=xy+zx2+y2+z2F_z=xy+\dfrac{z}{\sqrt{x^2+y^2+z^2}},则∂z∂x=−FxFz=−yz+xx2+y2+z2xy+zx2+y2+z2,∂z∂y=−FyFz=−z+xyx2+y2+z2xy+zx2+y2+z2\dfrac{\partial z}{\partial x}=-\dfrac{F_x}{F_z}=-\dfrac{yz+\dfrac{x}{\sqrt{x^2+y^2+z^2}}}{xy+\dfrac{z}{\sqrt{x^2+y^2+z^2}}},\dfrac{\partial z}{\partial y}=-\dfrac{F_y}{F_z}=-\dfrac{z+\dfrac{x}y{\sqrt{x^2+y^2+z^2}}}{xy+\dfrac{z}{\sqrt{x^2+y^2+z^2}}},

故∂z∂x∣(1,0,−1)=1,∂z∂y∣(1,0,−1)=−2,\dfrac{\partial z}{\partial x}\bigg|_{(1,0,-1)}=1,\dfrac{\partial z}{\partial y}\bigg|_{(1,0,-1)}=-\sqrt{2},

所以dz∣(1,0,−1)=dx−2dy.dz\bigg|_{(1,0,-1)}=dx-\sqrt{2}dy.

4.4、隐函数的二阶偏导数

在求出隐函数F(x,y,z)=0F(x,y,z)=0的一阶偏导数∂z∂x\dfrac{\partial z}{\partial x}和∂z∂y\dfrac{\partial z}{\partial y}后,按二阶偏导数的定义再对其求偏导,可得到相应的二阶偏导数。由于隐函数的一阶偏导数∂z∂x\dfrac{\partial z}{\partial x}和∂z∂y\dfrac{\partial z}{\partial y}中常常含有zz,故在求二阶偏导数时结果常需代入∂z∂x\dfrac{\partial z}{\partial x}或∂z∂y\dfrac{\partial z}{\partial y}的值.

例子1

设z=z(x,y)z=z(x,y)是由x2+y2+z2−4z=0x^2+y^2+z^2-4z=0所确定的隐函数,求∂2z∂x2.\dfrac{\partial ^2z}{\partial x^2}.

解:

令F(x,y,z)=x2+y2+z2−4zF(x,y,z)=x^2+y^2+z^2-4z,则Fx=2x,Fz=2z−4F_x=2x,F_z=2z-4,

当Fz≠0F_z\neq0时,则∂z∂x=−2x2z−4=x2−z\dfrac{\partial z}{\partial x}=-\dfrac{2x}{2z-4}=\dfrac{x}{2-z},

则∂2z∂x2=α(∂z∂x)αx=2−z−x⋅∂z∂x⋅(−1)(2−z)2=2−z+x22−z(2−z)2=(2−z)2+x2(2−z)3.\dfrac{\partial ^2z}{\partial x^2}=\dfrac{\alpha(\dfrac{\partial z}{\partial x})}{\alpha x}=\dfrac{2-z-x\cdot\dfrac{\partial z}{\partial x}\cdot(-1)}{(2-z)^2}=\dfrac{2-z+\dfrac{x^2}{2-z}}{(2-z)^2}=\dfrac{(2-z)^2+x^2}{(2-z)^3}.

注意:x⋅∂z∂x⋅(−1)x\cdot\dfrac{\partial z}{\partial x}\cdot(-1)可以看作是x⋅(1−z)′x\cdot(1-z)',由于z=z(x,y)z=z(x,y),故需要把zz对xx求偏导,复合求导.

多元函数微分学的几何应用

1、求空间曲线的切线及法平面

image-20230108165210416

切点为(x0,y0,z0)(x_0,y_0,z_0)求空间曲线{x=x(t)y=y(t)z=z(t)\begin{cases}x=x(t)\\y=y(t)\\z=z(t)\end{cases}的切线及法平面方程

解法:

①求导得向量:s⃗=(x′(t),y′(t),z′(t))=(m,n,p)\vec{s}=(x'(t),y'(t),z'(t))=(m,n,p)

n⃗=(x′(t),y′(t),z′(t))=(A,B,C)\vec{n}=(x'(t),y'(t),z'(t))=(A,B,C)

②套公式得出答案:“线除面乘”

切线:x−x0x′(t)=y−y0y′(t)=z−z0z′(t)\dfrac{x-x_0}{x'(t)}=\dfrac{y-y_0}{y'(t)}=\dfrac{z-z_0}{z'(t)}

法平面:x′(t)(x−x0)+y′(t)(y−y0)+z′(t)(z−z0)=0x'(t)(x-x_0)+y'(t)(y-y_0)+z'(t)(z-z_0)=0

例子1

求螺旋线x=acos⁡θx=a\cos\theta,y=asin⁡θy=a\sin\theta,z=bθz=b\theta,(aa、bb不同时为0)上在θ=π4\theta=\dfrac{\pi}{4}时的切线方程和法平面方程.

解:

令t=θt=\theta,则参数方程为{x=acos⁡ty=asin⁡tz=bt\begin{cases}x=a\cos t\\y=a\sin t\\z=bt\end{cases}

当θ=π4\theta=\dfrac{\pi}{4}时,x=2a2,y=2a2,z=π4bx=\dfrac{\sqrt{2}a}{2},y=\dfrac{\sqrt{2}a}{2},z=\dfrac{\pi}{4}b

故切点为(2a2,2a2,π4b)(\dfrac{\sqrt{2}a}{2},\dfrac{\sqrt{2}a}{2},\dfrac{\pi}{4}b)

x′=−asin⁡θ,y′=acos⁡θ,z′=bx'=-a\sin\theta,y'=a\cos\theta,z'=b

当θ=π4\theta=\dfrac{\pi}{4}时,x′=−2a2,y′=2a2,z′=π4x'=-\dfrac{\sqrt{2}a}{2},y'=\dfrac{\sqrt{2}a}{2},z'=\dfrac{\pi}{4}

则切线的方向向量为(−2a2,2a2,b)(-\dfrac{\sqrt{2}a}{2},\dfrac{\sqrt{2}a}{2},b)

法平面的法向量为(−2a2,2a2,b)(-\dfrac{\sqrt{2}a}{2},\dfrac{\sqrt{2}a}{2},b)

故切线方程为:x−2a2−2a2=y−2a22a2=z−π4bb\dfrac{x-\dfrac{\sqrt{2}a}{2}}{-\dfrac{\sqrt{2}a}{2}}=\dfrac{y-\dfrac{\sqrt{2}a}{2}}{\dfrac{\sqrt{2}a}{2}}=\dfrac{z-\dfrac{\pi}{4}b}{b}

法平面方程为:−2a2(x−2a2)+2a2(y−2a2)+b(z−π4b)=0-\dfrac{\sqrt{2}a}{2}(x-\dfrac{\sqrt{2}a}{2})+\dfrac{\sqrt{2}a}{2}(y-\dfrac{\sqrt{2}a}{2})+b(z-\dfrac{\pi}{4}b)=0

例子2

求曲线y=sin⁡xy=\sin x,z=x2z=\dfrac{x}{2}在点(π,0,π2)(\pi,0,\dfrac{\pi}{2})处的法平面方程.

解:

令x=tx=t,则参数方程为{x=ty=sin⁡tz=12t\begin{cases}x=t\\y=\sin t\\z=\dfrac{1}{2}t\end{cases}

x′=1,y′=cos⁡t,z′=12x'=1,y'=\cos t,z'=\dfrac{1}{2},

在点(π,0,π2)(\pi,0,\dfrac{\pi}{2})处x=t=πx=t=\pi

则y′=−1y'=-1

故曲线的法向量n⃗=(1,−1,12)\vec{n}=(1,-1,\dfrac{1}{2})

故法平面方程为:(x−π)−(y−0)+12(z−π2)=0(x-\pi)-(y-0)+\dfrac{1}{2}(z-\dfrac{\pi}{2})=0

例子3

求曲线LL:x=tx=t,y=t2y=t^2,z=t3z=t^3上的一点,使得曲线在该点处的切线平行于平面3x+3y+z=13x+3y+z=1.

解:

令曲线待求点为(t0,t02,t03)(t_0,t_0^2,t_0^3)

故曲线LL的方向向量为s⃗=(1,2t0,3t02)\vec{s}=(1,2t_0,3t_0^2)

平面3x+3y+z=13x+3y+z=1的法向量为n⃗=(3,3,1)\vec{n}=(3,3,1),

又曲线LL在该点处的切线平行于平面3x+3y+z=13x+3y+z=1,

所以n⃗⊥s⃗\vec{n}\perp\vec{s},

即n⃗⋅s⃗=3+6t0+3t02=0⇒t0=−1\vec{n}\cdot\vec{s}=3+6t_0+3t_0^2=0\Rightarrow t_0=-1

故该点为(−1,1,−1)(-1,1,-1),使得曲线在该点处的切线平行于平面3x+3y+z=13x+3y+z=1.

2、求空间曲面的切平面及法线

求空间曲面F(x,y,z)F(x,y,z)的切平面和法线方程

解法:

①求偏导得向量

n⃗=(A,B,C)=(Fx,Fy,Fz)\vec{n}=(A,B,C)=(F_x,F_y,F_z)

s⃗=(m,n,p)=(Fx,Fy,Fz)\vec{s}=(m,n,p)=(F_x,F_y,F_z)

②套公式:“线除面乘”

法线:x−x0Fx=y−y0Fy=z−z0Fz\dfrac{x-x_0}{F_x}=\dfrac{y-y_0}{F_y}=\dfrac{z-z_0}{F_z}

切平面:Fx(x−x0)+Fy(y−y0)+Fz(z−z0)=0F_x(x-x_0)+F_y(y-y_0)+F_z(z-z_0)=0

例子1

求空间曲面2z−ez+2xy=32z-e^z+2xy=3在(1,2,0)(1,2,0)处的法线方程.

解:

令F(x,y,z)=2z−ez+2xy−3F(x,y,z)=2z-e^z+2xy-3

则Fx=2y,Fy=2x,Fz=2−ezF_x=2y,F_y=2x,F_z=2-e^z

故在点(1,2,0)(1,2,0)处的方向向量为s⃗=(2y,2x,2−ez)∣(1,2,0)=(4,2,1)\vec{s}=(2y,2x,2-e^z)|_{(1,2,0)}=(4,2,1)

故该点处的法线方程为:x−14=y−22=z−01\dfrac{x-1}{4}=\dfrac{y-2}{2}=\dfrac{z-0}{1}

例子2

已知曲面z=4−x2−y2z=4-x^2-y^2上点PP处的切平面平行于平面2x+2y+z−1=02x+2y+z-1=0,则PP的坐标是?

解:

设PP点坐标为(x0,y0,z0)(x_0,y_0,z_0)

令F(x,y,z)=4−x2−y2−zF(x,y,z)=4-x^2-y^2-z

则Fx=−2x,Fy=−2y,Fz=−1F_x=-2x,F_y=-2y,F_z=-1

故曲面在点PP处切平面的法向量为n1⃗=(−2x0,−2y0,−1)\vec{n_1}=(-2x_0,-2y_0,-1)

又平面2x+2y+z−1=02x+2y+z-1=0的法向量为n2⃗=(2,2,1)\vec{n_2}=(2,2,1),且群面的切平面与平面平行,

故−2x02=−2y02=−11⇒x0=1,y0=1\dfrac{-2x_0}{2}=\dfrac{-2y_0}{2}=\dfrac{-1}{1}\Rightarrow x_0=1,y_0=1

故PP点的坐标是(1,1,2)(1,1,2)

方向导数与梯度

1、梯度

  1. 二元:函数z=f(x,y)z=f(x,y),在点P(x0,y0)P(x_0,y_0)的梯度,记为gradf(x0,y0)gradf(x_0,y_0)或∇f(x0,y0)\nabla f(x_0,y_0)

    gradf(x0,y0)=∂f∂x∣Pi⃗+∂f∂f∣Pj⃗=(∂f∂x∣P,∂f∂y∣P)gradf(x_0,y_0)=\dfrac{\partial f}{\partial x}\bigg|_P\vec{i}+\dfrac{\partial f}{\partial f}\bigg|_P\vec{j}=(\dfrac{\partial f}{\partial x}\bigg|_P,\dfrac{\partial f}{\partial y}\bigg|_P)

  2. 三元:函数u=f(x,y,z)u=f(x,y,z),在点P(x0,y0,z0)P(x_0,y_0,z_0)的梯度,记为gradf(x0,y0,z0)gradf(x_0,y_0,z_0)或∇f(x0,y0,z0)\nabla f(x_0,y_0,z_0)

    gradf(x0,y0,z0)=∂f∂x∣Pi⃗+∂f∂f∣Pj⃗+∂f∂z∣Pk⃗=(∂f∂x∣P,∂f∂y∣P,∂f∂z∣P)gradf(x_0,y_0,z_0)=\dfrac{\partial f}{\partial x}\bigg|_P\vec{i}+\dfrac{\partial f}{\partial f}\bigg|_P\vec{j}+\dfrac{\partial f}{\partial z}\bigg|_P\vec{k}=(\dfrac{\partial f}{\partial x}\bigg|_P,\dfrac{\partial f}{\partial y}\bigg|_P,\dfrac{\partial f}{\partial z}\bigg|_P)

例子1

f(x,y)=x2+y2f(x,y)=x^2+y^2,则gradf(1,−1)=gradf(1,-1)=?

解:

∂f∂x∣(1,−1)=2x∣(1,−1)=2,∂f∂y∣(1,−1)=2y∣(1,−1)=−2\dfrac{\partial f}{\partial x}\bigg|_{(1,-1)}=2x\bigg|_{(1,-1)}=2,\dfrac{\partial f}{\partial y}\bigg|_{(1,-1)}=2y\bigg|_{(1,-1)}=-2

故gradf(1,−1)=2i⃗−2j⃗=(2,−2)gradf(1,-1)=2\vec{i}-2\vec{j}=(2,-2)

例子2

已知f(x,y,z)=ln⁡x2+y2+z2f(x,y,z)=\ln\sqrt{x^2+y^2+z^2},则gradf(1,1,1)=gradf(1,1,1)=?

解:

∂f∂x∣(1,1,1)=xx2+y2+z2∣(1,1,1)=13\dfrac{\partial f}{\partial x}\bigg|_{(1,1,1)}=\dfrac{x}{x^2+y^2+z^2}\bigg|_{(1,1,1)}=\dfrac{1}{3},

∂f∂y∣(1,1,1)=yx2+y2+z2∣(1,1,1)=13\dfrac{\partial f}{\partial y}\bigg|_{(1,1,1)}=\dfrac{y}{x^2+y^2+z^2}\bigg|_{(1,1,1)}=\dfrac{1}{3},

∂f∂z∣(1,1,1)=zx2+y2+z2∣(1,1,1)=13\dfrac{\partial f}{\partial z}\bigg|_{(1,1,1)}=\dfrac{z}{x^2+y^2+z^2}\bigg|_{(1,1,1)}=\dfrac{1}{3}

故gradf(1,1,1)={13,13,13}gradf(1,1,1)=\{\dfrac{1}{3},\dfrac{1}{3},\dfrac{1}{3}\}

2、方向导数

  1. 二元:函数z=f(x,y)z=f(x,y),在点P(x0,y0)P(x_0,y_0)处沿着某个方向l⃗\vec{l}的偏导,记作∂z∂l⃗∣P\dfrac{\partial z}{\partial \vec{l}}\bigg|_P或∂f∂l⃗∣P\dfrac{\partial f}{\partial \vec{l}}\bigg|_P

    ∂f∂l⃗∣P=∂f∂x∣Pcos⁡α+∂f∂y∣Pcos⁡β\dfrac{\partial f}{\partial \vec{l}}\bigg|_P=\dfrac{\partial f}{\partial x}\bigg|_P\cos \alpha+\dfrac{\partial f}{\partial y}\bigg|_P\cos\beta

  2. 三元:函数z=f(x,y,z)z=f(x,y,z),在点P(x0,y0,z0)P(x_0,y_0,z_0)处沿着某个方向l⃗\vec{l}的偏导,记作∂z∂l⃗∣P\dfrac{\partial z}{\partial \vec{l}}\bigg|_P或∂f∂l⃗∣P\dfrac{\partial f}{\partial \vec{l}}\bigg|_P

    ∂f∂l⃗∣P=∂f∂x∣Pcos⁡α+∂f∂y∣Pcos⁡β+∂f∂z∣Pcos⁡γ\dfrac{\partial f}{\partial \vec{l}}\bigg|_P=\dfrac{\partial f}{\partial x}\bigg|_P\cos \alpha+\dfrac{\partial f}{\partial y}\bigg|_P\cos\beta+\dfrac{\partial f}{\partial z}\bigg|_P\cos\gamma

image-20230111180909651

例子1

z=x2+y2z=x^2+y^2在点(1,2)(1,2)处沿(1,2)(1,2)到(2,2+3)(2,2+\sqrt{3})方向的方向导数为?

解:

方向l⃗=(1,3),cos⁡α=12cos⁡β=32\vec{l}=(1,\sqrt{3}),\cos\alpha=\dfrac{1}{2}\cos\beta=\dfrac{\sqrt{3}}{2}

则方向导数为:

∂z∂l⃗∣(1,2)=∂z∂x∣(1,2)cos⁡α+∂z∂y∣(1,2)cos⁡β=2x∣(1,2)⋅12+2y∣(1,2)⋅32=1+23\begin{aligned}\dfrac{\partial z}{\partial\vec{l}}\bigg|_{(1,2)}&=\dfrac{\partial z}{\partial x}\bigg|_{(1,2)}\cos\alpha+\dfrac{\partial z}{\partial y}\bigg|_{(1,2)}\cos\beta\\&=2x\bigg|_{(1,2)}\cdot\dfrac{1}{2}+2y\bigg|_{(1,2)}\cdot\dfrac{\sqrt{3}}{2}\\&=1+2\sqrt{3}\end{aligned}

例子2

函数u=2xy−z2u=2xy-z^2在点A(2,−1,1)A(2,-1,1)处沿l⃗=(3,1,−1)\vec{l}=(3,1,-1)方向的方向导数为?

解:

cos⁡α=311,cos⁡β=111,cos⁡γ=−111\cos\alpha=\dfrac{3}{\sqrt{11}},\cos\beta=\dfrac{1}{\sqrt{11}},\cos\gamma=-\dfrac{1}{\sqrt{11}}

则方向导数为:

∂z∂l⃗∣(2,−1,1)=∂f∂x∣(2,−1,1)cos⁡α+∂f∂y∣(2,−1,1)cos⁡β+∂f∂z∣(2,−1,1)cos⁡γ=2y∣(2,−1,1)⋅311+2x∣(2,−1,1)⋅111+(−2)z∣(2,−1,1)⋅(−111)=−611+411+211=0\begin{aligned}\dfrac{\partial z}{\partial\vec{l}}\bigg|_{(2,-1,1)}&=\dfrac{\partial f}{\partial x}\bigg|_{(2,-1,1)}\cos \alpha+\dfrac{\partial f}{\partial y}\bigg|_{(2,-1,1)}\cos\beta+\dfrac{\partial f}{\partial z}\bigg|_{(2,-1,1)}\cos\gamma\\&=2y\bigg|_{(2,-1,1)}\cdot\dfrac{3}{\sqrt{11}}+2x\bigg|_{(2,-1,1)}\cdot\dfrac{1}{\sqrt{11}}+(-2)z\bigg|_{(2,-1,1)}\cdot(-\dfrac{1}{\sqrt{11}})\\&=\dfrac{-6}{\sqrt{11}}+\dfrac{4}{\sqrt{11}}+\dfrac{2}{\sqrt{11}}\\&=0\end{aligned}

3、导数与梯度的关系

方向导数的最值

最大值=∣gradf(x0,y0,z0)∣=|gradf(x_0,y_0,z_0)|(梯度的模)

最小值=−∣gradf(x0,y0,z0)∣=-|gradf(x_0,y_0,z_0)|(梯度的模)

例子1

f(x,y,z)=x2+y2+z2f(x,y,z)=x^2+y^2+z^2在点(1,1,1)(1,1,1)处的方向导数的最大值为?

解:

gradf(1,1,1)=(2,2,2)gradf(1,1,1)=(2,2,2)

则最大值为∣gradf(1,1,1)∣=4+4+4=23|gradf(1,1,1)|=\sqrt{4+4+4}=2\sqrt{3}

多元函数的极值及其求法

1、二元函数极值

1.1、无条件极值

①无条件极值:指对xx、yy取值范围无限制下,求z=f(x,y)z=f(x,y)的极值

②解法:

  1. 求一阶偏导,令其为0找驻点

    令{fx(x,y)=0,fy(x,y)=0⇒驻点(x0,y0)\begin{cases}f_x(x,y)=0,\\f_y(x,y)=0\end{cases}\Rightarrow\text{驻点}(x_0,y_0)

  2. 求二阶偏导,令A=fxx(x0,y0)A=f_{xx}(x_0,y_0),B=fxy(x0,y0)B=f_{xy}(x_0,y_0),C=fyy(x0,y0)C=f_{yy}(x_0,y_0)

  3. 判断:AC−B2=D⇒{D<0,则无极值D>0,有极值⇒{A<0,极大值,A>0,极小值.AC-B^2=D\Rightarrow\begin{cases}D<0,\text{则无极值}\\D>0,\text{有极值}\Rightarrow\begin{cases}A<0,\text{极大值},\\A>0,\text{极小值}.\end{cases}\end{cases}

例子1

求函数f(x,y)=x2−xy+y2−2x+yf(x,y)=x^2-xy+y^2-2x+y的极值.

解:

令{fx(x,y)=2x−y−2=0fy(x,y)=−x+2y+1=0⇒驻点(1,0)\begin{cases}f_x(x,y)=2x-y-2=0 \\f_y(x,y)=-x+2y+1=0\end{cases}\Rightarrow \text{驻点}(1,0)

A=fxx(x,y)=2,B=fxy(x,y)=−1,C=fyy(x,y)=2A=f_{xx}(x,y)=2,B=f_{xy}(x,y)=-1,C=f_{yy}(x,y)=2

AC−B2=3>0AC-B^2=3>0且A=2>0A=2>0,

故f(x,y)f(x,y)在点(1,0)(1,0)处有极小值

即f(1,0)=1−2=−1f(1,0)=1-2=-1

例子2

求f(x,y)=x3−4x2+2xy−y2f(x,y)=x^3-4x^2+2xy-y^2的极值.

解:

令{fx(x,y)=3x2−8x+2y=0,fy(x,y)=2x−2y=0驻点(2,2)与(0,0)\begin{cases}f_x(x,y)=3x^2-8x+2y=0,\\f_y(x,y)=2x-2y=0\end{cases}\text{驻点}(2,2)\text{与}(0,0)

A=fxx(x,y)=6x−8,B=fxy(x,y)=2,C=fyy(x,y)=−2A=f_{xx}(x,y)=6x-8,B=f_{xy}(x,y)=2,C=f_{yy}(x,y)=-2

在驻点(2,2)(2,2)处AC−B2=−12<0AC-B^2=-12<0,故在该点处无极值

在驻点(0,0)(0,0)处AC−B2=12>0AC-B^2=12>0且A=−8<0A=-8<0故在该点处有极大值

即f(0,0)=0f(0,0)=0

1.2、有条件极值

①条件极值:指对自变量xx、yy取值范围限制下,在条件ψ(x,y)=0\psi(x,y)=0下,求z=f(x,y)z=f(x,y)的极值。

②解法:

  1. ==确定目标函数f(x,y)f(x,y)及条件函数ψ(x,y)\psi(x,y)==

  2. 构造拉格朗日函数:F(x,y,λ)=f(x,y)+λψ(x,y)F(x,y,\lambda)=f(x,y)+\lambda\psi(x,y)

  3. 令{Fx=0Fy=0Fλ=0得驻点(x0,y0)\begin{cases}F_x=0\\ F_y=0 \\F_{\lambda}=0 \end{cases}\text{得驻点}(x_0,y_0)

  4. 得极值

例子1

求函数z=x2+y2z=x^2+y^2在直线x4+y3=1\dfrac{x}{4}+\dfrac{y}{3}=1上的极值.

解:

目标函数:x2+y2,x^2+y^2,条件函数:x4+y3−1=0\dfrac{x}{4}+\dfrac{y}{3}-1=0

令F(x,y,λ)=x2+y2+λ(x4+y3−1)F(x,y,\lambda)=x^2+y^2+\lambda(\dfrac{x}{4}+\dfrac{y}{3}-1)

令{Fx=2x+14λ=0Fy=2y+13λ=0Fλ=x4+y3−1=0得驻点(3625,4825)\begin{cases}F_x=2x+\dfrac{1}{4}\lambda=0\\ F_y=2y+\dfrac{1}{3}\lambda=0 \\F_{\lambda}=\dfrac{x}{4}+\dfrac{y}{3}-1=0 \end{cases}\text{得驻点}(\dfrac{36}{25},\dfrac{48}{25})

故z=x2+y2z=x^2+y^2在(3625,4825)(\dfrac{36}{25},\dfrac{48}{25})处取得极值为14425\dfrac{144}{25}

例子2

回顾:点(x0,y0)(x_0,y_0)到Ax+By+C=0Ax+By+C=0的距离dd

d=∣Ax0+By0+C∣A2+B2d=\dfrac{|Ax_0+By_0+C|}{\sqrt{A^2+B^2}}

问:在椭圆x2+4y2=4x^2+4y^2=4求一点,使其到直线2x+3y−6=02x+3y-6=0的距离最短.

解:

设P(x,y)P(x,y)在椭圆上,则该点到直线的距离为:

d=∣2x+3y−6∣13d=\dfrac{|2x+3y-6|}{\sqrt{13}}

目标方程2x+3y−62x+3y-6(目标是求距离,故dd为目标方程,由于dd的分母是常数,故分子做为目标方程方便计算),条件方程:x2+4y2−4=0x^2+4y^2-4=0(点PP在椭圆上,故其是条件方程)

令F(x,y,λ)=2x+3y−6+λ(x2+4y2−4)F(x,y,\lambda)=2x+3y-6+\lambda(x^2+4y^2-4)

{Fx=2+2xλ=0Fy=3+8yλ=0Fλ=x2+4y2−4=0得驻点(−85,−35)与(85,35)\begin{cases}F_x=2+2x\lambda=0\\ F_y=3+8y\lambda=0 \\F_{\lambda}=x^2+4y^2-4=0 \end{cases}\text{得驻点}(-\dfrac{8}{5},-\dfrac{3}{5})\text{与}(\dfrac{8}{5},\dfrac{3}{5})

①当(x,y)=(85,35)(x,y)=(\dfrac{8}{5},\dfrac{3}{5})时,代入d=113d=\dfrac{1}{\sqrt{13}}

②当(x,y)=(−85,−35)(x,y)=(-\dfrac{8}{5},-\dfrac{3}{5})时,代入d=1113d=\dfrac{11}{\sqrt{13}}

故x2+4y2=4x^2+4y^2=4在点(85,35)(\dfrac{8}{5},\dfrac{3}{5})取得唯一极值,使其到直线2x+3y−6=02x+3y-6=0的距离最短,为1313\dfrac{\sqrt{13}}{13}.

特训

1、下列四个命题正确的是( ).(\ ).

A.若函数f(x,y)f(x,y)在点(x,y)(x,y)处可微分,则f(x,y)f(x,y)在该点处一定连续

B.若函数f(x,y)f(x,y)在点(x,y)(x,y)处连续,则f(x,y)f(x,y)在该点处一定可微分

C.若函数f(x,y)f(x,y)在点(x,y)(x,y)的偏导数∂f∂x\dfrac{\partial f}{\partial x}及∂f∂y\dfrac{\partial f}{\partial y}均存在,则f(x,y)f(x,y)在该点处一定连续

D.若函数f(x,y)f(x,y)在点(x,y)(x,y)的偏导数∂f∂x\dfrac{\partial f}{\partial x}及∂f∂y\dfrac{\partial f}{\partial y}均存在,则f(x,y)f(x,y)在该点处一定可微分

解析:可微→\rightarrow连续,连续↛\nrightarrow可微,偏导存在↛\nrightarrow连续↛\nrightarrow可微分,故选A。

2、设f(x.y)f(x.y)在点(a,b)(a,b)处有偏导数,则lim⁡h→0f(a+h,b)−f(a−h,b)h=().\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a-h,b)}{h}=().

A.00 B.2fx(a,b)2f_x(a,b) C.fx(a,b)f_x(a,b) D.fy(a,b)f_y(a,b)

解析:

lim⁡h→0f(a+h,b)−f(a−h,b)h=lim⁡h→0f(a+h,b)−f(a,b)+f(a,b)−(f(a−h,b)h=lim⁡h→0f(a+h,b)−f(a,b)h+lim⁡h→0f(a,b)−f(a−h,b)h=lim⁡h→0f(a+h,b)−f(a,b)h−lim⁡h→0f(a−h,b)−f(a,b)h=lim⁡h→0f(a+h,b)−f(a,b)h+lim⁡h→0f(a−h,b)−f(a,b)−h=2fx(a,b).\begin{aligned}\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a-h,b)}{h}&=\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a,b)+f(a,b)-(f(a-h,b)}{h}\\&=\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a,b)}{h}+\lim\limits_{h\rightarrow 0}\dfrac{f(a,b)-f(a-h,b)}{h}\\&=\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a,b)}{h}-\lim\limits_{h\rightarrow 0}\dfrac{f(a-h,b)-f(a,b)}{h}\\&=\lim\limits_{h\rightarrow 0}\dfrac{f(a+h,b)-f(a,b)}{h}+\lim\limits_{h\rightarrow 0}\dfrac{f(a-h,b)-f(a,b)}{-h}\\&=2f_x(a,b).\end{aligned}

3、已知(axy3−y2cos⁡x)dx+(bysin⁡x+3x2y2)dy(axy^3-y^2\cos x)dx+(by\sin x+3x^2y^2)dy为函数f(x,y)f(x,y)的全微分,且函数f(x,y)f(x,y)的二阶偏导数连续,则aa和bb的值分别为( ).

A.-2和2 B.2和-2 C.2和2 D.-2和-2

解析:

由题意得∂f∂x=axy3−y2cos⁡x,∂f∂y=bysin⁡x+3x2y2\dfrac{\partial f}{\partial x}=axy^3-y^2\cos x,\dfrac{\partial f}{\partial y}=by\sin x+3x^2y^2,

又函数f(x,y)f(x,y)的二阶偏导数连续,即∂2f∂x∂y=∂2f∂y∂x\dfrac{\partial^2f}{\partial x\partial y}=\dfrac{\partial^2f}{\partial y\partial x},

∂2f∂x∂y=∂(axy3−y2cos⁡x)∂y=3axy2−2ycos⁡x\dfrac{\partial^2f}{\partial x\partial y}=\dfrac{\partial(axy^3-y^2\cos x)}{\partial y}=3axy^2-2y\cos x,

∂2f∂y∂x=∂(bysin⁡x+3x2y2)∂x=bycos⁡x+6xy2\dfrac{\partial^2f}{\partial y\partial x}=\dfrac{\partial(by\sin x+3x^2y^2)}{\partial x}=by\cos x+6xy^2,

3axy2−2ycos⁡x=bycos⁡x+6xy23axy^2-2y\cos x=by\cos x+6xy^2,对应项的系数相等,得a=2,b=−2a=2,b=-2.

4、设二元函数f(x,y)f(x,y)在点(2,0)(2,0)的某个领域有定义,且当x≠2x\neq 2时,f(x,0)−f(2,0)x−2=−(x+1)\dfrac{f(x,0)-f(2,0)}{x-2}=-(x+1),则fx(2,0)=_f_x(2,0)=\_.

解析:

fx(2,0)=lim⁡x→2f(x,0)−f(2,0)x−0=−lim⁡x→2(x+1)=−3f_x(2,0)=\lim\limits_{x\rightarrow 2}\dfrac{f(x,0)-f(2,0)}{x-0}=-\lim\limits_{x\rightarrow 2}(x+1)=-3.

5、设函数z=(x+2y)xz=(x+2y)^x,求∂z∂x\dfrac{\partial z}{\partial x},∂z∂y\dfrac{\partial z}{\partial y}.

解析:

∂z∂x=(x+2y)xln⁡(x+2y)⋅1=(x+2y)xln⁡(x+2y)\dfrac{\partial z}{\partial x}=(x+2y)^x\ln(x+2y)\cdot 1=(x+2y)^x\ln(x+2y),

∂z∂y=x⋅(x+2y)x−1⋅2=2x(x+2y)x−1.\dfrac{\partial z}{\partial y}=x\cdot(x+2y)^{x-1}\cdot 2=2x(x+2y)^{x-1}.

6、设函数z=xf(xy)+ϕ(x2,y)z=xf(xy)+\phi(x^2,y),且ff为二阶可导函数,ϕ(x,y)\phi(x,y)可微,求∂2z∂x∂y\dfrac{\partial^2z}{\partial x\partial y}.

解析:

∂z∂x=f(xy)+xyf′(xy)+2xϕ1′,\dfrac{\partial z}{\partial x}=f(xy)+xyf'(xy)+2x\phi'_1,

∂2z∂x∂y=f′(xy)⋅x+xf′(xy)+xyf′′(xy)⋅x+2xϕ12′′=2xf′(x,y)+x2yf′′(xy)+2xϕ12′′\dfrac{\partial^2z}{\partial x\partial y}=f'(xy)\cdot x+xf'(xy)+xyf''(xy)\cdot x+2x\phi''_{12}=2xf'(x,y)+x^2yf''(xy)+2x\phi''_{12}.

觉得有用就点个赞,不用登录也能读

评论

0 条
登录 之后才能评论
—
还没有评论,来说第一句