2.5 多元复合函数的求导法则

高等数学 A2



周维祺

简单情况:z=f(x(t),y(t))z=f(x(t),y(t)),求dzdt\frac{dz}{dt}

前提

  • z=f(x,y)z=f(x,y)的偏导数都存在且偏导数都连续

  • 参数方程x=x(t),y=y(t)x=x(t),y=y(t)对于tt可导

  • zz对于tt可导,导数为z(t)=zxdxdt+zydydtz'(t)=\frac{\partial z}{\partial x}\frac{dx}{dt}+\frac{\partial z}{\partial y}\frac{dy}{dt}

推导

  • z=f(x,y)z=f(x,y)的偏导数都存在且偏导数都连续
    \Rightarrow zz的全微分存在

  • tt有增量Δt\Delta t时,xyx,y也有增量:
    Δx=x(t+Δt)x(t)\Delta x=x(t+\Delta t)-x(t)
    Δy=y(t+Δt)y(t)\Delta y=y(t+\Delta t)-y(t)

  • zz的增量可以借助全微分:
    Δz=zxΔx+zyΔy+o((Δx)2+(Δy)2)\Delta z=\frac{\partial z}{\partial x}\Delta x+\frac{\partial z}{\partial y}\Delta y+o(\sqrt{(\Delta x)^2+(\Delta y)^2})

推导

  • 两边同除以Δt\Delta t,得
    ΔzΔt=zxΔxΔt+zyΔyΔt+o((Δx)2+(Δy)2)Δt\frac{\Delta z}{\Delta t}=\frac{\partial z}{\partial x}\frac{\Delta x}{\Delta t}+\frac{\partial z}{\partial y}\frac{\Delta y}{\Delta t}+\frac{o(\sqrt{(\Delta x)^2+(\Delta y)^2})}{\Delta t}

  • Δt0\Delta t\to 0
    ΔzΔtz(t)\frac{\Delta z}{\Delta t}\to z'(t), ΔxΔtx(t)=dxdt\quad\frac{\Delta x}{\Delta t}\to x'(t)=\frac{dx}{dt}, ΔyΔty(t)=dydt\quad\frac{\Delta y}{\Delta t}\to y'(t)=\frac{dy}{dt}

推导

  • (思考:为什么)
    (Δx)2+(Δy)2Δt=(ΔxΔt)2+(ΔyΔt)2(x(t))2+(y(t))2\frac{\sqrt{(\Delta x)^2+(\Delta y)^2}}{\Delta t}=\sqrt{(\frac{\Delta x}{\Delta t})^2+(\frac{\Delta y}{\Delta t})^2}\to\sqrt{\left(x'(t)\right)^2+\left(y'(t)\right)^2}

  • (思考:为什么)
    o((Δx)2+(Δy)2)Δt0\frac{o\left(\sqrt{(\Delta x)^2+(\Delta y)^2}\right)}{\Delta t}\to0

  • 综上,得 z(t)=zxdxdt+zydydtz'(t)=\frac{\partial z}{\partial x}\frac{dx}{dt}+\frac{\partial z}{\partial y}\frac{dy}{dt}

练习

  • z=exyz=e^{xy}

  • x=costx=\cos t, y=sinty=\sin t

  • 利用链式法则求z(t)z'(t)

一般情形

z=f(x1,,xn)z=f(x_1,\ldots,x_n)xk=xk(t1,,tm)x_k=x_k(t_1,\ldots,t_m),求ztj\frac{\partial z}{\partial t_j}

一般情形

  • z=f(x1,,xn)z=f(x_1,\ldots,x_n)的偏导数都存在且偏导数都连续

  • 各参数方程xk=xk(t1,,tm)x_k=x_k(t_1,\ldots,t_m)的各偏导都存在

  • zz对于tjt_j的偏导数为ztj=zx1x1tj++zxnxntj\frac{\partial z}{\partial t_j}=\frac{\partial z}{\partial x_1}\frac{\partial x_1}{\partial t_j}+\ldots+\frac{\partial z}{\partial x_n}\frac{\partial x_n}{\partial t_j}

需要掌握的情形

  • z=f(x,y)z=f(x,y)x=x(u,v),y=(u,v)x=x(u,v), y=(u,v)

  • z=f(x,y)z=f(x,y)的偏导数都存在且偏导数都连续

  • 设参数方程x=x(u,v)x=x(u,v), y=y(u,v)y=y(u,v)的各偏导都存在

例子

  • z=ex2+y2z=e^{x^2+y^2}, x=rcosθx=r\cos\theta, y=rsinθy=r\sin\theta

  • zx=2xex2+y2\frac{\partial z}{\partial x}=2xe^{x^2+y^2}, zy=2yex2+y2\frac{\partial z}{\partial y}=2ye^{x^2+y^2}
    xr=cosθ\frac{\partial x}{\partial r}=\cos\theta, xθ=rsinθ\frac{\partial x}{\partial \theta}=-r\sin\theta, yr=sinθ\frac{\partial y}{\partial r}=\sin\theta, yθ=rcosθ\frac{\partial y}{\partial\theta}=r\cos\theta

  • zr=zxxr+zyyr=2ex2+y2(xcosθ+ysinθ)=2rer2\frac{\partial z}{\partial r}=\frac{\partial z}{\partial x}\frac{\partial x}{\partial r}+\frac{\partial z}{\partial y}\frac{\partial y}{\partial r}=2e^{x^2+y^2}(x\cos\theta+y\sin\theta)=2re^{r^2}
    zθ=zxxθ+zyyθ=2rex2+y2(ycosθxsinθ)=0\frac{\partial z}{\partial \theta}=\frac{\partial z}{\partial x}\frac{\partial x}{\partial \theta}+\frac{\partial z}{\partial y}\frac{\partial y}{\partial \theta}=2re^{x^2+y^2}(y\cos\theta-x\sin\theta)=0

练习

  • z=exyz=e^{xy}

  • x=u+vx=u+v, y=uvy=u-v

  • 利用链式法则计算zzu,vu,v的偏导数

全微分的形式不变性

  • 给定z=f(x,y)z=f(x,y),全微分 dz=zxdx+zydydz=z_xdx+z_ydy

  • x=x(u,v)x=x(u,v), 则 dx=xudu+xvdvdx=x_udu+x_vdv
    y=y(u,v)y=y(u,v), 则 dy=yudu+yvdvdy=y_udu+y_vdv

  • zudu+zvdv=(zxxu+zyyu)du+(zxxv+zyyv)dv    =zx(xudu+xvdv)+zy(xvdv+yvdv)    =zxdx+zydy=dzz_udu+z_vdv=(z_xx_u+z_yy_u)du+(z_xx_v+z_yy_v)dv \\ \quad\quad\quad\quad\quad\;\;\,=z_x(x_udu+x_vdv)+z_y(x_vdv+y_vdv) \\ \quad\quad\quad\quad\quad\;\;\, =z_xdx+z_ydy=dz

全微分的形式不变性

dz=zxdx+zydy=zudu+zvdvdz=\frac{\partial z}{\partial x}dx+\frac{\partial z}{\partial y}dy=\frac{\partial z}{\partial u}du+\frac{\partial z}{\partial v}dv

练习

  • z=x+yz=x+y

  • x=uvx=uv, y=u2+v2y=u^2+v^2

  • 利用全微分的形式不变性计算zzu,vu,v的偏导数

小结

  • 单一变量的链式求导

  • 多个变量的链式求导

  • 全微分的形式不变性