欢迎光临散文网 会员登陆 & 注册

浅谈高等数学(8)

2022-02-25 21:14 作者:-YD-LM-  | 我要投稿

计算是数学之基础。

第八期  导数与微分的运算法则

       在高等数学的教材中,微分是放在导数完全结束后讲述的;但若是我们提前了解了微分,那么涉及导数的许多公式也就明晰许多了。然而,我们不得不使用许多严谨的计算,初见时可能会有一些枯燥。与此同时,也希望读者能体会严谨中蕴含的整洁与美观。

一、函数和的导数

设函数u%3Du(x)%2C%5C%20v%3Dv(x),求(u%2Bv)'。推导如下:

(u%2Bv)'%3D%5Cfrac%7Bu(x%2B%5Cmathrm%20dx)%2Bv(x%2B%5Cmathrm%20dx)-u(x)-v(x)%7D%7B%5Cmathrm%20dx%7D

                 %3D%5Cfrac%7Bu(x%2B%5Cmathrm%20dx)-u(x)%7D%7B%5Cmathrm%20dx%7D%2B%5Cfrac%7Bv(x%2B%5Cmathrm%20dx)-v(x)%7D%7B%5Cmathrm%20dx%7D

                 %3Du'%2Bv'.

因此,函数和的导数等于它们导数的和。使用数学归纳法可以证明多个函数的情况,发现仍旧如此;不了解数学归纳法的请往下翻。

二、函数积的导数

设据同上,求(uv)'。推导如下:

(uv)'%3D%5Cfrac%7Bu(x%2B%5Cmathrm%20dx)v(x%2B%5Cmathrm%20dx)-u(x)v(x)%7D%7B%5Cmathrm%20dx%7D

           %3D%5Cfrac%7Bu(x%2B%5Cmathrm%20dx)%C2%B7%5Bv(x%2B%5Cmathrm%20dx)-v(x)%5D%2Bv(x)%C2%B7%5Bu(x%2B%5Cmathrm%20dx)-u(x)%5D%7D%7B%5Cmathrm%20dx%7D

           %3Du(x%2B%5Cmathrm%20dx)%C2%B7v'%2Bv(x)%C2%B7u'

           %3Du'v%2Buv'.

通过这个公式,我们对多元情况进行尝试,容易得到

(uvw)'%3Du'vw%2Buv'w%2Buvw'%2C

(uvwx)'%3Du'vwx%2Buv'wx%2Buvw'x%2Buvwx'.

于是,我们猜想:

(y_1y_2y_3%5C%20...y_n)'%3D%5Csum_%7B%5Calpha_1%2B%5Calpha_2%2B%5Calpha_3%2B...%2B%5Calpha_n%3D1%5C%5C%20%0A%20%5C%20%5C%20%5C%20%5Calpha_1%2C%5Calpha_2%2C%5Calpha_3%2C...%2C%5Calpha_n%5Cge0%7Dy_1%5E%7B(%5Calpha_1)%7Dy_2%5E%7B(%5Calpha_2)%7Dy_3%5E%7B(%5Calpha_3)%7D......y_n%5E%7B(%5Calpha_n)%7D.

先解释一下这个式子:由于%5Calpha_1%2B%5Calpha_2%2B...%2B%5Calpha_n%3D1且他们之和均为非负数,又因为求和号的意义,他们均为自然数——因此只有可能是其中一个为1,其余的均为0.式子中,我们定义y%5E%7B(0)%7D%3Dy%2Cy%5E%7B(1)%7D%3Dy'.至于%5Calpha%0A取其他数时的情况,我们以后会讲述。这个式子同样能用数学归纳法证明,数学归纳法是说:

若(1)n%3D1时,命题p为真;

(2)若n%3Dk%5C%20(k%5Cin%20%5Cmathrm%20N%5E*)p为真,则n%3Dk%2B1p为真,

%5Cforall%20n%5Cin%5Cmathrm%20N%5E*p为真。自然数集亦可类比。

下面是证明:首先,n%3D1时显然成立。

(y_1y_2y_3%5C%20...y_n)'%3D%5Csum_%7B%5Calpha_1%2B%5Calpha_2%2B%5Calpha_3%2B...%2B%5Calpha_n%3D1%5C%5C%20%0A%20%5C%20%5C%20%5C%20%5Calpha_1%2C%5Calpha_2%2C%5Calpha_3%2C...%2C%5Calpha_n%5Cge0%7Dy_1%5E%7B(%5Calpha_1)%7Dy_2%5E%7B(%5Calpha_2)%7Dy_3%5E%7B(%5Calpha_3)%7D......y_n%5E%7B(%5Calpha_n)%7D,则

(y_1y_2y_3...y_%7Bn%2B1%7D)'%3Dy_%7Bn%2B1%7D%5E%7B(1)%7D%C2%B7%5Csum_%7B%5Calpha_1%2B%5Calpha_2%2B%5Calpha_3%2B...%2B%5Calpha_n%3D1%5C%5C%20%0A%20%5C%20%5C%20%5C%20%5Calpha_1%2C%5Calpha_2%2C%5Calpha_3%2C...%2C%5Calpha_n%5Cge0%7Dy_1%5E%7B(%5Calpha_1)%7Dy_2%5E%7B(%5Calpha_2)%7Dy_3%5E%7B(%5Calpha_3)%7D......y_n%5E%7B(%5Calpha_n)%7D

                                     %2By_1%5E%7B(0)%7Dy_2%5E%7B(0)%7Dy_3%5E%7B(0)%7D%E2%80%A6y_n%5E%7B(0)%7Dy_%7Bn%2B1%7D%5E%7B(1)%7D

%3D%5Csum_%7B%5Calpha_1%2B%5Calpha_2%2B%5Calpha_3%2B...%2B%5Calpha_%7Bn%2B1%7D%3D1%5C%5C%20%0A%20%5C%20%5C%20%5C%20%5Calpha_1%2C%5Calpha_2%2C%5Calpha_3%2C...%2C%5Calpha_%7Bn%2B1%7D%5Cge0%7Dy_1%5E%7B(%5Calpha_1)%7Dy_2%5E%7B(%5Calpha_2)%7Dy_3%5E%7B(%5Calpha_3)%7D......y_%7Bn%2B1%7D%5E%7B(%5Calpha_%7Bn%2B1%7D)%7D.

命题即得证。

三、复合函数的导数

设函数y%3Df%5B%20g(x)%5Dy%3Df(u)u%3Dg(x)两个可导函数复合而成,求%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20dx%7D

%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20dx%7D%3D%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20du%7D%C2%B7%5Cfrac%7B%5Cmathrm%20du%7D%7B%5Cmathrm%20dx%7D%3Df'(u)%C2%B7g'(x).这是较为直观的一条结论。

同样地,可以证明由多个函数u_%7Bn-1%7D%3Df_n(x)%2Cu_%7Bn-2%7D%3Df_%7Bn-1%7D(u_%7Bn-1%7D)...%2Cu_1%3Df_2(u_2)%2Cy%3Df_1(u_1)复合而成的函数y%3D(f_1%5Ccirc%20f_2%5Ccirc%E2%80%A6%5Ccirc%20f_n)(x)的导数为%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20dx%7D%3D%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20du_1%7D%C2%B7%5Cfrac%7B%5Cmathrm%20du_1%7D%7B%5Cmathrm%20du_2%7D%C2%B7%5C%20......%5C%20%C2%B7%5Cfrac%7B%5Cmathrm%20du_%7Bn-2%7D%7D%7B%5Cmathrm%20du_%7Bn-1%7D%7D%C2%B7%5Cfrac%7B%5Cmathrm%20du_%7Bn-1%7D%7D%7B%5Cmathrm%20dx%7D.

四、函数商的导数

设函数u%3Du(x)%2Cv%3Dv(x),且v(x)%5Cnot%3D0,求%5Cbigg(%5Cfrac%20uv%5Cbigg)'

它是可以使用与函数积的导数相同的方法求解的,只是多了通分的步骤。读者可以尝试证明,并且教材上也有。这里给出另一种方法:

首先,求y%3D%5Cfrac1x的导数:

y'%3D%5Cfrac%7B%5Cfrac1%7Bx%2B%5Cmathrm%20dx%7D-%5Cfrac1x%7D%7B%5Cmathrm%20dx%7D%3D%5Cfrac%7Bx-(x%2B%5Cmathrm%20dx)%7D%7Bx(x%2B%5Cmathrm%20dx)%5Cmathrm%20dx%7D%3D-%5Cfrac%7B%5Cmathrm%20dx%7D%7Bx%5E2%5Cmathrm%20dx%7D%3D-%5Cfrac1%7Bx%5E2%7D.

于是,y%3D%5Cfrac1%7Bv%7Dx的导数为

-%5Cfrac%7Bv'%7D%7Bv%5E2%7D.

最后,y%3D%5Cfrac%20uvx的导数为

y'%3D%5Cbig(u%C2%B7%5Cfrac1v%5Cbig)'%3D%5Cfrac%7Bu'%7Dv-%5Cfrac%7Buv'%7D%7Bv%5E2%7D%3D%5Cfrac%7Bu'v-uv'%7D%7Bv%5E2%7D.

五、反函数的导数

其与复合函数一样也非常直观。设函数x%3Df(y)在某区间内具有反函数y%3Df%5E%7B-1%7D(x)

%5Cfrac%7B%5Cmathrm%20dy%7D%7B%5Cmathrm%20dx%7D%3D%5Cfrac%7B1%7D%7B%5Cfrac%7B%5Cmathrm%20dx%7D%7B%5Cmathrm%20dy%7D%7D%3D%5Cfrac%7B1%7D%7Bf'(y)%7D.

需要注意的是,这里是用原函数对y求导,而非x

对于其几何意义,还有一种深入的理解:

图1  反函数求导的几何意义

如图是y_1%3D%5Csin%20x(蓝)与y_2%3D%5Carcsin%20x(红)的图像,由反函数的意义可知它们的图像是关于y%3Dx对称的。假如我们要考察反函数在B(%5Csin%20x_0%2Cx_0)(为了方便起见,我们只考虑(0%2C%5Cfrac%5Cpi2)内的部分)处的导数,则对应的,原函数上的点应为A(x_0%2C%5Csin%20x_0)。原函数切线交x轴于C,交y轴于E;反函数切线交x轴于F,交y轴于D。由于关于y%3Dx对称,则可知%E2%88%A0ECO%3D%5Cfrac%5Cpi2-%E2%88%A0FDO.,即%5Ctan%E2%88%A0ECO%5Ctan%E2%88%A0FDO%3D1.而这两个角恰好分别是原函数与反函数的倾角,故有

y_1'%5Cbigg%7C_%7Bx%3Dx_0%7D%C2%B7y_2'%5Cbigg%7C_%7Bx%3D%5Csin%20x_0%7D%3D1.


至于微分运算法则,只需利用%5Cmathrm%20dy%3Df'(x)%5Cmathrm%20dx变形即可,此处就不详述了。

浅谈高等数学(8)的评论 (共 条)

分享到微博请遵守国家法律