zoukankan      html  css  js  c++  java
  • 【数学】求导

    导数

    对于函数 (fleft(x ight)),定义其导数(f'left(x ight))

    (f'left(x ight)) 可以看作 (fleft(x ight)) 在相应位置的斜率。

    多项式函数的求导

    对于一个多项式函数 (fleft(x ight) = sum_{i=0} a_i x^i),这个函数的导数(f'left(x ight) = sum_{i = 0} i cdot a_i x ^ {i - 1})(系数乘上指数,指数上减一)。

    特殊函数导数

    [left(sinleft(x ight) ight)' = cosleft(x ight) ]

    [left(cosleft(x ight) ight)' = -sinleft(x ight) ]

    [left(lnleft(x ight) ight)' = dfrac{1}{x} ]

    [ ext{指数函数:}left(a^x ight)' = a^xlnleft(a ight) ext{(将会在后文给出证明)} ]

    [ ext{特别的:}left(e^x ight)' = e^xlnleft(e ight) = e^x cdot 1 = e^x ]

    一般函数的求导

    对于一个一般函数 (fleft(x ight)),其在 (x) 处的导数为

    [lim_{Delta x o 0} dfrac{fleft(x + Delta x ight) - fleft(x ight)}{Delta x} ]

    和积商求导法则

    和差

    [left(fleft(x ight) pm gleft(x ight) ight)' = f'left(x ight) pm g'left(x ight) ]

    证明略

    [left(fleft(x ight)gleft(x ight) ight)' = f'left(x ight)gleft(x ight) + fleft(x ight)g'left(x ight) ]

    证明:

    [left(fleft(x ight)gleft(x ight) ight)' = lim_{Delta x o 0} dfrac{fleft(x + Delta x ight)gleft(x + Delta x ight) - fleft(x ight)gleft(x ight)}{Delta x} ]

    [= lim_{Delta x o 0} dfrac{left(fleft(x + Delta x ight)gleft(x + Delta x ight) - fleft(x ight)gleft(x + Delta x ight) ight) + left(fleft(x ight)gleft(x + Delta x ight) - fleft(x ight)gleft(x ight) ight)}{Delta x} ]

    [= lim_{Delta x o 0} dfrac{fleft(x + Delta x ight) - fleft(x ight)}{Delta x}gleft(x + Delta x ight) + fleft(x ight)lim_{Delta x o 0}dfrac{gleft(x + Delta x ight) - gleft(x ight)}{Delta x} ]

    [= f'left(x ight)gleft(x ight) + fleft(x ight)g'left(x ight) ]

    [left(dfrac{fleft(x ight)}{gleft(x ight)} ight)' = dfrac{f'left(x ight)gleft(x ight) - fleft(x ight)g'left(x ight)}{g^2left(x ight)} ]

    证明:

    [left(dfrac{fleft(x ight)}{gleft(x ight)} ight)' = lim_{Delta x o 0} dfrac{dfrac{fleft(x + Delta x ight)}{gleft(x + Delta x ight)} - dfrac{fleft(x ight)}{gleft(x ight)}}{Delta x} ]

    [= lim_{Delta x o 0} dfrac{fleft(x + Delta x ight)gleft(x ight) - gleft(x + Delta x ight)fleft(x ight)}{gleft(x +Delta x ight)gleft(x ight)Delta x} ]

    [= lim_{Delta x o 0} dfrac{left(fleft(x + Delta x ight)gleft(x ight) - fleft(x ight)gleft(x ight) ight) - left(gleft(x + Delta x ight)fleft(x ight) - fleft(x ight)gleft(x ight) ight)}{gleft(x +Delta x ight)gleft(x ight)Delta x} ]

    [= lim_{Delta x o 0} dfrac{dfrac{fleft(x + Delta x ight) - fleft(x ight)}{Delta x}gleft(x ight) - dfrac{gleft(x + Delta x ight) - gleft(x ight)}{Delta x}fleft(x ight)}{gleft(x +Delta x ight)gleft(x ight)} ]

    [= dfrac{f'left(x ight)gleft(x ight) - fleft(x ight)g'left(x ight)}{g^2left(x ight)} ]

    复合函数

    [left(fleft(gleft(x ight) ight) ight)' = f'left(gleft(x ight) ight)g'left(x ight) ]

    证明:

    其他

    那么,关于前文“特殊函数求导”的指数函数求导,可以简单给出证明了。

    [ ext{设} y = a^x ]

    [ ext{则有} lnleft(y ight) = xlnleft(a ight) ]

    [left(lnleft(y ight) ight)' = left(xlnleft(a ight) ight)' ]

    [dfrac{y'}{y} = 1 cdot lnleft(a ight) + 0 ]

    (这一步左边使用了复合函数求导法则,右边使用了积的求导法则。)

    再将右边的分母上的 (y) 乘到右边,并替换回 (a^x),就得到了:

    [y' = a^xlnleft(a ight) ]

    鸣谢

    基本求导法则证明 by E-748

  • 相关阅读:
    从读者角度来看Blog
    NDuiker项目第3天
    IssueVision的List控件源码分析
    测试一个网站的想法
    IssueVision的PaneCaption控件源码分析
    技术研究的时候不要忘了“集成创新”
    人脸识别活体检测之张张嘴和眨眨眼
    jsp>Session 小强斋
    jsp>Request对象 小强斋
    jsp>四种作用域 小强斋
  • 原文地址:https://www.cnblogs.com/zimujun/p/14527700.html
Copyright © 2011-2022 走看看