(设f(x)是[a,b]上连续函数,则f(x)在[a,b]上必然一致连续\)
(证明:因为f(x)在[a,b]上连续,所以任取[a,b]内一点x_{0},任给frac{epsilon}{2}>0)
(existsdelta(x_{0})>0,对于任何xin[a,b],且异于x_{0},若|x-x_{0}|<delta,有|f(x)-f(x_{0})|<epsilon)
(因为这个delta与x_{0}的选取有关,对于同一个epsilon,不同位置的点,其对应的delta不同)
(设frac{delta_{x}}{2}对应的邻域为U(x,frac{delta(x)}{2}))
(设[a,b]上所有点的邻域集合为I={U(x,frac{delta}{2}),xin[a,b]})
(则I构成对区间[a,b]的完全开覆盖)
(因为[a,b]是闭区间,所以,根据有限开覆盖原理,在I内,存在有限个开覆盖,可以完全覆盖[a,b])
(设这个有限覆盖组成的集合为M={U(x_{k},frac{delta}{2}),k in N^+,x_{k}in[a,b]})
(forall x_{1},x_{2}in[a,b],设|x_{1}-x_{2}|<frac{delta}{2})
(因为M完全覆盖[a,b],所以x_{1}必属于某点x_{k}的邻域U(x_{k},frac{delta}{2})cap[a,b],)
(因此,|x_{1}-x_{k}|<frac{delta}{2})
(|x_{2}-x_{k}|<|x_{2}-x_{1}|+|x_{1}-x_{l}|=frac{delta}{2}+frac{delta}{2}=delta)
(因为x_{1},x_{2}均在x_{k}的邻域U(x_{k},delta)内,由函数的连续性,可知|f(x_{1})-f(x_{k})|<frac{epsilon}{2},|f(x_{2})-f(x_{k})|<frac{epsilon}{2})
(可得:|f(x_{1})-f(x_{2})|<|f(x_{1})-f(x_{k})|+|f_x{2}-f_x{k}|=frac{epsilon}{2}+frac{epsilon}{2}=epsilon)
(即forallepsilon>0,存在frac{delta}{2},若x_{1},x_{2}in [a,b],且|x_{1}-x_{2}|<frac{delta}{2})
(则quad |f(x_{1})-f(x_{2})|<epsilon)
证毕
(说明,只要给定epsilon,则对应的delta即确定,任何[a,b]内距离小于delta的两点,其函数差必然<epsilon)
(与这两点位置无关,仅与两点距离有关)
(只有闭区间才能使用有限覆盖)
下图是知乎网友提供的课本证明https://www.zhihu.com/question/56393706
其中的k,应为R