manthanein 发表于 2016-11-28 22:13:41

这样的直线是否一定是切线呢?

在坐标平面上给定曲线\(x=\phi(t), y=\psi(t)\)。\(t\)的定义域另外指定为\(D\)。\(\phi(t)\)和\(\psi(t)\)在定义域\(D\)上都连续。
取\(t=a \in D\),\(a\)为常数。构造直线\(L\),\(A\)、\(B\)均为常数:
\(L:A+B=0\)
取\(u \in D\),构造:
\(F(u)=[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2\)
\(G(u)=\D \frac{\D \abs{A[\phi(u)-\phi(a)]+B[\psi(u)-\psi(a)]}}{\sqrt{A^2+B^2}\cdot F(u)}\)
已知:
对于任意给定的正数\(\epsilon\),总存在正数\(\delta\),使得\(F(u) \lt \delta\)时,一定有\(G(u) \lt \epsilon\)。
求可能的直线\(L\)。
从几何角度看,直线\(L\)应该是\(t=a\)处的切线。
但是几何直观未必正确,能否有一个严格的代数证明呢?
当然,也有可能不是切线,如果不是切线,又会是什么?

补充内容 (2016-12-1 22:52):
\(F(u)=\sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}\)更正

补充内容 (2016-12-1 22:53):
\(D\)为一个区间

补充内容 (2016-12-2 21:52):
\(u\)不在给定直线上

manthanein 发表于 2016-12-1 21:59:01

本帖最后由 manthanein 于 2016-12-1 22:19 编辑

先试试吧。
先假设:\(\phi(u) \ne \phi(a)\),\(\psi(u) \ne \psi(a)\),\(B \ne0\),\(a\)不在定义域边界上
令\(k=\D -\frac{A}{B}\)
\(G(u)\)
\(=\D \frac{\D \abs{A[\phi(u)-\phi(a)]+B[\psi(u)-\psi(a)]}}{\sqrt{A^2+B^2}\cdot F(u)}\)
\(=\D \frac{\D \abs{-Bk[\phi(u)-\phi(a)]+B[\psi(u)-\psi(a)]}}{\sqrt{B^{2}k^{2}+B^2}\cdot F(u)}\)
\(=\D \frac{\D \abs{B}\abs{-k[\phi(u)-\phi(a)]+[\psi(u)-\psi(a)]}}{\abs{B} \sqrt{k^{2}+1}\cdot F(u)}\)
\(=\D \frac{\D \abs{-k[\phi(u)-\phi(a)]+[\psi(u)-\psi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)

manthanein 发表于 2016-12-1 22:19:18

微积分理论告诉我们,此时存在切线的充分必要条件是存在常数\(C\)满足:
\(\D \lim_{u \to a} \D \frac{\psi(u)-\psi(a)}{\phi(u)-\phi(a)}=C\)

manthanein 发表于 2016-12-1 22:25:33

也就是说:
对于任意给定的正数\(\lambda_1\),存在正数\(\gamma_1\)
使得只要\(u\)满足不等式\(0 \lt \abs{u-a} \lt \gamma_1\),一定有\(\abs{\D \frac{\psi(u)-\psi(a)}{\phi(u)-\phi(a)}-C} \lt \lambda_1\)

manthanein 发表于 2016-12-1 22:42:39

令\(k=C+\zeta\)
\(G(u)\)
\(=\D \frac{\D \abs{-(C+\zeta)[\phi(u)-\phi(a)]+[\psi(u)-\psi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)
\(= \D \frac{\D \abs{[\psi(u)-\psi(a)]-(C+\zeta)[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)
\(= \D \frac{\D \abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}-\zeta[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)
\(\le \D \frac{\D \abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}}+\abs{\zeta[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)
\(= \D \frac{\D \abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}}+\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)

manthanein 发表于 2016-12-1 22:45:58

另一方面:
\(G(u) \ge \D \frac{\D \abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}}-\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)

manthanein 发表于 2016-12-1 23:00:04

\(\abs{\D \frac{\psi(u)-\psi(a)}{\phi(u)-\phi(a)}-C} \lt \lambda_1\)
\(\D \frac{\abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}}}{\abs{\phi(u)-\phi(a)}} \lt \lambda_1\)
\(\abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}} \lt \lambda_1{\abs{\phi(u)-\phi(a)}}\)

manthanein 发表于 2016-12-1 23:05:41

\(G(u) \le \D \frac{\D \abs{[\psi(u)-\psi(a)]-C{[\phi(u)-\phi(a)]}}+\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)} \lt \frac{\lambda_1{\abs{\phi(u)-\phi(a)}}+\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)} \)

manthanein 发表于 2016-12-1 23:37:12

\(\D \frac{\lambda_1{\abs{\phi(u)-\phi(a)}}+\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot F(u)}\)
\(=\D \frac{\lambda_1{\abs{\phi(u)-\phi(a)}}+\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot \sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}}\)
\(=\D \frac{\lambda_1{\abs{\phi(u)-\phi(a)}}}{\sqrt{k^{2}+1}\cdot \sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}}+\frac{\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot \sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}}\)

manthanein 发表于 2016-12-1 23:43:45

\(\D \frac{\lambda_1{\abs{\phi(u)-\phi(a)}}}{\sqrt{k^{2}+1}\cdot \sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}}+\frac{\abs{\zeta}\abs{[\phi(u)-\phi(a)]}}{\sqrt{k^{2}+1}\cdot \sqrt{[\phi(u)-\phi(a)]^2+[\psi(u)-\psi(a)]^2}}\)
\(= \D \frac{\lambda_1}{\sqrt{k^{2}+1}\cdot \sqrt{1+[\D \frac{\psi(u)-\psi(a)}{\phi(u)-\phi(a)}]^2}}+\frac{\abs{\zeta}}{\sqrt{k^{2}+1}\cdot \sqrt{1+[\D \frac{\psi(u)-\psi(a)}{\phi(u)-\phi(a)}]^2}}\)
页: [1] 2
查看完整版本: 这样的直线是否一定是切线呢?