Excercise. (Principle of Mathematical Analysis 5.14)
Let $f$ be a differentiable real function defined in $(a, b)$. Prove that $f$ is convex if and only if $f'$ is monotonically increasing.
Answer.
Recall: Definition of convex - $f$ is convex if and only if
\[(1-\lambda)f(a)+\lambda f(b) \geq f((1-\lambda)a+\lambda b) \] for all $0\leq\lambda\leq1$ and $a, b$. WLOG, set $a\leq b$. Define
\[L(x):=f(a)\cfrac{b-x}{b-a}+f(b)\cfrac{x-a}{b-a}\]
Suppose $f$ is convex and $a\leq x\leq b$, and $\lambda=\cfrac{x-a}{b-a}$, then $0\leq\lambda\leq1$ and $L(x)=(1-\lambda)f(a)+\lambda f(b)\geq f((1-\lambda)a+\lambda b)=f(x)$. Note that $L'(x)=\cfrac{f(b)-f(a)}{b-a}$. Define
\[g(x):=L(x)-f(x)\]
Then $g(x)\geq0$ for $a\leq x\leq b$ and $g(a)=0, g(b)=0$. Also, $g$ is a differentiable real function. If $g'(a)<0$, then there exists $x_1>a$ s.t. $g(x_1)<0$. But $g(x)\geq0$ for $a\leq x\leq b$, so $g'(a)\geq0$. By the same logic, $g'(b)\leq0$. So, $g'(a)=L'(a)-f'(a)\geq0, f'(a)\leq L'(a)$ and $g'(b)=L'(b)-f'(b)\leq0, f'(b)\geq L'(b)$, $f'(a)\leq L'(a)=L'(b)\leq f'(b)$, concluding that if $a<b$, then $f'(a)\leq f'(b)$.
Now, if $f'$ is not convex, for some $a, b$, there exists an $x\in(a,b)$ s.t. $L(x)<f(x)$, so $g(x)<0$. So $g(x)-g(a)=g(x)<0, g(b)-g(x)=-g(x)>0$, and there exist $x_1\in(a,x), x_2\in(x,b)$ s.t. $g'(x_1)=\cfrac{g(x)-g(a)}{x-a}<0, g'(x_2)=\cfrac{g(b)-g(x)}{b-x}>0$. So $g'(x_2)>g'(x_1), f
(x_1)>f'(x_2)$. This concludes $f'$ is not monotonically increasing.
End of Proof.
'math > analysis' 카테고리의 다른 글
| If f' is monotonically increasing, f(x)/x is too. (0) | 2025.10.17 |
|---|