I would like to demonstrate that for a concave downward function passing through the origin that the value of the derivative $\frac{dy}{dx}$ is always higher than $\frac{y}{x}$. As shown in figure, it is simple to understand graphically, but for a generical concave downward function, without having a specific function, I don't know how to start.
-
1$\begingroup$ the graph shows that the inequality is reversed $\endgroup$– G CabNov 15, 2021 at 17:54
-
$\begingroup$ The thing you are trying to prove is not true $\endgroup$– AndreiNov 15, 2021 at 18:00
-
1$\begingroup$ If you assume that the function passes through the origin then you should state so in the question. $\endgroup$– dxivNov 15, 2021 at 18:30
-
$\begingroup$ Sorry @GCab I wrote it badly in the figure, Now I fixed it. $\endgroup$– Luca MirtaniniNov 15, 2021 at 19:38
-
$\begingroup$ @dxiv sorry I forgot to mention it. $\endgroup$– Luca MirtaniniNov 15, 2021 at 19:46
1 Answer
Let's use $$y=-(x-2)^2+C$$ Then at $x=1$ one gets $$\frac yx=C-1$$ and $$\frac{dy}{dx}=-2(1-2)=2$$ Notice that you can change the $y/x$ ratio to any number, while the derivative is always constant. So you can't prove that it is smaller or larger.
EDIT With the new addition to the question that the function passes through origin
The definition of concave function $f$ is that $\forall \lambda\in(0,1)$ $$f((1-\lambda)a+\lambda b)\ge(1-\lambda)f(a)+\lambda f(b)$$ So let's choose $a=0$ and $b=x$. We know that $f(0)=0$, so $$f(\lambda x)\ge\lambda f(x)$$ Then we want to calculate $$\frac{f(x)-f(x-h)}h$$ We can choose $x-h=\lambda x$ or $h=(1-\lambda)x$. $$\frac{f(x)-f(x-h)}h=\frac{f(x)-f(\lambda x)}{(1-\lambda)x}\le\frac{f(x)-\lambda f(x)}{(1-\lambda)x}=\frac{f(x)}{x}$$
-
$\begingroup$ I forgot to say that the function should pass from the origin. I would like to do a generic demonstration $\endgroup$ Nov 15, 2021 at 19:44
-
1$\begingroup$ That's a very important missing part. I will add an edit to my answer $\endgroup$– AndreiNov 15, 2021 at 19:46
-
$\begingroup$ Thank you for your demonstration. I just want to ask you two questions. 1)Where in this demonstration, is pointed out the fact that the fucntion passes from the origin? 2) Since the derivative is a limit, the fact that you assume (1−λ)x tending to 0, do not affect the right member of the disequation? Sorry again $\endgroup$ Nov 16, 2021 at 11:58
-
$\begingroup$ I used $f(0)=0$ to get from definition to the second equation in the edited part. If my final formula is true for every $\alpha$, it's true in the limit as well $\endgroup$– AndreiNov 16, 2021 at 13:11
-
$\begingroup$ which parameter is $\alpha$? This demonstration is true also if I choose different values of a and b? $\endgroup$ Nov 16, 2021 at 13:49