1
$\begingroup$

Let $h : [0,1] \to [0,1]$ be a $\mathcal{C}^1$ function such that $h'(x)<0$ for all $x \in (0,1)$. Consider the function $$ \psi(x) = \frac{\int_0^1yh(|x-y|) dy}{\int_0^1 h(|x-y|)dy} $$ I am trying to show that $\psi$ is monotonically increasing (i.e. $\psi'(x)>0$ for all $x \in (0,1)$). This property seems to be working for many examples for $h$ but I don't know how to generalize it.

Note that we can easily show that $\psi(x)+\psi(1-x)=1$ for all $x\in [0,1]$. This might help?

$\endgroup$

1 Answer 1

3
$\begingroup$

Unfortunately, one can construct an $h(x)$ such that it is untrue that $\psi^{\prime } (x) >0$ for sufficiently small $x$. To begin with, in order to get a better sense of what the moving parts are here, separate the integration intervals into $y<x$ and $y>x$, and substitute $t=x-y$ or $t=y-x$ such that $h$ appears with argument $t$, $h(t)$ everywhere. Then, $\psi $ takes the form $$ \psi (x) = x + \frac{\int_{x}^{1-x} dt\, t\, h(t)}{2\int_{0}^{x} dt\, h(t) + \int_{x}^{1-x} dt\, h(t) } $$ and therefore, $$ \psi^{\prime } (x) = 1 - \left. \frac{1}{\left[ 2\int_{0}^{x} dt\, h(t) + \int_{x}^{1-x} dt\, h(t) \right]^{2} } \right[ \\ ( (1-x) h(1-x)+xh(x)) \left( 2\int_{0}^{x} dt\, h(t) + \int_{x}^{1-x} dt\, h(t) \right) + \left. (h(x)-h(1-x)) \int_{x}^{1-x} dt\, t\, h(t) \right] $$ So, for $\psi^{\prime } (x)$ to remain positive, the denominator must exceed the numerator. Consider $h(t)=(1/2)(1+\theta (x_0+\epsilon -t))-\kappa t$, with, in order to formally satisfy the conditions of the OP, an arbitrarily small downward slope $\kappa $, and with the step function smeared out to continuously drop from $1$ to $0$ on the arbitrarily narrow interval $]x_0 , x_0 +2\epsilon [$. Evaluating at $x=x_0 $, we have, up to arbitrarily small corrections, $$ \int_{0}^{x_0 } dt\, h(t) = x_0 $$ $$ \int_{x_0 }^{1-x_0 } dt\, h(t) = \frac{1}{2} -x_0 $$ $$ \int_{x_0 }^{1-x_0 } dt\, t\,h(t) = \frac{1}{4} (1-2 x_0 ) $$ and so, $$ \psi^{\prime } (x=x_0 ) = 1-\frac{3/8 + x_0/2 + x_0^2 /2}{1/4 +x_0 +x_0^2 } $$ which is negative for sufficiently small $x_0 $.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .