The question is as follows. Let us consider a positive integer number $x \in \{1,2,3,...\}$ and a positive real number $q \in [1,x]$. Show that \begin{equation} \sum_{m=0}^{x-1} \left( \frac{x+2-q}{m+1} -\frac{x+q}{\min(m+q,x)} \right) \binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m \geq 0. \end{equation}

I've tried to solve the inequality as follows. First, let us denote the LHS as $f(x,q)$
\begin{equation} f(x,q)=\sum_{m=0}^{x-1} \left( \frac{x+2-q}{m+1} -\frac{x+q}{\min(m+q,x)} \right) \binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m. \end{equation} I want to show that $f(x,q) \geq 0$ for $x \in \{1,2,3,...\}$ and any positive real number $q \in [1,x]$. It is easy to verify that $f(x,1)=0$ and $f(x,x)=0$.

The first idea I pursued was to try by induction over $x$, while considering $f(x,q)$ with a fixed $q$. In fact, numerical results seem to verify that $f(x+1,q) \geq f(x,q)$ (the inequality is strict for $q > 1$). I tried to prove it by induction but I could not succed as I could not manage the terms $\binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m$, which are very difficult to manipolate.

The second way I tried was to consider $f(x,q)$ by fixing $x$ and letting $q$ to vary in $[1,x]$. In this case $f(x,q)=0$ at the two extremes of the interval ($q=1$ and $q=x$), while numerical results show that $f(x,q)$ is greater than zero inside the interval, suggesting that $f(x,q)$ is quasiconvex in $q \in [1,x]$. My idea was to consider the different intervals of $q$ given by $q \in [1,2)$, $[2,3)$ (where $f(x,q)$ is differentiable), and show that in each of this interval the function is either convex or concave, while being quasiconcave all over the interval. In fact, by considering $q \in [a,a+1)$, for any integer $a$ in $[1,x]$, we have that
\begin{equation} f(x,q)=\sum_{m=0}^{x-a-1} \left( \frac{x+2-q}{m+1} -\frac{x+q}{m+q} \right) \binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m + \sum_{m=x-a}^{x-1} \left( \frac{x+2-q}{m+1} -\frac{x+q}{x} \right) \binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m . \end{equation} However, taking the derivate with respect to $q$ is very difficult and I wasn't able to conclude anything after this.

To sum up, it looks like that the issues are given by the terms $\binom{x-1}{m} \left ( \frac{x-q}{x+q}\right)^m$, which are very difficult to manage. If someone can help me by providing new directions to look at, pointing me out similar inequalities in the literature which I don't know, or give me new ideas, I would be grateful.


After many attemps, I have simplified the inequality as follows: Show that, for any positive integer $x \in \{1,2,3,... \}$ and $q \in [1,x]$, the following inequality holds: \begin{equation} \sum_{m=1}^x \frac{m}{\min(m+q-1,x)} \binom{x}{m} \left( \frac{x-q}{x+q}\right)^m \leq \frac{x-q+2}{x+q} \left( \left(\frac{2x}{x+q} \right)^x -1\right). \end{equation} Note that the LHS and the RHS are equal for $q=1$ and $q=x$. Moreover, note that \begin{equation} \sum_{m=1}^x \binom{x}{m} \left( \frac{x-q}{x+q}\right)^m = \left(\frac{2x}{x+q} \right)^x -1 . \end{equation} This inequality looks more tractable. However, even here, all the classical techniques as CS, AM-GM, concavity/convexity seem to fail.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.