What is the estimation for the positive root of the following equation $$ ax^k = (x+1)^{k-1} $$ where $a > 0$ (specifically $0 < a \leq 1$).

Could you point out some reference related to the question?

  • $\begingroup$ Is $k$ a positive integer? What kind of estimates are you looking for? $\endgroup$ – Aryabhata Apr 7 '11 at 16:37
  • $\begingroup$ Yes, $k$ is positive integer. If $a = 1$ then an estimation on $x$ is $\Theta(k/\log k)$. I am looking for a similar kind of estimation where $a$ is fixed (an estimation as a function of $k$ (of course that depends also on $a$) where $k$ is large). $\endgroup$ – ogn Apr 7 '11 at 16:44
  • $\begingroup$ is it possible to verify this equation on a contraction property? $\endgroup$ – Ilya Apr 7 '11 at 16:53

Put $\displaystyle z = 1 + \frac{1}{x}$ and we get the equation

$$z^{n-1}(z-1) = a$$

(I prefer to use $\displaystyle n$ instead of $\displaystyle k$)

We can easily see that $\displaystyle z \in (1,2)$

Assume that $\displaystyle z = 1 + \frac{g(n)}{n-1}$

Thus we have that

$$\left(1 + \frac{g(n)}{n-1}\right)^{n-1} g(n) = (n-1)a$$

Now we have that $\displaystyle e^{x/2} \lt 1 + x \lt e^x$ for $\displaystyle x \in (0,1)$

And so we get $$e^{g(n)} g(n) \gt (n-1)a \gt e^{g(n)/2} g(n)$$

Now since $\displaystyle xe^x$ is increasing, and the root of $\displaystyle xe^x = y$ is given by the LambertW function: $\displaystyle W(y)$.

Thus we get that

$\displaystyle g(n) \gt W((n-1)a)$


$\displaystyle g(n) \lt 2 W\left(\frac{(n-1)a}{2}\right)$

It is well known that $\displaystyle W(x) = \theta(\log x)$ (as $\displaystyle x \to \infty$) and so we get that

$\displaystyle g(n) = \theta (\log (n-1)a) = \theta(\log na)$

Thus the root $\displaystyle r(n)$ is $$\theta\left(\frac{n}{\log na}\right)$$

In fact, I would go so far as to guess that

$$ \lim_{n \to \infty} \dfrac{r(n)W((n-1)a)}{n-1} = 1$$

  • $\begingroup$ Nice! But I am wondering what happens if $a$ is a function of $n$, say $a = 2^{-n}$. Now $\log(na)$ is negative, but the equation has positive root. $\endgroup$ – ogn Apr 8 '11 at 8:25
  • $\begingroup$ Put $y = ax$, we have $(ax)^n = (ax + a)^{n-1} \leq (ax + 1)^{n-1}$. Thus, $x \leq c \cdot \frac{n}{a \log n}$ for some constant $c$. % However, I think the bound could be strengthened. $\endgroup$ – ogn Apr 8 '11 at 11:59
  • $\begingroup$ @ogn: Even if $a$ is $2^{-n}$, $W((n-1)a)$ will be positive, you will just have to pick the asymptotics for $W(x)$ as $x \to 0$ (here we picked the one where $x \to \infty$). So I don't see the problem. For constant $a$, the bound I gave is tight (i.e. $\theta$). $\endgroup$ – Aryabhata Apr 8 '11 at 14:15
  • 1
    $\begingroup$ @CogitoErgoCogitoSum. It is an asymptotic estimate. That itself should tell you that plugging in values is just wasting your time (at least for small n). $\endgroup$ – Aryabhata Mar 6 '17 at 19:29
  • 1
    $\begingroup$ @CogitoErgoCogitoSum: You haven't shown any work at all and don't seem to have understood what I was saying (in the answer and comment). I suspect you are doing this deliberately and it is going to be hard for me to be polite if I continue this discussion. So bye. $\endgroup$ – Aryabhata Mar 16 '17 at 22:57

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.