0
$\begingroup$

I have asked about a limit before, which is still unanswered. With few modifications we can arrive to the following limit, which experimentally can be be shown to converge to $0$. However, I can't prove it.

$$\lim_\limits{x\to\infty}{f(x)}=x^a{\left(1-\frac {0.1\prod_\limits{t=1}^x{0.5}^{1/t}}{10}\right)}^{x-a},$$

where $a\in\mathbb Z^+$ is constant and $x\in\mathbb Z^+$.

I think we can solve this limit by solving the following one:

$$\lim_\limits{x\to\infty}{f(x)}=x^a{\left(1-\frac {0.1\cdot {0.5}^{\ln k}}{10}\right)}^{x-a}.$$

$\endgroup$
1
$\begingroup$

For large $x$ we know that $H(x)=\sum_{t=1}^{x}\frac{1}{t}\approx \ln(x)$. Your limit will be equal to $$x^a\left(1-\frac{1}{100\cdot x^{\ln(2)}}\right)^{x-a}=x^a e^{(x-a)\ln\left(1-\frac{1}{100\cdot x^{\ln(2)}}\right)}$$ Since $\ln(1+y)=y+O(y^2)$ when $y\to 0$, we have:

$$(x-a)\ln\left(1-\frac{1}{100\cdot x^{\ln(2)}}\right)=-\frac{(x-a)}{100\cdot x^{\ln(2)}}+O\left(x^{1-2\ln(2)}\right)$$ when $x\to \infty$, so we have $$\lim_{x\to\infty} f(x)=\lim_{x\to\infty}e^{-\frac{a(x-a)\ln(x)}{100\cdot x^{\ln(2)}}+O\left(x^{1-2\ln(2)}\ln(x)\right)}$$ Obviously that when $x\to\infty$ exponent's power $-\frac{a(x-a)\ln(x)}{100\cdot x^{\ln(2)}}+O\left(x^{1-2\ln(2)}\ln(x)\right)\to -\infty$ and our limit will be zero

$\endgroup$
1
$\begingroup$

Write $f(x)=x^a{\left(1-\frac {.1\prod_\limits{t=1}^x{0.5}^{1/t}}{10}\right)}^{x-a} $ as $f(x)=x^a{\left(1-c\prod_\limits{t=1}^xd^{1/t}\right)}^{x-a} $ where $c = 1/100$ and $d = 0.5$.

Since $\prod_\limits{t=1}^xd^{1/t} =\exp(\sum_\limits{t=1}^x\frac{\ln(d)}{t}) =\exp(\ln(d)\sum_\limits{t=1}^x\frac{1}{t}) =\exp(\ln(d)H_x) =d^{H_x} $ where, for large $x$, $H_x = \sum_\limits{t=1}^x\frac{1}{t} =\ln(x)+\gamma+O(1/x) $ (the well-known harmonic sum),

$\begin{array}\\ f(x) &=x^a{\left(1-cd^{H_x}\right)}^{x-a}\\ &=x^a{\left(1-cd^{\ln(x)+\gamma+O(1/x)}\right)}^{x-a}\\ &=x^a{\left(1-cd^{\ln(x)}d^{\gamma}d^{O(1/x)}\right)}^{x-a}\\ &=x^a{\left(1-cx^{\ln(d)}d^{\gamma}d^{O(1/x)}\right)}^{x-a}\\ \end{array} $

so, for large $x$,

$\begin{array}\\ \ln(f(x)) &=a\ln(x)+(x-a)\ln(1-cx^{\ln(d)}d^{\gamma}d^{O(1/x)})\\ &=a\ln(x)+(x-a)\ln(1-cx^{\ln(d)}d^{\gamma}(1+O(1/x)))\\ &=a\ln(x)+(x-a)\ln(1-cx^{\ln(d)}d^{\gamma}+O(x^{\ln(d)-1}))\\ &=a\ln(x)-(x-a)(cx^{\ln(d)}d^{\gamma}+O(x^{\ln(d)-1})+O(x^{2\ln(d)}))\\ &=a\ln(x) -x(cx^{\ln(d)}d^{\gamma}+O(x^{\ln(d)-1})+O(x^{2\ln(d)})) +a(cx^{\ln(d)}d^{\gamma}+O(x^{\ln(d)-1})+O(x^{2\ln(d)}))\\ &=a\ln(x) -cx^{1+\ln(d)}d^{\gamma}+O(x^{\ln(d)})+O(x^{1+2\ln(d)})) +a(cx^{\ln(d)}d^{\gamma}+O(x^{\ln(d)-1})+O(x^{2\ln(d)}))\\ &=a\ln(x) -cx^{1+\ln(d)}d^{\gamma}+o(1)\\ &\to -\infty\\ \end{array} $

since $1+\ln(d) \approx 0.3 \gt 0$ and $\dfrac{\ln(x)}{x^{1+\ln(d)}} \to 0$.

More precisely, $f(x) \approx \dfrac{x^a}{\exp\left(cx^{1+\ln(d)}d^{\gamma}\right)}(1+o(1)) $.

By keeping more terms in the various expansions, we could get a more accurate result.

$\endgroup$

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.