Searcharxiv⌕ Search

arXiv · 0706.2219

On the derivative of the Minkowski question mark function $?(x)$

Abstract

Let $ x = [0;a_1,a_2,...]$ be the decomposition of the irrational number $x \in [0,1]$ into regular continued fraction. Then for the derivative of the Minkowski function $?(x)$ we prove that $?'(x) = +\infty$ provided $ \limsup_{t\to \infty}\frac{a_1+...+a_t}{t} <κ_1 =\frac{2\log λ_1}{\log 2} = 1.388^+$, and $?'(x) = 0$ provided $ \liminf_{t\to \infty}\frac{a_1+...+a_t}{t} >κ_2 = \frac{4L_5-5L_4}{L_5-L_4}= 4.401^+$ (here $ L_j = \log (\frac{j+\sqrt{j^2+4}}{2}) - j\cdot\frac{\log 2}{2}$). Constants $κ_1,κ_2$ are the best possible. Also we prove that $?'(x) = +\infty$ holds for all $x$ with partial quotients bounded by 4.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Anna A. Dushistova, Nikolai G. Moshchevitin. 2007-12-17. On the derivative of the Minkowski question mark function $?(x)$. https://arxiv.org/abs/0706.2219

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Root numbers for twisted Fermat quotient curves

Let $\ell$ be an odd prime, $N \geq 1$ be an integer, and $δ\geq 1$ be a $\ell^N$-th power free integer such that ${\rm ord}_{\ell}(δ) = 0$ or $\ell \nmid {\rm ord}_{\ell}(δ)$. In this paper, we give an explicit formula for the root number of the Hecke character associated with a certain quotient curve of the twisted Fermat curve $X^{\ell^N} + Y^{\ell^N} = δ$. This result gives a generalization of Stoll (2002) and Shu (2021).

math.NT↗

Dirichlet $L$-functions on the critical line and multiplicative chaos

In this article, we prove that the Dirichlet $L$-functions $L(1/2+ix,χ_q)$, where $χ_q$ is a uniformly random Dirichlet character modulo $q$ and $x\in \mathbb{R}$, converge as $q\to \infty$ to a random Schwartz distribution $ζ_{\mathrm{rand}}$, which is related to complex Gaussian multiplicative chaos (GMC). This is the same limiting object that appeared in \cite{SaWe20a}, where the authors proved that the random shifts of the Riemann zeta function on the critical line $ζ(1/2+ix+iωT)$, where $ω\sim \mathrm{Unif} ([0,1])$, converge as $T\to \infty$. As an application of our method, we also prove that to the right of the critical line, that is, for $\Re(s)>1/2$, the convergence of the random $L$-functions $L(s,χ_q)$ occurs in the space of analytic functions, if we exclude the zero probability (in the limit $q\to \infty$) event that $χ_q$ is the principal character. This was already proven in \cite{Ba81a}, but our method provides a novel proof of this fact. Our results hold for moduli $q=q_n$ such that the smallest prime divisor of $q_n$ tends to $\infty$ as $n\to\infty$.

math.NT↗

Finite fields whose members are the sum of a potent and a 5-potent

We show that there are only finitely many finite fields whose members are the sum of an $n$-potent element and a $5$-potent element. Combining this with the algorithmic results provided by S.D. Cohen {\it et al.}, we confirm in the affirmative the conjecture posed in \cite{Cohen} concerning all finite fields satisfying this condition. Furthermore, we obtain several elementary results related to the more general problem of decomposing every element of a finite field into the sum of an $m$-potent and an $n$-potent (as formulated in \cite{Cohen}), proving that the number of finite fields satisfying this condition is finite.

math.NT↗