SearcharxivSearch

arXiv subjects

Cheuk Fung Lau

Publications and source records attributed to Cheuk Fung Lau.

4 recordsLinked to original sources

Residue Class Patterns of Consecutive Primes

Dickson's conjecture and the Hardy--Littlewood prime tuple conjecture predict that every pattern of reduced residue classes modulo $q$ is attained by infinitely many strings of $m$ consecutive primes. At present, however, even proving that a single non-constant residue class pattern of length $m$ occurs infinitely often is beyond the reach of existing methods. Combining Dirichlet's theorem on primes in arithmetic progressions with a theorem of Shiu (2000) shows that, for any $m,q\in\mathbb N$ with $q \ge 3$, at least $mφ(q)$ residue class patterns of length $m$ are attained by infinitely many consecutive primes. In this paper, we prove that if $q$ is squarefree, every prescribed sequence of at least $60m\log m$ reduced residue classes mod $q$ contains, in order, an $m$-term block pattern that occurs infinitely often among consecutive primes, with each constant block of length at most $\lceil\log m\rceil$. A recursive combinatorial argument then shows that if $q$ is squarefree and $q \gg (\log m)^2$, then at least \[ \gg \frac{m}{(\log m)^{10}} φ(q)^2 \] residue class patterns of length $m$ occur infinitely often among consecutive primes. Moreover, we also show that if $q$ is squarefree and $q \gg (\log m)^2$, then at least \[ \gg e^{-O(m \log_2 m/\log m)} φ(q)^{m/\lceil \log m \rceil} \] residue class patterns of length $m$ occur infinitely often among consecutive primes. The proof consists of a modification of the Maynard--Tao sieve found in Banks, Freiberg, and Maynard (2016), by considering the $r$-th moment instead of the 2nd moment for an integer $r$ depending on $m$, which is then combined with an Erdős--Rankin type construction.

math.NT

Smoothed Shifted Convolutions of Generalised Divisor Functions

We prove an asymptotic formula for the smoothed shifted convolution of the generalised divisor function $d_k(n)$ and the divisor function $d(n)$ for $k \ge 4$, with a power-saving error term whose exponent is independent of $k$. In particular, for sufficiently large $k$, this improves on the result of Topacogullari (2018).

math.NT

On the Number of Prime Factors of Consecutive Integers

We prove that there are infinitely many $n$ such that $ω(n+k) \ll \log k$ for all integers $k \ge 2$. This improves on a result of Tao-Teräväinen (2025), who has $O(k)$ in place of $O(\log k)$. As corollaries, we make progress on a number of questions posed by Erdős. The proof is based on a quantitative refinement of the Tao-Teräväinen probabilistic argument, combining a more efficient sieve procedure with stronger exponential concentration-of-measure estimates. Moreover, we formulate a conjecture on integers with many prime factors based on Cramér-type random models. Assuming this conjecture, the main bound is essentially sharp.

math.NT

Simultaneously Small Fractional Parts of Polynomials

Let $f_1,\dots,f_k \in \mathbb{R}[X]$ be polynomials of degree at most $d$ with $f_1(0)=\dots=f_k(0)=0$. We show that there is an $n<x$ such that $\|f_i(n)\|\ll x^{-1/10.5kd(d-1)+o(1)}$ for all $1\le i\le k$. This improves on an earlier result of Maynard, who obtained the same exponent dependency on $k$ but not on $d$.

math.NT