The least common multiple of a bivariate quadratic sequence. (English) Zbl 07771583
Writing the Chebyshev function \(\psi(N)\) in the following form,
\[
\psi(N) =\log\left(\text{LCM}\left\lbrace1,2,3,4, \dots ,N\right\rbrace\right) =\log\left(\mathop{\text{LCM}}_{n\leq N}\left\lbrace n\right\rbrace\right),
\]
in the paper under review, the author considers
\[
\psi_F(N) = \log\left( \mathop{\text{LCM}}_{0<F(x,y)\leq N} \left\lbrace F(x,y)\right\rbrace \right),
\]
where \(F(x,y)\in\mathbb{Z}[x,y]\) is a polynomial of degree 2 of the general form
\[
F(x,y) = ax^2+bxy+cy^2+ex+fy+g
\]
with discriminant \(\Delta = b^2-4ac\), and assuming that \(F\) represents arbitrarily large integers. The author shows that if \(\frac{\partial F}{\partial x},\frac{\partial F}{\partial y}\) are linearly dependent, then
\[
\delta_F(N) := \frac{1}{N}\,\#\left\lbrace 0<n\leq N \;\middle|\; n=F(x,y) \right\rbrace \sim \frac{\eta}{N},
\]
where \(\eta=1/\sqrt{\text{GCD}(a,c)}\). In this case, if \(F\) is irreducible then \(\psi_F(N) \sim (\eta/2)\sqrt{N} \log N\), and if \(F\) is reducible then \(\psi_F(N) \sim c_F \sqrt{N}\), where the constant \(c_F\) is explicitly computable. The author also considers the case when \(\frac{\partial F}{\partial x},\frac{\partial F}{\partial y}\) are linearly independent and shows that in this case, if \(\Delta\) is a perfect square, then \(\delta_F(N)\asymp 1\) and \(\psi_F(N)\asymp N\), and if \(\Delta\) is not a perfect square and \(D:=af^2 + ce^2 -bfe + \Delta g = 0\) then \(\delta_F(N)\asymp 1/\sqrt{\log N}\) and \(\psi_F(N)\asymp N\). Moreover, if \(\Delta\) is not a perfect square and \(D\neq 0\), then \(\delta_F(N)\asymp 1/\sqrt{\log N}\), and \(\psi_F(N)\asymp (N \log \log N)/\sqrt{\log N}\).
Reviewer: Mehdi Hassani (Zanjan)
MSC:
11N32 | Primes represented by polynomials; other multiplicative structures of polynomial values |
11N36 | Applications of sieve methods |