Nesterenko's linear independence criterion for vectors

Lattices and convex bodies (number-theoretic aspects) (11H06) Simultaneous homogeneous approximation, linear forms (11J13) Irrationality; linear independence over a field (11J72)

Abstract: In this paper we deduce a lower bound for the rank of a family of

p

vectors in

R^{k}

(considered as a vector space over the rationals) from the existence of a sequence of linear forms on

R^{p}

, with integer coefficients, which are small at

k

points. This is a generalization to vectors of Nesterenko's linear independence criterion (which corresponds to

k = 1

), used by Ball-Rivoal to prove that infinitely many values of Riemann zeta function at odd integers are irrational. The proof is based on geometry of numbers, namely Minkowski's theorem on convex bodies.

Review text

Let \(k\) and \(p\) be integers with \(1\leq k\leq p-1\), and \(e_1,\dots ,e_k\in\mathbb R^p\). Let \((v_1,\dots v_p)\) denote a basis of \(\mathbb R^p\). Let \(\tau_1,\dots ,\tau_k >0, \sigma_1\geq\dots \geq\sigma_p>0, \omega_1,\dots ,\omega_k, \phi_1,\dots ,\phi_k\) be real numbers with \(\tau_1,\dots ,\tau_k\) pairwise distinct. Assume that there exist infinitely many \(n\) such that for any \(j\in\{ 1,\dots ,k\}\) we have \(n\omega_j+\phi_j\not\equiv \frac {\pi}2 \mod \pi\). Let \((Q_n)_{n\geq 1}\) be an increasing sequence of positive integers such that \(Q_{n+1}=Q_n^{1+O(\frac 1n)}\); if \(\omega_1=\dots =\omega_k=0\) this assumption can be weakened to \(Q_{n+1}=Q_n^{1+o(1)}\). For any \(n\geq 1\), let \(L_n=\sum_{i=1}^p l_{i,n}X_j\) be a linear form on \(\mathbb R^p\) with integer coefficients \( l_{i,n}\) such that, as \(n\to\infty\) \[ \mid L_n(e_j)\mid =Q_n^{-\tau_j+o(1)}\mid \cos (n\omega_j+\phi_j)+o(1)\mid \] for any \(j\in\{ 1,\dots ,k\}\) and \[ \mid L_n(v_i)\mid \leq Q_n^{\sigma_i+o(1)} \] for any \(i\in\{ 1,\dots ,p\}\). For all \(n\geq 1\) and \(i\in\{ 1,\dots ,p\}\), let \(\delta_{i,n}\) be a positive divisor of \(l_{i,n}\) such that \(\delta_{i,n}\) divides \(\delta_{i+1,n}\) for any \(i\in\{ 1,\dots ,p-1\}\), \(\frac{\delta_{j,n}}{\delta_{i,n}}\) divides \(\frac{\delta_{j+1,n}}{\delta_{i+1,n}}\) for any \(0\leq i<j\leq p\}\) with \(\delta_{0,n}=1\) and \(\delta_{i,n}=Q_n^{d_i+o(1)}\) as \(n\to\infty\) with real numbers \(d_i\) such that \(0\leq d_1\leq\dots \leq d_p\leq\sigma_p\). Then the author proves the following results. {\parindent=6mm \begin{itemize} \item[1.] If \(\mathbb F\) is a subspace of \(\mathbb R^p\) defined over \(\mathbb Q\) which contains \(e_1,\dots ,e_k\) then \(s=\dim_{\mathbb F}\) satisfies \(s\geq k+1\) and \(\sum_{i=1}^{s-k}\sigma_i \geq \sum_{i=1}^k\tau_i+\sum_{i=1}^sd_i\). \item [2.] The vectors \(e_1,\dots ,e_k\) are \(\mathbb R\)-linearly independent in \(\mathbb R^p\) and the \(\mathbb R\)-subspace they span does not intersect \(\mathbb Q^p\setminus\{ (0,\dots ,0)\}\). \item [3.] Let \(\varepsilon >0\) and \(Q\) be sufficiently large (in terms of \(\varepsilon >0\)). Let \(C(\varepsilon ,Q)\) denote the set of all vectors that can be written as \(\sum_{i=1}^k\lambda_ie_i+u\) with \(\lambda_1,\dots ,\lambda_k\in\mathbb R\) such that \(\mid\lambda_j\mid \leq Q^{\tau_j-\varepsilon}\) for any \(j\in\{ 1,\dots ,k\}\) and \(u\in (\text{Span}_{\mathbb R}(e_1,\dots ,e_k))^{\perp}\) such that \(u=\sum_{i=1}^pu_iv_i\) with \(\mid u_i\mid \leq Q^{\sigma_i-\varepsilon}\) for any \(i\in\{ 1,\dots ,p\}\). Let \(\Lambda(Q)\) denote the set of all \((x_1,\dots ,x_p)\in\mathbb Q^p\) such that \(\delta_{i,\Psi(Q)}x_i\in\mathbb Z\) for any \(i\in\{ 1,\dots ,p\}\), where \(\Psi(Q)\) is the largest integer \(n\) such that \(Q_n\leq Q\). Then \(C(\varepsilon ,Q)\cap \Lambda(Q)=\{ (0,\dots ,0)\}\). \end{itemize}} The proof is based on the geometry of numbers.

Recommendations

Cites work

Cited in

(9)

This page was built for publication: Nesterenko's linear independence criterion for vectors

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2354037)