Can we recover an integral quadratic form by representing all its subforms? (Q6058891)

Let \(\mathfrak o\) denote the ring of integers of an algebraic number field \(F\). Let \(L\) be an \(\mathfrak o\)-lattice; i.e., a finitely generated \(\mathfrak o\)-module in a quadratic space over \(F\). In this paper, it is first shown that if \(M\) is an indefinite \(\mathfrak o\)-lattice that represents all proper sublattices of \(L\), then \(M\) represents \(L\). By contrast, it is proven that if \(F\) is totally real and \(L\) is a positive definite indecomposable \(\mathfrak o\)-lattice, then there exists a positive definite \(\mathfrak o\)-lattice that represents all proper sublattices of \(L\), but not \(L\) itself. For \(\mathbb Z\)-lattices of rank less than 4, this result was proven in a recent paper of \textit{K. Kim} et al. [J. Number Theory 238, 37--59 (2022; Zbl 1506.11051)]. The general result is obtained here as a corollary from the following characterization of decomposable lattices: a positive definite \(\mathfrak o\)-lattice is decomposable if and only if there exist proper sublattices \(L_1,\ldots,L_t\) of \(L\) such that \(L\) is represented by \(L_1\perp\cdots\perp L_t\). The authors then prove that, for a totally real number field \(F\), every infinite set \(\mathcal S\) of isometry classes of positive definite \(\mathfrak o\)-lattices of a fixed rank contains a finite subset \(\mathcal S_0\) such that if a positive definite \(\mathfrak o\)-lattice represents all classes of lattices in \(\mathcal S_0\), then it represents all classes of lattices in \(\mathcal S\). \textit{B. M. Kim} et al. [J. Reine Angew. Math. 581, 23--30 (2005; Zbl 1143.11011)] previously proved this result for the special case \(F=\mathbb Z\). The strategy employed in the proof here follows the one used in that previous paper, but additional effort is required to handle the Humbert reduction needed over more general totally real fields. As an interesting special case, it follows that for a totally real number field \(F\), there exists a finite universality criterion set. That is, there is a finite set \(\mathcal U(F)\) of totally positive integers of \(F\) such that any positive definite integer-valued quadratic form over \(F\) that represents all elements of \(\mathcal U(F)\) is universal.