Effective estimates on integral quadratic forms: Masser's conjecture, generators of orthogonal groups, and bounds in reduction theory (Q311145)

Consider two integral quadratic forms \(Q_A\), \(Q_B\) over \(\mathbb{Z}\) of rank \(n\) given by symmetric matrices \(A\), \(B\in\mathbb{Z}^{n\times n}\). They are called equivalent if there exists \(\gamma\in\mathrm{GL}_n(\mathbb{Z})\) such that \(A=\gamma^tB\gamma\). If the forms are equivalent, it is an important algorithmic question whether one can find such \(\gamma\) such that \(\|\gamma\|\), the maximum of the absolute values of the coefficients of \(\gamma\), can be kept below an effective bound. In [Nachr. Akad. Wiss. Gött., II. Math.-Phys. Kl. 1972, 21--46 (1972; Zbl 0252.10019)], \textit{C. L. Siegel} showed that for nonsingular equivalent forms as above, there exists a nonnegative function \(F_n=F_n(A,B)\) such that there always exists such \(\gamma\) with \(\|\gamma\|<F_n(A,B)\). Since then, efforts have been undertaken to make such an \(F_n\) explicit. Siegel's arguments can be used to find such \(F_n\) that is essentially exponential in \(\|A\|\), \(\|B\|\), and it can be shown that in the indefinite rank \(n=2\) case, one cannot do better than exponential bounds. In the definite case, the authors point out that it is not too hard to show that one can achieve \[ \|\gamma\|<2(4n)^{(n-1)/2}\|A\|^{1/2}\|B\|^{(n-1)/2}. \] So the hard case is rank \(\geq 3\) and indefinite. The present paper constitutes an important contribution to this and related problems. The first main theorem states that for \(n\geq 3\) and nonsingular forms as above, one can find a bound \[ \|\gamma\|<C_n|\det B|^{-2f_1(n)/n}\|A\|^{f_2(n)}\|B\|^{f_3(n)}, \] where \(C_n>0\) is a constant and the \(f_i(n)\) are explicitly given polynomials of degree \(3\) of the shape \(\frac{13}{40}n^3+\ldots\) while still keeping the determinant of a given transformation. If the forms are indefinite and anisotropic, the authors show that one can obtain \[ \|\gamma\|<C_n|\det B|^{f(n)/n}\|A\|^{1/2}\|B\|^{(n-1)/2} \] with \(f(n)\) quadratic of shape \(\frac{13}{10}n^2+\ldots\). This settles a conjecture by \textit{D. W. Masser} posed in [in: A panorama in number theory or The view from Baker's garden. Based on a conference in honor of Alan Baker's 60th birthday, Zürich, Switzerland, 1999. Cambridge: Cambridge University Press, 247--259 (2002; Zbl 1066.11014)]. Previously, this conjecture was known to hold only in the case \(n=3\) and some special higher rank cases due to \textit{R. Dietmann} [Proc. Lond. Math. Soc. (3) 86, No. 3, 545--582 (2003; Zbl 1036.11009)]. The proof of Masser's conjecture in the present paper is reduced to a study of the quantitative behavior of the translates in \(\mathrm{SL}_n(\mathbb{R})\backslash\mathrm{SL}_n(\mathbb{Z})\) and uses methods from homogeneous dynamics, arithmetic groups and the spectral theory of automorphic representations. Several related theorems are also proved. For example, it is shown that the orthogonal group \(\mathbf O_{Q_B}(\mathbb{Z})\) for \(Q_B\) nonsingular and of rank \(\geq 3\) is generated by those \(\gamma\in\mathbf O_{Q_B}(\mathbb{Z})\) with \(\gamma<C_n|\det B|^{n^6}\|B\|^{3n^4}\) for some constant \(C_n>0\) (Theorem 2), thus also giving a new proof that \(\mathbf O_{Q_B}(\mathbb{Z})\) is finitely generated. Theorem 3 states that any nonsingular indefinite integral form \(Q_A\) of rank \(\geq 3\) is equivalent to a form \(Q_B\) with \(\|B\|<C_n\|A\|^{1/n}\). The final main theorem (Theorem 4) concerns representations of one nonsingular integral quadratic form \(Q_A\) of rank \(m\) by another such form \(Q_B\) of rank \(n\geq m\) with \(n\geq 3\), i.e., the case when there exists \(\tau_0\in\mathbb{Z}^{n\times m}\) with \(\tau_0^tB\tau_0=A\). If this holds, then it is shown that one can find such \(\tau\) with \(\tau^tB\tau=A\) and \(\|\tau\|<C_n|\det A|^{n^4}|\det B|^{n^3}\|A\|^{n^3}\|B\|^{n^3}\) for some constant \(C_n>0\). If \(\tau_0\) is primitive, i.e., if the greatest common divisor of its maximal minors is \(1\), the \(\tau\) can also be chosen primitive.