On the \(u\)-invariant of Hermitian forms (Q369288)

Let \(A=M_n(D)\) be a central simple algebra over a field \(K\) of characteristic not \(2\) where \(D\) is a central division algebra over \(K\), let \(\sigma\) be an involution on \(A\) and let \((V,h)\) be an \(\varepsilon\)-Hermitian space over \((A,\sigma)\) of finite \(K\)-dimension. Then \(n\dim_K(D)\) divides \(\dim_K V\) and one defines the dimension of the form \(h\) to be \(\dim h=\dim_KV/(n\dim_K(D))\). The \(u\)-invariant \(u(A,\sigma,\varepsilon)\) is defined to be the supremum of the dimensions of those forms \(h\) as above that are anisotropic. In the case \(A=K\), \(\sigma =id\) and \(\varepsilon=1\) one recovers \(u(K)\), the usual \(u\)-invariant of a field \(K\). In [Manuscr.\ Math.\ 116, No. 4, 493--516 (2005; Zbl 1155.11321)], \textit{M. G. Mahmoudi} proved that \(u(A,\sigma,\varepsilon)\leq r(r+1)u(k)/(2m^2[K:k])\) where \(k\) is the fixed field of \(K\) under \(\sigma\), \(m=\deg_KA\) and \(r\) is the \(k\)-dimension of the space of \(\varepsilon\)-symmetric elements in \(A\). The main purpose of the present paper is to improve these bounds in the situation where \((A,\sigma)\) is given by the tensor product of quaternion algebras with their canonical involution. To this effect, the authors first give new upper bounds for \(u(A,\sigma,\varepsilon)\) in terms of \(u\)-invariants of certain subalgebras with involution of \(A\) where they employ a new exact sequence of Witt groups of Hermitian forms over certain central simple algebras with involution of the second kind. For tensor products \((H,\tau):=(H_1,\tau_1)\otimes\cdots\otimes (H_m,\tau_m)\) where the \((H_i,\tau_i)\) are quaternion algebras with canonical involution, it is shown that \(u(H,\tau,\varepsilon)\leq c_mu(k)\) with \(c_1=7/8\), \(c_2=29/16\) for \(\varepsilon =1\) and \(c_2=17/16\) for \(\varepsilon=-1\), and \(c_m=213\cdot 3^{2m-6}4^{-m}\) for \(m\geq 3\). There are no examples given that would indicate how good these bounds really are, so this leaves room for further exploration of this nice subject.