Some remarks on units in Grothendieck-Witt rings (Q1703093): Difference between revisions

Let \(k\) be a field of characteristic not 2, and let \(S\) be the space of isomorphism classes of quadratic forms equipped with the operation of direct sum and tensor product. By the usual Grothendieck completion procedure, we obtain the Grothendieck-Witt ring \(GW(k)\) of \(k\). In this ring, we have the group of units \(GW^\times(k)\). With a general commutatitive ring, there isn't an obvious way to make its group of units into a module over the ring itself. However, the author of the title under review has found a way to do that for the ring \(GW(k)\). That is, he gives \(GW^\times(k)\) a \(GW(k)\)-module structure, defining a scalar multiplication \(xy\) for \(x\in GW(k)\) and \(y\in GW^\times(k)\), which the author writes as exponentiation for good reason. Moreover, if \(A/k\) is a finite étale \(k\)-algebra, then this module structure is compatible with the projection formula, so that for \(x\in GW(A)\) and \(y\in GW^\times(k)\), the formula \(\mathrm{tr}_{A/k}(x)y = N_{A/k}(x(y|_A))\). Taking this idea further, the author sheafifies this idea and proves that the sheafified \(GW^\times\) is homotopy invariant. A short exact sequence involving \(GW(k)\) and of the corresponding sheaves is derived as well involving the fundamental ideal functor. This allows the author to give a new presentation of the group of units \(GW^\times(k)\), which is a rather interesting result.