Quantitative \(K\)-theory related to spin Chern numbers (Q401269)

Let \(U\) and \(V\) be two unitary matrices that almost commute in the sense that their commutator is small in operator norm. \(K\)-theory gives integers that are obstructions for \(U\) and \(V\) to be close in operator norm to unitary matrices that commute exactly. The simplest formula for this obstruction is as a winding number. Other formulas for this obstruction involve a concrete formula for the Bott generator of the \(K\)-theory of the 2-torus. A third way to express the obstruction involves logarithms of unitary matrices. In each case, the obstruction says that there are \(c>0\) and \(\delta>0\) so that two pairs of unitaries with commutator of norm at most \(\delta\) and with different numerical invariants have distance at least \(c\). This article finds reasonably large choices of \(\delta\) and \(c\) for each of the three methods, with \(\delta\) between \(2\) and \(0.1\). This is important to use the obstructions in practical computations. Some applications in quantum physics are mentioned. We may ask the same question for pairs of unitaries with some extra symmetries. For self-dual unitaries with respect to some involution, the numerical obstruction is given by the sign of the Pfaffian and so belongs to \(\{\pm1\}\). The winding number formula for the obstruction does not work in this case, but the other two methods still do. This is a reason to consider these methods, although they need smaller values of \(\delta\) and \(c\).