The authors propose a new, state-based way to classify topological phases of quantum matter. Instead of classifying operators such as spectral projectors of Hamiltonians, they study the topology of the space of pure quantum states of a chosen algebra of observables. Two configurations of states are considered the same phase when one can continuously deform one into the other through allowed states. This viewpoint aims to extend classification ideas to settings where the usual operator-based tools may be hard to use.

Their classification is built from a simple recipe. Fix a C*-algebra of observables and a symmetry group. Pick a “sample space” of pure states that respect those symmetries or other structural conditions. Often this sample space forms a fiber bundle over a compact parameter space X (the quantum parameter space). A continuous choice of a state over X is a section of that bundle. Topological phases are then the homotopy classes of those sections — that is, sections that can be deformed into each other without leaving the sample space.

They apply this scheme to the Weyl C*-algebra, a standard model for extended quantum systems, and to three physically motivated types of invariant pure states. One family is states invariant under all continuous translations. A second family is Bloch-wave states: pure, semi-regular states invariant under a lattice of translations; these live over the Brillouin zone, a compact torus B_Γ (the usual momentum space for a lattice). For the Bloch-wave case the bundle has a typical fiber given by a Hilbert Grassmannian of rank-1 projections equipped with a weak operator topology. A third family, called Zak states, is invariant under both spatial translations and momentum translations by the dual lattice.

Using these concrete bundles the authors recover known K-theory results for topological insulators in this state-based language. In particular, the classification scheme reproduces the K-theoretic classification of gapped spectral projectors for topological insulators of type A (see Theorem 4.11 and related results) and, when time-reversal symmetry is imposed, for type AI (see Theorem 6.6). They also find that in a spinless model the fully translation-invariant states lead to trivial topological phases, while including spin (by enlarging the algebra of observables) produces the expected nontrivial phases for homogeneous systems. They further report that Zak states do not naturally show nontrivial topology, though restricting the parameter space can produce “spurious” phases.