Maximal Unobservable Subspace
Abstract
The all-future readout kernel is the maximal invariant hidden subspace.
Theorem 1.1 (The future kernel is maximal among invariant hidden subspaces).
Proof. Machine-checked in Lean as D5/S3/ObserverMemory/Dynamics/MaximalUnobservableSubspace.future_kernel_is_maximal_invariant (✓ std3). ∎
Source. Repository-derived.
Commentary.
Let V and Y be finite-dimensional inner-product spaces over a real or complex scalar field. Let T evolve V linearly and let C read V linearly into Y.
The hidden subspace is constructed canonically as the intersection of the kernels of C composed with every power of T. This is the source all-future readout test, not a definition by maximality.
The public theorem states all maximality clauses: the future kernel lies inside ker(C), T maps it into itself, and every T-invariant subspace inside ker(C) is contained in it.
The zero iterate proves current invisibility, shifting an iterate proves invariance, and induction keeps every iterate of a point in any competing invariant subspace.
References
- Truth anchor:
D5/S3/ObserverMemory/Dynamics/MaximalUnobservableSubspace.future_kernel_is_maximal_invariant - Dependency: D5/S3/ObserverMemory/Dynamics/ObservableKrylovGrowthBound