Abstract Wolfram's Principle of Computational Irreducibility asserts that for most non-trivial discrete systems there exists no analytical shortcut: to determine the state at time T, one must simulate all T steps. For finite-dimensional unitary graph systems, the evolution admits an exact spectral representation for a specific class of systems: finite quantum graphs with unitary evolution. Knowledge of the spectral slice - the quantum state |Ψ(t₀)⟩ and the spectrum of the evolution operator Û - fully and exactly determines the state at any moment T without stepping through intermediate times. The theorem is a principled, not a practical result: algorithmic complexity O( ) remains exponential. Practical significance is achieved in the coarse-graining limit at scale ℓ >> a, where the problem reduces to polynomial complexity. Three conditions - finiteness, unitarity, linearity - constitute the exact domain of applicability.
We study the requirements for and implications of the existence of finite-dimensional approximations of quantum systems with an infinite number of degrees of freedom. We find a close relationship between these physical questions and the embedding problem of Alain Connes, appearing in the theory of von Neumann algebras. We are mainly interested in the bipartite scenario known from quantum information theory, where two independent parties act on a joint physical system, i.e. by performing measurements. This latter example leads to Tsirelson's problem, which asks whether the modeling of bipartite situations using the usual approach of tensor product of Hilbert spaces compared to the situation of only commuting observables leads to the same correlation tables. We model this situation using the theory of operator systems and explain the connection to Connes' embedding problem. We go on by employing the language of operator systems to elaborate on the requirements for the existence of finite-dimensional approximations for general quantum systems. We furthermore introduce the concept of ultraproducts of operator systems and use it to study the implications of two different kinds of finite-dimensional approximations. Note from Publisher: This article contains the abstract only.