Mathematical formulation of quantum mechanics
The mathematical formulation of quantum mechanics represents physical states, observables, transformations, and measurement statistics through structures drawn primarily from functional analysis, operator theory, and probability theory. In the standard Hilbert-space formulation, a physical system is associated with a complex Hilbert space. Pure states correspond to rays in that space, mixed states correspond to normalized positive operators, and observables are represented by self-adjoint operators. The theory’s probabilistic content follows from the Born rule, while its temporal content is expressed through unitary evolution or, for open systems, through more general completely positive maps.
Several mathematically equivalent formulations arose during the initial development of quantum theory. Werner Heisenberg, Max Born, and Pascual Jordan developed matrix mechanics, whereas Erwin_Schrödinger introduced wave mechanics. Their equivalence was incorporated into the transformation theory associated with Paul Dirac and Jordan, and subsequently into the abstract operator formulation developed by John von Neumann. Later mathematical treatments distinguish carefully between bounded and unbounded operators, between vectors and rays, and between ordinary eigenvectors and generalized states belonging to larger distribution spaces.
Historical synthesis
Early quantum theory used discrete transition amplitudes and differential wave equations as apparently different mathematical languages. In matrix mechanics, the position and momentum of a system were represented by noncommuting arrays satisfying the canonical relation
[ [\hat Q,\hat P]=i\hbar I, ]
where (I) denotes the identity operator and (\hbar) is the reduced Planck constant. Wave mechanics represented the same quantities by multiplication and differentiation operators acting on square-integrable functions. Schrödinger established the correspondence between the energy eigenvalue problem in wave mechanics and the spectral problem underlying matrix mechanics.
Dirac and Jordan developed transformation theory to express physical predictions independently of a chosen basis. This treatment introduced a systematic relation between amplitudes, operator representations, and changes of coordinates in state space. Dirac’s notation placed vectors and continuous-spectrum eigenstates within a common symbolic framework, although the distributional interpretation of those expressions was formalized later through rigged Hilbert spaces.
Von Neumann supplied a unified account based on Hilbert spaces, projection operators, and the spectral theorem. During the same period, the statistical description of non-pure states acquired an operator form. In 1927, Lev Landau, John von Neumann, and You Watanabe independently represented statistical ensembles by positive trace-class operators of unit trace. This density-operator formulation separated uncertainty arising from quantum superposition from uncertainty associated with ensemble preparation and became the standard language for subsystems and incomplete state information.
The subsequent development of mathematical quantum mechanics refined the domains of unbounded operators and the convergence of operator sequences. Hermann Weyl replaced the formal canonical commutation relation with an exponentiated relation between unitary operators, while Marshall Stone related continuous one-parameter unitary groups to self-adjoint generators. These results clarified the mathematical conditions under which Hamiltonians generate time evolution.
States and Hilbert-space structure
A complex Hilbert space (\mathcal H) consists of a complex vector space equipped with an inner product and complete in the norm induced by that product. Two nonzero vectors (\lvert\psi\rangle) and (c\lvert\psi\rangle), where (c) is a nonzero complex number, determine the same pure physical state. Normalized vectors therefore retain an arbitrary overall phase, and the actual pure-state space is the projective Hilbert space associated with (\mathcal H).
For normalized representatives, the transition probability between pure states is
[ P(\phi\mid\psi)=|\langle\phi\mid\psi\rangle|^2. ]
This expression depends only on the corresponding rays. Transformations preserving all transition probabilities are represented, according to Wigner's theorem, by either unitary or antiunitary operators. Continuous time evolution belongs to the unitary case.
A mixed state is represented by a density operator (\rho) satisfying
[ \rho\geq 0, \qquad \operatorname{Tr}\rho=1. ]
Every convex decomposition
[ \rho=\sum_j p_j\lvert\psi_j\rangle\langle\psi_j\rvert ]
describes an ensemble yielding the same expectation values whenever the probabilities (p_j) are nonnegative and sum to unity. Distinct decompositions of a density operator do not in general correspond to empirically distinct states. A density operator represents a pure state precisely when it is a rank-one projection, equivalently when (\rho^2=\rho).
In the algebraic formulation, states are normalized positive linear functionals on a C*-algebra of observables. Density operators describe the normal states of the standard operator algebra on a Hilbert space, while the functional formulation also accommodates representations encountered in systems with infinitely many degrees of freedom.
Observables and the Born rule
An observable is represented in the conventional formulation by a self-adjoint operator (A) on (\mathcal H). Bounded observables are defined on the entire Hilbert space, whereas operators representing position, momentum, and energy are commonly unbounded and require explicitly specified dense domains. Self-adjointness is stronger than symmetry and provides the spectral structure needed for real measurement values and unitary evolution.
The spectral theorem associates a projection-valued measure (E_A) with each self-adjoint operator:
[ A=\int_{\mathbb R}\lambda,dE_A(\lambda). ]
For a state (\rho), the Born rule assigns the probability
[ \Pr(A\in\Delta)=\operatorname{Tr}!\left(\rho E_A(\Delta)\right) ]
to the result lying in a Borel set (\Delta\subseteq\mathbb R). When the state is pure, the same probability becomes
[ \Pr(A\in\Delta)
\langle\psi\mid E_A(\Delta)\psi\rangle. ]
The expectation value of an observable is
[ \langle A\rangle_\rho=\operatorname{Tr}(\rho A), ]
provided that the operator product has the required trace properties. For an unbounded observable in a pure state, the corresponding expression requires that the state vector belong to the appropriate operator or quadratic-form domain.
Projection-valued measures describe idealized sharp observables. A positive-operator-valued measure replaces projections with positive operators (F(\Delta)) satisfying countable additivity and (F(\mathbb R)=I). The probability law then takes the form
[ \Pr(F\in\Delta)=\operatorname{Tr}!\left(\rho F(\Delta)\right). ]
This generalized structure includes measurements whose outcomes combine quantum statistics with finite resolution or indirect coupling to an auxiliary system. The change of state associated with a measurement is described more fully by a quantum instrument, which assigns a completely positive operation to each measurable outcome set.
Dynamics
For a closed system, time evolution is represented by a strongly continuous one-parameter unitary group (U(t)). Stone’s theorem associates such a group with a self-adjoint Hamiltonian (H):
[ U(t)=e^{-iHt/\hbar}. ]
In the Schrödinger picture, a pure state evolves according to
[ i\hbar\frac{d}{dt}\lvert\psi(t)\rangle
H\lvert\psi(t)\rangle, ]
while a density operator satisfies the von Neumann equation,
[ i\hbar\frac{d\rho}{dt}=[H,\rho]. ]
In the Heisenberg picture, the state remains fixed and an observable without explicit time dependence evolves as
[ A(t)=U(t)^\dagger A,U(t), ]
which formally yields
[ \frac{dA(t)}{dt}
\frac{i}{\hbar}[H,A(t)]. ]
The Schrödinger and Heisenberg pictures assign the same probabilities because they relocate the unitary transformation between states and observables without changing their pairing.
An open quantum system is described by a reduced state obtained after disregarding environmental degrees of freedom. Its admissible state transformations are completely positive and trace preserving. Under assumptions of time homogeneity and Markovian composition, a continuous dynamical semigroup has a generator of Gorini–Kossakowski–Sudarshan–Lindblad form:
[ \frac{d\rho}{dt}
-\frac{i}{\hbar}[H,\rho] + \sum_k \left( L_k\rho L_k^\dagger -\frac12 L_k^\dagger L_k\rho -\frac12\rho L_k^\dagger L_k \right). ]
The first term generates unitary dynamics, while the remaining terms represent nonunitary changes produced by coupling to unobserved degrees of freedom.
Composite systems and entanglement
The state space of a composite system is the tensor product
[ \mathcal H_{AB}=\mathcal H_A\otimes\mathcal H_B. ]
A product state has the form (\lvert\psi_A\rangle\otimes\lvert\psi_B\rangle). States that cannot be expressed as product states are entangled, and their subsystem statistics are described by reduced density operators. For a joint state (\rho_{AB}), the state of subsystem (A) is
[ \rho_A=\operatorname{Tr}B(\rho{AB}), ]
where (\operatorname{Tr}_B) denotes the partial trace over subsystem (B).
A pure composite state can therefore induce a mixed state on either component. The Schmidt decomposition expresses a bipartite pure state as
[ \lvert\Psi\rangle
\sum_j \sqrt{\lambda_j}, \lvert a_j\rangle\otimes\lvert b_j\rangle, ]
where the coefficients (\lambda_j) are nonnegative and sum to unity. More than one nonzero Schmidt coefficient characterizes entanglement for a bipartite pure state.
For identical particles, the physical state space is restricted by permutation symmetry. Integer-spin particles occupy the symmetric subspace associated with Bose–Einstein statistics, whereas half-integer-spin particles occupy the antisymmetric subspace associated with Fermi–Dirac statistics. In relativistic quantum theory, this relation is formalized by the spin–statistics theorem.
Canonical commutation relations and representations
For a finite number of canonical degrees of freedom, the Weyl form of the commutation relations is expressed through unitary operators generated by position and momentum. Under regularity and irreducibility assumptions, the Stone–von Neumann theorem states that these representations are unitarily equivalent to the Schrödinger representation. This result provides a precise sense in which matrix and wave formulations describe the same finite-dimensional canonical structure.
The theorem does not extend to systems with infinitely many degrees of freedom. Quantum field theory consequently admits unitarily inequivalent representations of the observable algebra. Different representations can encode distinct vacuum structures, thermal states, or phases of an extended system. The algebraic formulation treats the observable algebra as primary and obtains Hilbert-space realizations from states through the Gelfand–Naimark–Segal construction.
Unbounded operators create additional domain questions even in nonrelativistic systems. Formal products and commutators need not be defined on every vector, while self-adjoint extensions of the same symmetric differential operator can correspond to different boundary conditions. The mathematical specification of a quantum model therefore includes the Hilbert space, the operator domains, and the representation of the relevant observable algebra.
Relation among formulations
The Hilbert-space, path-integral, and phase-space formulations organize the same quantum content through different mathematical objects. The path integral formulation represents transition amplitudes through a weighted sum over histories and connects quantum dynamics with classical action functionals. The Wigner quasiprobability distribution maps density operators to functions on phase space, where operator multiplication becomes a noncommutative star product. The algebraic formulation emphasizes relations among observables and remains applicable when no single preferred Hilbert-space representation exists.
These formulations agree when their mathematical domains and correspondence maps are defined consistently. Their differences concern representation, analytic control, and the types of systems for which each language supplies a direct description. The common structure consists of noncommutative observables, positive normalized states, probabilistic evaluation rules, and transformations preserving the admissible state space.