| Kraus representation | |
|---|---|
| Name | Kraus representation |
| Field | Quantum mechanics |
| Introduced | 1971 |
| Introduced by | Karl Kraus |
Kraus representation
The Kraus representation is a mathematical description of general quantum operations acting on the state of a quantum system. It expresses a linear, completely positive map on density operators as a sum over operator elements (Kraus operators), providing a bridge between unitary evolution on a larger Hilbert space and effective, non-unitary dynamics of subsystems. The representation underlies most treatments of open quantum systems, quantum information theory, and quantum noise modeling.
The Kraus representation characterizes the allowed transformations of a density matrix ρ under physical processes that include interactions with an environment, measurements, or noise. Physically, it corresponds to starting with a global unitary operator acting on the combined system and an ancillary environment (often modeled as a bath), then tracing out the environment degrees of freedom to get the reduced state of the system. This procedure links the representation to concepts such as decoherence, quantum measurement described by positive operator-valued measure (POVM) elements, and the Stinespring dilation theorem which formalizes the ancillary construction. The representation clarifies when non-unitary effects still preserve positivity and probabilistic interpretation of quantum states, reflecting constraints from complete positivity and trace preservation.
Mathematically, a Kraus representation of a linear map E on density operators on a Hilbert space H is given by E(ρ) = Σ_i K_i ρ K_i†, where {K_i} are the Kraus operators acting on H. The operators are not unique: different sets {K_i} related by an isometry or unitary mixing produce the same map. The representation exposes how rank, operator-sum decomposition, and spectral properties determine dynamical behavior. The first rigorous study of operator-sum forms appears in works by Karl Kraus and is connected to operator algebra results from GNS construction and the Choi–Jamiolkowski isomorphism that maps superoperators to bipartite states in H⊗H.
Kraus representations precisely describe completely positive maps: a linear map is completely positive if and only if it admits an operator-sum (Kraus) decomposition. If in addition Σ_i K_i† K_i = I, the map is trace-preserving (CPTP), representing a deterministic quantum channel. This class of maps coincides with quantum channels as used in quantum communication and quantum computing models such as the quantum circuit model. The requirement of complete positivity ensures physical consistency under inclusion of ancillary systems as emphasized in foundational studies by various authors and in formulations by A. S. Holevo and G. Lindblad for open-system dynamics.
Constructive methods to obtain Kraus operators include performing a Stinespring dilation using an explicit unitary on system-plus-environment, diagonalizing the Choi matrix via the Choi matrix (the Choi–Jamiolkowski state) to extract operators, or using operator-spectral decompositions. Minimal Kraus representations correspond to the rank of the Choi matrix and yield the smallest number of nonzero Kraus operators. Canonical forms include diagonal or operator-Schmidt decompositions tied to singular value decomposition techniques. For dynamical semigroups, the Lindblad equation provides a generator whose finite-time maps admit Kraus representations when exponentiated; derivations often involve tools from semigroup theory and C*-algebra methods.
Kraus representations are central to modeling quantum noise channels such as depolarizing channel, amplitude damping channel, and phase damping. They underpin analysis of quantum error correction codes, where noise is represented by Kraus errors and syndromes are designed to detect operator-sum components. In quantum tomography, process tomography reconstructs Kraus operators or the Choi matrix from experimental data; methods include maximum likelihood estimation and compressed sensing techniques developed at institutions like IBM and Google quantum laboratories. The representation is also used in studies of entanglement dynamics, quantum thermodynamics of open devices, and implementations in platforms such as trapped ion quantum computers, superconducting qubits, and photonic quantum information systems.
Standard illustrative examples expressed with explicit Kraus operators include: - The amplitude damping channel modeling energy relaxation, with operators derived from a two-level system coupling to a bath. - The bit flip, phase flip, and depolarizing channel families used in theoretical error models for quantum error correction. - The measure-and-prepare channel describing classical communication of measurement outcomes, implemented via POVMs and conditional state preparation. Each example can be related to an underlying system–environment unitary; experimental realizations are reported in literature from groups at Caltech, MIT, and University of Oxford.
Extensions of the Kraus formalism include descriptions for infinite-dimensional Hilbert spaces using forms from operator theory and von Neumann algebras, frameworks for non-Markovian dynamics that go beyond CPTP semigroups (e.g., memory kernels and time-convolutionless master equations), and categorical formulations in quantum foundations and categorical quantum mechanics. Related mathematical structures include the Choi–Jamiolkowski isomorphism, Stinespring dilation theorem, and the Lindblad master equation. Computational and algorithmic approaches to manipulate Kraus decompositions connect to software projects such as Qiskit, QuTiP, and research groups at Perimeter Institute and National Institute of Standards and Technology (NIST). The representation remains a foundational tool in both theoretical and experimental studies of quantum channels, decoherence control, and quantum technologies.