| tensor product | |
|---|---|
| Name | Tensor product |
| Caption | Schematic of tensor product of vector spaces |
| Field | Linear algebra; Functional analysis; Quantum mechanics |
| Introduced | 19th century |
| Related | Kronecker product; Tensor (mathematics); Hilbert space |
tensor product
The tensor product is a bilinear operation that combines two linear spaces or modules into a new one whose elements represent formal multilinear combinations of pairs of vectors. In Quantum mechanics and Quantum information theory, tensor products provide the mathematical structure for composing independent subsystems, encoding correlations and enabling phenomena such as quantum entanglement. The construction is central to formulations using Hilbert spaces and operators.
The tensor product V ⊗ W of two vector spaces V and W (over the same field) is defined by a universal bilinear map V × W → V ⊗ W such that any bilinear map from V × W factors uniquely through a linear map from V ⊗ W. Algebraic properties include bilinearity, associativity up to canonical isomorphism, and distributivity over direct sums: V ⊗ (W ⊕ X) ≅ (V ⊗ W) ⊕ (V ⊗ X). For finite-dimensional spaces dim(V ⊗ W) = dim(V)·dim(W). The construction generalizes to modules over rings, C*-algebras, and topological vector spaces. Important named results and tools include the universal property, the Kronecker product for matrices, and coherence isomorphisms used in category-theoretic treatments such as in Mac Lane's work on Category theory.
For Hilbert spaces H1 and H2 the Hilbert space tensor product H1 ⊗ H2 is obtained by completing the algebraic tensor product with respect to the inner product defined on simple tensors by ⟨u1⊗u2, v1⊗v2⟩ = ⟨u1,v1⟩⟨u2,v2⟩ and extending by linearity. The resulting separable Hilbert spaces underpin the mathematical formulation of composite quantum systems in the Dirac notation formalism popularized by Paul Dirac. In infinite dimensions, choices of completion and topologies (projective vs injective tensor norms) matter; constructions are often formulated in functional analysis and in the theory of von Neumann algebra tensor products for operator algebras studied at institutions such as von Neumann's foundational work and later by researchers at Institute for Advanced Study and Princeton University.
In quantum theory the state space of two subsystems A and B with Hilbert spaces HA and HB is the tensor product HA ⊗ HB. Product states are simple tensors ψA ⊗ ψB, while superpositions of such tensors can be entangled. Entanglement was formalized in foundational studies by John Bell and exemplified by the Einstein–Podolsky–Rosen paradox and Bell's theorem, with experimental tests by groups including those at Alain Aspect's laboratory and later Anton Zeilinger's teams. Tensor structure also defines locality, subsystem observables, and supports notions like Schmidt decomposition, which gives a canonical form for pure bipartite states and connects to the concept of reduced density matrixs via the partial trace.
Linear operators on tensor product spaces include simple tensor operators A ⊗ B acting as (A ⊗ B)(ψ ⊗ φ) = (Aψ) ⊗ (Bφ). The algebraic span of such operators generates the full operator algebra B(HA ⊗ HB). The partial trace Tr_B is the unique linear map satisfying Tr_B((A ⊗ I) ρ) = A Tr_B(ρ) for density operators ρ; it produces reduced states and is essential in defining locality and entropic measures such as von Neumann entropy. In quantum computing, tensoring of gates corresponds to composing quantum circuits; tensor methods underlie operator-sum representations and channels studied in completely positive map theory.
The algebraic tensor product is constructed from the free vector space on V × W modulo the subspace enforcing bilinearity; this yields the universal property used to define tensors in algebraic contexts. For topological or metric structures one introduces tensor norms and completes; for Hilbert spaces the inner product completion yields the Hilbert tensor product. Category-theoretic formulations treat ⊗ as a monoidal product; coherence and natural isomorphisms are described in texts by Saunders Mac Lane and in monoidal category theory used in quantum foundations research by authors such as Bob Coecke. For operator algebras, minimal and maximal C*-tensor products and spatial tensor products of von Neumann algebras are distinct constructions with implications for quantum statistical mechanics and the Connes embedding problem.
Tensor product structure is foundational in quantum computing for representing multi-qubit registers (tensoring Pauli matrices, Hadamard gate, etc.). It enables the formalism of quantum error correction (stabilizer codes developed by Daniel Gottesman), entanglement measures, and protocols such as quantum teleportation and superdense coding. In many-body physics tensor networks (e.g., Matrix product state, Projected entangled pair state, Tensor network states) exploit approximate low-rank structure of large tensor products for efficient simulation; relevant research groups include those at Max Planck Institute for Quantum Optics and Caltech. Tensor products also appear in quantum field theory for Fock space constructions and in condensed matter models like the Heisenberg model.
Given bases {e_i} of V and {f_j} of W, the set {e_i ⊗ f_j} forms a basis of V ⊗ W and expansion coefficients follow multilinear algebra rules used in numerical linear algebra packages such as LAPACK and libraries in NumPy and MATLAB. For finite-dimensional operators, the Kronecker product of matrices represents the matrix of A ⊗ B in product bases; this is widely used in simulations of quantum circuits and many-body Hamiltonians. Computational methods include singular value decomposition and Schmidt decomposition for bipartite systems, tensor decomposition algorithms (CP decomposition, Tucker decomposition) used in data-driven quantum chemistry and in software developed by groups at IBM Quantum and Google Quantum AI.
Category:Linear algebra Category:Quantum mechanics