Unitary matrices

Unitary matrices
Name	Unitary matrices
Field	Linear algebra
Properties	Invertible, norm-preserving, diagonalizable by unitary similarity
Related	Orthogonal matrix, Hermitian matrix, Normal matrix

Unitary matrices

Unitary matrices are square complex matrices U that satisfy U*U† = I and serve as the complex analogue of Orthogonal matrix in contexts spanning Quantum mechanics, Signal processing, and Representation theory. They preserve the standard inner product, implement change of orthonormal basis in Hilbert space, and appear naturally in the study of symmetry groups such as the Unitary group and in decompositions like the Singular value decomposition. Historically they connect to work by David Hilbert, John von Neumann, Eugene Wigner, and methods developed at institutions like Institute for Advanced Study and Bell Labs.

Definition

A matrix U ∈ C^{n×n} is unitary if U†U = UU† = I, where † denotes the conjugate transpose; this condition parallels definitions used by Carl Friedrich Gauss in quadratic forms and by Évariste Galois in group actions. The set of all n×n unitary matrices forms the Lie group U(n), studied in the context of Hermann Weyl’s representation theory and in classifications used by Élie Cartan and Claude Chevalley. Unitarity implies invertibility with U^{-1} = U†, a fact exploited in algorithms developed at AT&T Bell Laboratories and in numerical libraries originating from work at Argonne National Laboratory.

Basic properties

Unitary matrices preserve Euclidean norms and inner products, so for vectors x,y they satisfy ⟨Ux,Uy⟩ = ⟨x,y⟩, a property central to applications in Paul Dirac’s bra–ket formalism, Richard Feynman path-integral techniques, and in optical systems studied by Augustin-Jean Fresnel. Unitaries are normal (UU† = U†U) and thus diagonalizable by a unitary similarity transformation, a consequence used in the Spectral theorem treatments by Issai Schur and John von Neumann. The determinant of a unitary lies on the complex unit circle; determinants feature in topological classifications like those by Michael Atiyah and Raoul Bott and in invariants studied by Simon Donaldson.

Eigenvalues and spectral decomposition

All eigenvalues of a unitary matrix have modulus one, so they lie on the complex unit circle; this property figures in analyses by Hermann Weyl and in the study of time evolution operators formalized by Max Born and Werner Heisenberg. A unitary U admits a spectral decomposition U = VDV† with D diagonal of unimodular eigenvalues and V unitary; this decomposition underpins techniques in John von Neumann’s quantum measurement theory and in matrix functional calculus used in Nicholas Metropolis’s computational frameworks. Continuous functional calculus for unitaries is applied in index theorems developed by Atiyah–Singer collaborators and in scattering theory elaborated by Ludwig Faddeev.

Examples and special classes

Standard examples include permutation matrices studied by Évariste Galois through permutation groups, discrete Fourier transform matrices central to work by Joseph Fourier and adopted by Claude Shannon in information theory, and diagonal phase matrices used in constructions by Paul Dirac. Special classes include the special unitary group SU(n) studied by Murray Gell-Mann and Yoichiro Nambu in particle physics, block-diagonal unitaries appearing in decomposition results by Alfred Horn, and Householder reflectors implemented in numerical linear algebra packages influenced by Gene H. Golub and Lloyd Trefethen. Circulant unitary matrices tie to harmonic analysis traditions from Norbert Wiener and Andrey Kolmogorov.

Operations and calculus (products, tensor, exponentials)

The product of unitary matrices is unitary, forming the group structure used in gauge theories by Chen Ning Yang and Robert Mills, and inverses are given by conjugate transpose as in group treatments by Élie Cartan. Tensor products of unitaries produce larger unitaries; this is fundamental in multi-qubit constructions by Peter Shor and in quantum circuit theory advanced by Richard Jozsa and David Deutsch. Matrix exponentials of skew-Hermitian operators yield unitaries, connecting to Lie algebra–Lie group correspondences studied by Sophus Lie and used in time-evolution operators developed by Paul Dirac. Numerical exponentiation algorithms trace lineage to methods at Los Alamos National Laboratory and to Krylov-subspace techniques elaborated by Youcef Saad.

Applications

Unitary matrices implement quantum gates in Quantum computing architectures proposed by Peter Shor, Lov Grover, and laboratories at IBM and Google. They diagonalize normal operators in signal processing routines used by Alan Turing’s successors and in image compression principles influenced by Abraham Gershun and David Marr. In control theory and vibration analysis they appear in modal decompositions developed at Massachusetts Institute of Technology, and in statistics they underpin orthogonal designs studied by Ronald Fisher and Jerzy Neyman. Unitaries are crucial in scattering matrices of particle physics explored by Murray Gell-Mann and in cryptographic transforms employed in standards from National Institute of Standards and Technology.

Generalizations and related operators

Generalizations include isometries between Hilbert spaces studied by John von Neumann and bounded unitary operators on infinite-dimensional spaces analyzed by Marshall Stone and Nelson Dunford. Related classes are orthogonal matrices over R connected to Carl Friedrich Gauss, symplectic matrices appearing in Hamiltonian mechanics studied by William Rowan Hamilton, and normal matrices central to work by Issai Schur. Nonunitary generalizations appear in contraction semigroups used in dissipative quantum systems investigated by Ilya Prigogine and in quasi-unitary extensions considered in scattering theory by Mark Krein.

Category:Linear algebra