Other meanings of Kronecker product
Mathematics
The Kronecker product, denoted by ⊗, is a matrix operation that produces a block matrix from two matrices of arbitrary sizes. If A is an m×n matrix and B is a p×q matrix, then the Kronecker product A⊗B is an mp×nq matrix defined by replacing each element aij of A with the block aijB. This operation, named after the German mathematician Leopold Kronecker, is also known as the matrix direct product or the tensor product of matrices. It is fundamental in linear algebra and finds applications in statistics, physics, and engineering, particularly in the study of linear matrix equations and the vectorization of matrices.
The Kronecker product is defined for any two matrices, regardless of their dimensions. For A = [aij] of size m×n and B of size p×q, the product A⊗B is the mp×nq block matrix [aijB]. For example, if A is 2×2 and B is 2×2, the result is a 4×4 matrix. This operation is bilinear and associative, meaning (A⊗B)⊗C = A⊗(B⊗C) and (A+B)⊗C = A⊗C + B⊗C, provided the sums are defined.1
Key properties include the mixed-product rule: (A⊗B)(C⊗D) = (AC)⊗(BD), when the ordinary products exist. The transpose satisfies (A⊗B)T = AT⊗BT, and the inverse, when A and B are invertible, is (A⊗B)−1 = A−1⊗B−1. The eigenvalues of A⊗B are all products λiμj, where λi are eigenvalues of A and μj of B.
The Kronecker product is indispensable in solving linear matrix equations such as AXB = C. By vectorizing the matrices, the equation becomes (BT⊗A)vec(X) = vec(C), transforming it into a standard linear system.2 This technique is central to the study of Sylvester and Lyapunov equations, which arise in control theory and stability analysis.
In statistics, the Kronecker product appears in the covariance structure of multivariate normal distributions, particularly in the analysis of repeated measures and spatial data. For example, a separable covariance matrix can be expressed as Σ = Σtime⊗Σspace, simplifying estimation and computation.3 It also underlies the design of experiments, where the Kronecker product of design matrices represents factorial designs.
In quantum mechanics, the Kronecker product, often called the tensor product, is used to describe composite systems. The state space of a system composed of two subsystems is the tensor product of their individual state spaces, and operators on the composite system are Kronecker products of operators on the subsystems.4 This formalism is essential for understanding entanglement and quantum computation.
In signal processing, the Kronecker product is used in the construction of multidimensional filter banks and in the representation of separable transforms, such as the two-dimensional discrete Fourier transform, which can be expressed as a Kronecker product of one-dimensional transforms. This property enables efficient algorithms for image and video processing.
The Kronecker product is sometimes called the Zehfuss product, after the German mathematician Johann Georg Zehfuss, who studied it in 1858, independently of Kronecker. Kronecker himself never published a systematic treatment; the name became standard through later works.
An edge case: the Kronecker product of a matrix with a 0×0 matrix yields a 0×0 matrix, which is consistent with the definition but often overlooked. Also, the Kronecker product of two vectors is equivalent to the outer product, and the vectorization of a matrix product relates to the Kronecker product via vec(ABC) = (CT⊗A)vec(B). This identity is widely used in matrix calculus but is frequently misremembered.5
The Kronecker product is also known as the matrix direct product.
Help improve the encyclopedia. Reports go straight to the site manager.