In linear algebra, the outer product typically refers to the tensor product of two vectors. The result of applying the outer product to a pair of coordinate vectors is a matrix. The name contrasts with the inner product, which takes as input a pair of vectors and produces a scalar.
The outer product of vectors can be also regarded as a special case of the Kronecker product of matrices.
Some authors use the expression "outer product of tensors" as a synonym of "tensor product". The outer product is also a higher-order function in some computer programming languages such as APL and Mathematica.
Definition (matrix multiplication)
The outer product u ⊗ v is equivalent to a matrix multiplication uvT, provided that u is represented as a m × 1 column vector and v as a n × 1 column vector (which makes vT a row vector). For instance, if m = 4 and n = 3, then
Contrast with inner product
If m = n, then one can take the matrix product the other way, yielding a scalar (or 1 × 1 matrix):
Definition (vectors and tensors)
Given the vectors
For complex vectors, the complex conjugate of v (denoted v∗ or v̅). Namely, matrix A is obtained by multiplying each element of u by the complex conjugate of each element of v.
The outer product on tensors is typically referred to as the tensor product. Given a tensor a with rank q and dimensions (i1, ..., iq), and a tensor b with rank r and dimensions (j1, ..., jr), their outer product c has rank q + r and dimensions (k1, ..., kq+r) which are the i dimensions followed by the j dimensions. It is denoted in coordinate-free notation using ⊗ and components are defined index notation by:
similarly for higher order tensors:
For example, if A has rank 3 and dimensions (3, 5, 7) and B has rank 2 and dimensions (10, 100), their outer product c has rank 5 and dimensions (3, 5, 7, 10, 100). If A has a component A[2, 2, 4] = 11 and B has a component B[8, 88] = 13, then the component of C formed by the outer product is C[2, 2, 4, 8, 88] = 143.
To understand the matrix definition of outer product in terms of the definition of tensor product:
- The vector v can be interpreted as a rank 1 tensor with dimension M, and the vector u as a rank 1 tensor with dimension N. The result is a rank 2 tensor with dimension (M, N).
- The rank of the result of an inner product between two tensors of rank q and r is the greater of q + r − 2 and 0. Thus, the inner product of two matrices has the same rank as the outer product (or tensor product) of two vectors.
- It is possible to add arbitrarily many leading or trailing 1 dimensions to a tensor without fundamentally altering its structure. These 1 dimensions would alter the character of operations on these tensors, so any resulting equivalences should be expressed explicitly.
- The inner product of two matrices V with dimensions (d, e) and U with dimensions (e, f) is , where i = 1, 2, ..., d and k = 1, 2, ..., f. For the case where e = 1, the summation is trivial (involving only a single term).
- The outer product of two matrices V with dimensions (m, n) and U with dimensions (p, q) is , where s = 1, 2, ..., mp − 1, mp and t = 1, 2, ..., nq − 1, nq.
Here y∗(w) denotes the value of the linear functional y∗ (which is an element of the dual space of W) when evaluated at the element w ∈ W. This scalar in turn is multiplied by x to give as the final result an element of the space V.
If V and W are finite-dimensional, then the space of all linear transformations from W to V, denoted Hom(W, V), is generated by such outer products; in fact, the rank of a matrix is the minimal number of such outer products needed to express it as a sum (this is the tensor rank of a matrix). In this case Hom(W, V) is isomorphic to W∗ ⊗ V.
Contrast with inner product
If W = V, then one can also pair the covector w∗ ∈ V∗ with the vector v ∈ V via (w∗, v) → w∗(v), which is the duality pairing between V and its dual, sometimes called the inner product.
The outer product is useful in computing physical quantities (e.g., the tensor of inertia), and performing transform operations in digital signal processing and digital image processing. It is also useful in statistical analysis for computing the covariance and auto-covariance matrices for two random variables.
- Linear Algebra (4th Edition), S. Lipcshutz, M. Lipson, Schaum’s Outlines, McGraw Hill (USA), 2009, ISBN 978-0-07-154352-1
- Encyclopaedia of Physics (2nd Edition), R.G. Lerner, G.L. Trigg, VHC publishers, 1991, (Verlagsgesellschaft) 3-527-26954-1, (VHC Inc.) 0-89573-752-3
- Mathematical methods for physics and engineering, K.F. Riley, M.P. Hobson, S.J. Bence, Cambridge University Press, 2010, ISBN 978-0-521-86153-3