- This article is about the transpose of a matrix. For other uses, see Transposition
- reflect A over its main diagonal (which runs top-left to bottom-right) to obtain AT
- write the rows of A as the columns of AT
- write the columns of A as the rows of AT
- visually rotate A 90 degrees clockwise, and mirror the image in a vertical line to obtain AT
Formally, the (i,j) element of AT is the (j,i) element of A.
- [AT]ij = [A]ji
If A is an m × n matrix then AT is a n × m matrix. The transpose of a scalar is the same scalar.
For matrices A, B and scalar c we have the following properties of transpose:
- The transpose respects addition.
- Note that the order of the factors reverses. From this one can deduce that a square matrix A is invertible if and only if AT is invertible, and in this case we have (A−1)T = (AT)−1. It is relatively easy to extend this result to the general case of multiple matrices, where we find that (ABC...XYZ)T = ZTYTXT...CTBTAT.
- The determinant of a square matrix is the same as that of its transpose.
- The dot product of two column vectors a and b can be computed as
- If A has only real entries, then ATA is a positive-semidefinite matrix.
- The transpose of an invertible matrix is also invertible, and its inverse is the transpose of the inverse of the original matrix. The notation is often used to represent either of these equivalent expressions.
- If A is a square matrix, then its eigenvalues are equal to the eigenvalues of its transpose.
Special transpose matrices
A square matrix whose transpose is equal to itself is called a symmetric matrix; that is, A is symmetric if
A square matrix whose transpose is also its inverse is called an orthogonal matrix; that is, G is orthogonal if
- the identity matrix, i.e. GT = G-1.
A square matrix whose transpose is equal to its negative is called skew-symmetric matrix; that is, A is skew-symmetric if
Transpose of linear maps
Here, BV and BW are the bilinear forms on V and W respectively. The matrix of the transpose of a map is the transposed matrix only if the bases are orthonormal with respect to their bilinear forms.
Over a complex vector space, one often works with sesquilinear forms instead of bilinear (conjugate-linear in one argument). The transpose of a map between such spaces is defined similarly, and the matrix of the transpose map is given by the conjugate transpose matrix if the bases are orthonormal. In this case, the transpose is also called the Hermitian adjoint.
If V and W do not have bilinear forms, then the transpose of a linear map f: V→W is only defined as a linear map tf : W*→V* between the dual spaces of W and V.
This means that the transpose (and even the orthogonal group) can be defined abstractly, and completely without reference to matrices (nor the components thereof). If f: V→W then for any o:W→F (that is, any o belonging to W*), if Tf(o) is defined as o composed with f then it will map V→F (that is, Tf will map W* to V*). If the vector spaces have metrics then V* can be uniquely mapped to V, etc, such that we can immediately consider whether or not fT : W→V is equal to f - 1:W→V.
As a shorthand for contraction with the metric tensor
Introductory linear algebra generally does not distinguish between the notion of a vector and a dual vector. Once that distinction is made, many common expressions seem to be freely transposing vectors to create dual vectors, in seeming disregard for the distinction. For example, this is the case in defining the inner product as
What is going on here is that is a notational shortcut for tensor contraction with the metric tensor. Using the Einstein summation convention, with regular (contravariant) vectors having upper indices, this is computing
with the metric tensor for the Euclidean metric being the Kronecker delta. In other words, the notation to create a dual vector is really shorthand:
with the assumption that gij = δij.
Implementation of matrix transposition on computers
On a computer, one can often avoid explicitly transposing a matrix in memory by simply accessing the same data in a different order. For example, software libraries for linear algebra, such as BLAS, typically provide options to specify that certain matrices are to be interpreted in transposed order to avoid the necessity of data movement.
However, there remain a number of circumstances in which it is necessary or desirable to physically reorder a matrix in memory to its transposed ordering. For example, with a matrix stored in row-major order, the rows of the matrix are contiguous in memory and the columns are discontiguous. If repeated operations need to be performed on the columns, for example in a fast Fourier transform algorithm, transposing the matrix in memory (to make the columns contiguous) may improve performance by increasing memory locality.
Ideally, one might hope to transpose a matrix with minimal additional storage. This leads to the problem of transposing an N × M matrix in-place, with O(1) additional storage or at most storage much less than MN. For N ≠ M, this involves a complicated permutation of the data elements that is non-trivial to implement in-place. Therefore efficient in-place matrix transposition has been the subject of numerous research publications in computer science, starting in the late 1950s, and several algorithms have been developed.
- MIT Linear Algebra Lecture on Matrix Transposes
- Transpose, mathworld.wolfram.com
- Transpose, planetmath.org
- Khan Academy introduction to matrix transposes
Topics related to linear algebraScalar · Vector · Vector space · Vector projection · Linear span · Linear map · Linear projection · Linear independence · Linear combination · Basis · Column space · Row space · Dual space · Orthogonality · Rank · Minor · Kernel · Eigenvalues and eigenvectors · Least squares regressions · Outer product · Inner product space · Dot product · Transpose · Gram–Schmidt process · Matrix decomposition
Wikimedia Foundation. 2010.