Moore–Penrose inverse

From Wikipedia, the free encyclopedia
  (Redirected from Moore–Penrose pseudoinverse)
Jump to navigation Jump to search

In mathematics, and in particular linear algebra, a pseudoinverse A+ of a matrix A is a generalization of the inverse matrix.[1] The most widely known type of matrix pseudoinverse is the Moore–Penrose inverse,[2][3][4][5] which was independently described by E. H. Moore[6] in 1920, Arne Bjerhammar[7] in 1951, and Roger Penrose[8] in 1955. Earlier, Erik Ivar Fredholm had introduced the concept of a pseudoinverse of integral operators in 1903. When referring to a matrix, the term pseudoinverse, without further specification, is often used to indicate the Moore–Penrose inverse. The term generalized inverse is sometimes used as a synonym for pseudoinverse.

A common use of the pseudoinverse is to compute a 'best fit' (least squares) solution to a system of linear equations that lacks a unique solution (see below under § Applications). Another use is to find the minimum (Euclidean) norm solution to a system of linear equations with multiple solutions. The pseudoinverse facilitates the statement and proof of results in linear algebra.

The pseudoinverse is defined and unique for all matrices whose entries are real or complex numbers. It can be computed using the singular value decomposition.

Notation[edit]

In the following discussion, the following conventions are adopted.

  • will denote one of the fields of real or complex numbers, denoted , , respectively. The vector space of matrices over is denoted by .
  • For , and denote the transpose and Hermitian transpose (also called conjugate transpose) respectively. If , then .
  • For , denotes the range (image) of (the space spanned by the column vectors of ) and denotes the kernel (null space) of .
  • Finally, for any positive integer , denotes the identity matrix.

Definition[edit]

For , a pseudoinverse of is defined as a matrix satisfying all of the following four criteria, known as the Moore-Penrose conditions:[8][9]

  1.       (AA+ need not be the general identity matrix, but it maps all column vectors of A to themselves);
  2.       (A+ is a weak inverse for the multiplicative semigroup);
  3.       (AA+ is Hermitian); and
  4.       (A+A is also Hermitian).

exists for any matrix , but when the latter has full rank, can be expressed as a simple algebraic formula.

In particular, when has linearly independent columns (and thus matrix is invertible), can be computed as:

This particular pseudoinverse constitutes a left inverse, since, in this case, .

When has linearly independent rows (matrix is invertible), can be computed as:

This is a right inverse, as .

Properties[edit]

Proofs for some of these facts may be found on a separate page, Proofs involving the Moore–Penrose inverse.

Existence and uniqueness[edit]

  • The pseudoinverse exists and is unique: for any matrix , there is precisely one matrix , that satisfies the four properties of the definition.[9]

A matrix satisfying the first condition of the definition is known as a generalized inverse. If the matrix also satisfies the second definition, it is called a generalized reflexive inverse. Generalized inverses always exist but are not in general unique. Uniqueness is a consequence of the last two conditions.

Basic properties[edit]

  • If has real entries, then so does .
  • If is invertible, its pseudoinverse is its inverse. That is: .[10]:243
  • The pseudoinverse of a zero matrix is its transpose.
  • The pseudoinverse of the pseudoinverse is the original matrix: .[10]:245
  • Pseudoinversion commutes with transposition, conjugation, and taking the conjugate transpose:[10]:245
  • The pseudoinverse of a scalar multiple of A is the reciprocal multiple of A+:
for

Identities[edit]

The following identities can be used to cancel certain subexpressions or expand expressions involving pseudoinverses. Proofs for these properties can be found in the proofs subpage.

Reduction to Hermitian case[edit]

The computation of the pseudoinverse is reducible to its construction in the Hermitian case. This is possible through the equivalences:

as and are Hermitian.

Products[edit]

If , and if

  • has orthonormal columns (i.e., ), or
  • has orthonormal rows (i.e., ), or
  • has all columns linearly independent (full column rank) and has all rows linearly independent (full row rank), or
  • (i.e., is the conjugate transpose of ),

then

.

The last property yields the equivalences:

Projectors[edit]

and are orthogonal projection operators – that is, they are Hermitian (, ) and idempotent ( and ). The following hold:

  • and
  • is the orthogonal projector onto the range of (which equals the orthogonal complement of the kernel of ).
  • is the orthogonal projector onto the range of (which equals the orthogonal complement of the kernel of ).
  • is the orthogonal projector onto the kernel of .
  • is the orthogonal projector onto the kernel of .[9]

The last two properties imply the following identities:

Another property is the following: if is Hermitian and idempotent (true if and only if it represents an orthogonal projection), then, for any matrix the following equation holds:[11]

This can be proven by defining matrices , , and checking that is indeed a pseudoinverse for by verifying that the defining properties of the pseudoinverse hold, when is Hermitian and idempotent.

From the last property it follows that, if is Hermitian and idempotent, for any matrix

Finally, it should be noted that if is an orthogonal projection matrix, then its pseudoinverse trivially coincides with the matrix itself, i.e. .

Geometric construction[edit]

If we view the matrix as a linear map over a field then can be decomposed as follows. We write for the direct sum, for the orthogonal complement, for the kernel of a map, and for the image of a map. Notice that and . The restriction is then an isomorphism. These imply that is defined on to be the inverse of this isomorphism, and on to be zero.

In other words: To find for given b in Km, first project b orthogonally onto the range of A, finding a point p(b) in the range. Then form A−1({p(b)}), i.e. find those vectors in Kn that A sends to p(b). This will be an affine subspace of Kn parallel to the kernel of A. The element of this subspace that has the smallest length (i.e. is closest to the origin) is the answer we are looking for. It can be found by taking an arbitrary member of A−1({p(b)}) and projecting it orthogonally onto the orthogonal complement of the kernel of A.

This description is closely related to the Minimum norm solution to a linear system.

Subspaces[edit]

Limit relations[edit]

  • The pseudoinverse are limits:
(see Tikhonov regularization). These limits exist even if or do not exist.[9]:263

Continuity[edit]

  • In contrast to ordinary matrix inversion, the process of taking pseudoinverses is not continuous: if the sequence converges to the matrix A (in the maximum norm or Frobenius norm, say), then (An)+ need not converge to A+. However, if all the matrices have the same rank, (An)+ will converge to A+.[12]

Derivative[edit]

The derivative of a real valued pseudoinverse matrix which has constant rank at a point may be calculated in terms of the derivative of the original matrix:[13]

Examples[edit]

Since for invertible matrices the pseudoinverse equals the usual inverse, only examples of non-invertible matrices are considered below.

  • For the pseudoinverse is (Generally, the pseudoinverse of a zero matrix is its transpose.) The uniqueness of this pseudoinverse can be seen from the requirement , since multiplication by a zero matrix would always produce a zero matrix.
  • For the pseudoinverse is
    Indeed, and thus
    Similarly, and thus
  • For
  • For (The denominators are .)
  • For
  • For the pseudoinverse is
    Note that for this matrix, the left inverse exists and thus equals , indeed,

Special cases[edit]

Scalars[edit]

It is also possible to define a pseudoinverse for scalars and vectors. This amounts to treating these as matrices. The pseudoinverse of a scalar x is zero if x is zero and the reciprocal of x otherwise:

Vectors[edit]

The pseudoinverse of the null (all zero) vector is the transposed null vector. The pseudoinverse of a non-null vector is the conjugate transposed vector divided by its squared magnitude:

Linearly independent columns[edit]

If the columns of are linearly independent (so that ), then is invertible. In this case, an explicit formula is:[1]

.

It follows that is then a left inverse of :   .

Linearly independent rows[edit]

If the rows of are linearly independent (so that ), then is invertible. In this case, an explicit formula is:

.

It follows that is a right inverse of :   .

Orthonormal columns or rows[edit]

This is a special case of either full column rank or full row rank (treated above). If has orthonormal columns () or orthonormal rows (), then:

.

Orthogonal projection matrices[edit]

If is an orthogonal projection matrix, i.e. and , then the pseudoinverse trivially coincides with the matrix itself:

.

Circulant matrices[edit]

For a circulant matrix , the singular value decomposition is given by the Fourier transform, that is the singular values are the Fourier coefficients. Let be the Discrete Fourier Transform (DFT) matrix, then[14]

Construction[edit]

Rank decomposition[edit]

Let denote the rank of . Then can be (rank) decomposed as where and are of rank . Then .

The QR method[edit]

For or computing the product or and their inverses explicitly is often a source of numerical rounding errors and computational cost in practice. An alternative approach using the QR decomposition of may be used instead.

Consider the case when is of full column rank, so that . Then the Cholesky decomposition , where is an upper triangular matrix, may be used. Multiplication by the inverse is then done easily by solving a system with multiple right-hand sides,

which may be solved by forward substitution followed by back substitution.

The Cholesky decomposition may be computed without forming explicitly, by alternatively using the QR decomposition of , where has orthonormal columns, , and is upper triangular. Then

,

so R is the Cholesky factor of .

The case of full row rank is treated similarly by using the formula and using a similar argument, swapping the roles of and .

Singular value decomposition (SVD)[edit]

A computationally simple and accurate way to compute the pseudoinverse is by using the singular value decomposition.[1][9][15] If is the singular value decomposition of A, then . For a rectangular diagonal matrix such as , we get the pseudoinverse by taking the reciprocal of each non-zero element on the diagonal, leaving the zeros in place, and then transposing the matrix. In numerical computation, only elements larger than some small tolerance are taken to be nonzero, and the others are replaced by zeros. For example, in the MATLAB, GNU Octave, or NumPy function pinv, the tolerance is taken to be t = ε⋅max(m,n)⋅max(Σ), where ε is the machine epsilon.

The computational cost of this method is dominated by the cost of computing the SVD, which is several times higher than matrix–matrix multiplication, even if a state-of-the art implementation (such as that of LAPACK) is used.

The above procedure shows why taking the pseudoinverse is not a continuous operation: if the original matrix A has a singular value 0 (a diagonal entry of the matrix above), then modifying A slightly may turn this zero into a tiny positive number, thereby affecting the pseudoinverse dramatically as we now have to take the reciprocal of a tiny number.

Block matrices[edit]

Optimized approaches exist for calculating the pseudoinverse of block structured matrices.

The iterative method of Ben-Israel and Cohen[edit]

Another method for computing the pseudoinverse (cf. Drazin inverse) uses the recursion

which is sometimes referred to as hyper-power sequence. This recursion produces a sequence converging quadratically to the pseudoinverse of if it is started with an appropriate satisfying . The choice (where , with denoting the largest singular value of ) [16] has been argued not to be competitive to the method using the SVD mentioned above, because even for moderately ill-conditioned matrices it takes a long time before enters the region of quadratic convergence.[17] However, if started with already close to the Moore–Penrose inverse and , for example , convergence is fast (quadratic).

Updating the pseudoinverse[edit]

For the cases where A has full row or column rank, and the inverse of the correlation matrix ( for A with full row rank or for full column rank) is already known, the pseudoinverse for matrices related to can be computed by applying the Sherman–Morrison–Woodbury formula to update the inverse of the correlation matrix, which may need less work. In particular, if the related matrix differs from the original one by only a changed, added or deleted row or column, incremental algorithms[18][19] exist that exploit the relationship.

Similarly, it is possible to update the Cholesky factor when a row or column is added, without creating the inverse of the correlation matrix explicitly. However, updating the pseudoinverse in the general rank-deficient case is much more complicated.[20][21]

Software libraries[edit]

The Python package NumPy provides a pseudoinverse calculation through its functions matrix.I and linalg.pinv; its pinv uses the SVD-based algorithm. SciPy adds a function scipy.linalg.pinv that uses a least-squares solver. High quality implementations of SVD, QR, and back substitution are available in standard libraries, such as LAPACK. Writing one's own implementation of SVD is a major programming project that requires a significant numerical expertise. In special circumstances, such as parallel computing or embedded computing, however, alternative implementations by QR or even the use of an explicit inverse might be preferable, and custom implementations may be unavoidable.

The MASS package for R provides a calculation of the Moore–Penrose inverse through the ginv function.[22] The ginv function calculates a pseudoinverse using the singular value decomposition provided by the svd function in the base R package. An alternative is to employ the pinv function available in the pracma package.

The Octave programming language provides a pseudoinverse through the standard package function pinv as well as the pseudo_inverse() method.

Applications[edit]

Linear least-squares[edit]

The pseudoinverse provides a least squares solution to a system of linear equations.[23] For , given a system of linear equations

in general, a vector that solves the system may not exist, or if one does exist, it may not be unique. The pseudoinverse solves the "least-squares" problem as follows:

  • , we have where and denotes the Euclidean norm. This weak inequality holds with equality if and only if for any vector w; this provides an infinitude of minimizing solutions unless A has full column rank, in which case is a zero matrix.[24] The solution with minimum Euclidean norm is [24]

This result is easily extended to systems with multiple right-hand sides, when the Euclidean norm is replaced by the Frobenius norm. Let .

  • , we have where and denotes the Frobenius norm.

Obtaining all solutions of a linear system[edit]

If the linear system

has any solutions, they are all given by[25]

for arbitrary vector . Solution(s) exist if and only if .[25] If the latter holds, then the solution is unique if and only if A has full column rank, in which case is a zero matrix. If solutions exist but A does not have full column rank, then we have an indeterminate system, all of whose infinitude of solutions are given by this last equation.

Minimum norm solution to a linear system[edit]

For linear systems with non-unique solutions (such as under-determined systems), the pseudoinverse may be used to construct the solution of minimum Euclidean norm among all solutions.

  • If is satisfiable, the vector is a solution, and satisfies for all solutions.

This result is easily extended to systems with multiple right-hand sides, when the Euclidean norm is replaced by the Frobenius norm. Let .

  • If is satisfiable, the matrix is a solution, and satisfies for all solutions.

Condition number[edit]

Using the pseudoinverse and a matrix norm, one can define a condition number for any matrix:

A large condition number implies that the problem of finding least-squares solutions to the corresponding system of linear equations is ill-conditioned in the sense that small errors in the entries of A can lead to huge errors in the entries of the solution.[26]

Generalizations[edit]

In order to solve more general least-squares problems, one can define Moore–Penrose inverses for all continuous linear operators A : H1H2 between two Hilbert spaces H1 and H2, using the same four conditions as in our definition above. It turns out that not every continuous linear operator has a continuous linear pseudoinverse in this sense.[26] Those that do are precisely the ones whose range is closed in H2.

In abstract algebra, a Moore–Penrose inverse may be defined on a *-regular semigroup. This abstract definition coincides with the one in linear algebra.

See also[edit]

Notes[edit]

  1. ^ a b c Ben-Israel & Greville 2003.
  2. ^ Ben-Israel & Greville 2003, p. 7.
  3. ^ Campbell & Meyer, Jr. 1991, p. 10.
  4. ^ Nakamura 1991, p. 42.
  5. ^ Rao & Mitra 1971, p. 50–51.
  6. ^ Moore, E. H. (1920). "On the reciprocal of the general algebraic matrix". Bulletin of the American Mathematical Society. 26 (9): 394–95. doi:10.1090/S0002-9904-1920-03322-7.
  7. ^ Bjerhammar, Arne (1951). "Application of calculus of matrices to method of least squares; with special references to geodetic calculations". Trans. Roy. Inst. Tech. Stockholm. 49.
  8. ^ a b Penrose, Roger (1955). "A generalized inverse for matrices". Proceedings of the Cambridge Philosophical Society. 51 (3): 406–13. doi:10.1017/S0305004100030401.
  9. ^ a b c d e Golub, Gene H.; Charles F. Van Loan (1996). Matrix computations (3rd ed.). Baltimore: Johns Hopkins. pp. 257–258. ISBN 978-0-8018-5414-9.
  10. ^ a b c Stoer, Josef; Bulirsch, Roland (2002). Introduction to Numerical Analysis (3rd ed.). Berlin, New York: Springer-Verlag. ISBN 978-0-387-95452-3..
  11. ^ Maciejewski, Anthony A.; Klein, Charles A. (1985). "Obstacle Avoidance for Kinematically Redundant Manipulators in Dynamically Varying Environments". International Journal of Robotics Research. 4 (3): 109–117. doi:10.1177/027836498500400308.
  12. ^ Rakočević, Vladimir (1997). "On continuity of the Moore–Penrose and Drazin inverses" (PDF). Matematički Vesnik. 49: 163–72.
  13. ^ Golub, G. H.; Pereyra, V. (April 1973). "The Differentiation of Pseudo-Inverses and Nonlinear Least Squares Problems Whose Variables Separate". SIAM Journal on Numerical Analysis. 10 (2): 413–32. doi:10.1137/0710036. JSTOR 2156365.
  14. ^ Stallings, W. T.; Boullion, T. L. (1972). "The Pseudoinverse of an r-Circulant Matrix". Proceedings of the American Mathematical Society. 34 (2): 385–88. doi:10.2307/2038377. JSTOR 2038377.
  15. ^ Linear Systems & Pseudo-Inverse
  16. ^ Ben-Israel, Adi; Cohen, Dan (1966). "On Iterative Computation of Generalized Inverses and Associated Projections". SIAM Journal on Numerical Analysis. 3 (3): 410–19. doi:10.1137/0703035. JSTOR 2949637.pdf
  17. ^ Söderström, Torsten; Stewart, G. W. (1974). "On the Numerical Properties of an Iterative Method for Computing the Moore–Penrose Generalized Inverse". SIAM Journal on Numerical Analysis. 11 (1): 61–74. doi:10.1137/0711008. JSTOR 2156431.
  18. ^ Tino Gramß (1992). "Worterkennung mit einem künstlichen neuronalen Netzwerk". Georg-August-Universität zu Göttingen.
  19. ^ , Mohammad Emtiyaz, "Updating Inverse of a Matrix When a Column is Added/Removed"[1]
  20. ^ Meyer, Jr., Carl D. (1973). "Generalized inverses and ranks of block matrices". SIAM J. Appl. Math. 25 (4): 597–602. doi:10.1137/0125057.
  21. ^ Meyer, Jr., Carl D. (1973). "Generalized inversion of modified matrices". SIAM J. Appl. Math. 24 (3): 315–23. doi:10.1137/0124033.
  22. ^ "R: Generalized Inverse of a Matrix".
  23. ^ Penrose, Roger (1956). "On best approximate solution of linear matrix equations". Proceedings of the Cambridge Philosophical Society. 52: 17–19. doi:10.1017/S0305004100030929.
  24. ^ a b Planitz, M. (October 1979). "Inconsistent systems of linear equations". Mathematical Gazette. 63 (425): 181–85. doi:10.2307/3617890. JSTOR 3617890.
  25. ^ a b James, M. (June 1978). "The generalised inverse". Mathematical Gazette. 62 (420): 109–14. doi:10.1017/S0025557200086460.
  26. ^ a b Hagen, Roland; Roch, Steffen; Silbermann, Bernd (2001). "Section 2.1.2". C*-algebras and Numerical Analysis. CRC Press.

References[edit]

External links[edit]