矩阵分解学习详细介绍SVD
MATLAB以svd函数来执行svd分解法， 其语法为[S,V,D]=svd(A)。
Singular value decomposition
In linear algebra, the singular value decomposition (SVD) is a factorization of a real or complex matrix, with many useful applications in signal processing and statistics.
Formally, the singular value decomposition of an m×n real or complex matrix M is a factorization of the form
where U is an m×m real or complex unitary matrix, Σ is an m×n rectangular diagonal matrix with nonnegative real numbers on the diagonal, and V* (the conjugate transpose of V) is an n×n real or complex unitary matrix. The diagonal entries Σ_{i,i} of Σ are known as the singular values of M. The m columns of U and the n columns of V are called the left singular vectors and right singular vectors of M, respectively.
The singular value decomposition and the eigendecomposition are closely related. Namely:

 The left singular vectors of M are eigenvectors of
 The right singular vectors of M are eigenvectors of
 The nonzero singular values of M (found on the diagonal entries of Σ) are the square roots of the nonzero eigenvalues of both and
Applications which employ the SVD include computing the pseudoinverse, least squares fitting of data, matrix approximation, and determining the rank, range and null space of a matrix.
[edit] Statement of the theorem
Suppose M is an m×n matrix whose entries come from the field K, which is either the field of real numbers or the field of complex numbers. Then there exists a factorization of the form
where U is an m×m unitary matrix over K, the matrix Σ is an m×n diagonal matrix with nonnegative real numbers on the diagonal, and the n×n unitary matrix V* denotes the conjugate transpose of V. Such a factorization is called the singular value decomposition of M.
The diagonal entries of Σ are known as the singular values of M. A common convention is to list the singular values in descending order. In this case, the diagonal matrix Σ is uniquely determined by M (though the matrices U and V are not).
[edit] Intuitive interpretations
[edit] Rotation, scaling, rotation
In the special but common case in which M is just an m×m square matrix with positive determinant whose entries are plain real numbers, then U, V*, and Σ are m×m matrices of real numbers as well, Σ can be regarded as a scaling matrix, and U and V* can be viewed as rotation matrices.
If the above mentioned conditions are met, the expression can thus be intuitively interpreted as a composition (or sequence) of three geometrical transformations: a rotation, a scaling, and another rotation. For instance, the figure above explains how a shear matrix can be described as such a sequence.
[edit] Singular values as semiaxis of an ellipse or ellipsoid
As shown in the figure, the singular values can be interpreted as the semiaxes of an ellipse in 2D. This concept can be generalized to ndimensional Euclidean space, with the singular values of any n×n square matrix being viewed as the semiaxes of an ndimensional ellipsoid. See below for further details.
[edit] U and V are orthonormal bases
Since U and V* are unitary, the columns of each of them form a set of orthonormal vectors, which can be regarded as basis vectors. By the definition of unitary matrix, the same is true for their conjugate transposes U* and V. In short, U, U*, V, and V* are orthonormal bases.
[edit] Example
Consider the 4×5 matrix
A singular value decomposition of this matrix is given by
Notice contains zeros outside of the diagonal. One diagonal element is zero. Furthermore, because the matrices and are unitary, multiplying by their respective conjugate transposes yields identity matrices, as shown below. In this case, because and are real valued, they each are an orthogonal matrix.
and
It should also be noted that this particular singular value decomposition is not unique. Choosing such that
is also a valid singular value decomposition.
[edit] Singular values, singular vectors, and their relation to the SVD
A nonnegative real number σ is a singular value for M if and only if there exist unitlength vectors u in K^{m} and v in K^{n} such that
The vectors u and v are called leftsingular and rightsingular vectors for σ, respectively.
In any singular value decomposition
the diagonal entries of Σ are equal to the singular values of M. The columns of U and V are, respectively, left and rightsingular vectors for the corresponding singular values. Consequently, the above theorem implies that:
 An m × n matrix M has at least one and at most p = min(m,n) distinct singular values.
 It is always possible to find an orthogonal basis U for K^{m} consisting of leftsingular vectors of M.
 It is always possible to find an orthogonal basis V for K^{n} consisting of rightsingular vectors of M.
A singular value for which we can find two left (or right) singular vectors that are linearly dependent is called degenerate.
Nondegenerate singular values always have unique left and right singular vectors, up to multiplication by a unit phase factor e^{iφ} (for the real case up to sign). Consequently, if all singular values of M are nondegenerate and nonzero, then its singular value decomposition is unique, up to multiplication of a column of U by a unit phase factor and simultaneous multiplication of the corresponding column of V by the same unit phase factor.
Degenerate singular values, by definition, have nonunique singular vectors. Furthermore, if u_{1} and u_{2} are two leftsingular vectors which both correspond to the singular value σ, then any normalized linear combination of the two vectors is also a left singular vector corresponding to the singular value σ. The similar statement is true for right singular vectors. Consequently, if M has degenerate singular values, then its singular value decomposition is not unique.
[edit] Applications of the SVD
[edit] Pseudoinverse
The singular value decomposition can be used for computing the pseudoinverse of a matrix. Indeed, the pseudoinverse of the matrix M with singular value decomposition is
where Σ^{+} is the pseudoinverse of Σ, which is formed by replacing every nonzero diagonal entry by its reciprocal and transposing the resulting matrix. The pseudoinverse is one way to solve linear least squares problems.
[edit] Solving homogeneous linear equations
A set of homogeneous linear equations can be written as for a matrix and vector . A typical situation is that is known and a nonzero is to be determined which satisfies the equation. Such an belongs to 's null space and is sometimes called a (right) null vector of . can be characterized as a right singular vector corresponding to a singular value of that is zero. This observation means that if is a square matrix and has no vanishing singular value, the equation has no nonzero as a solution. It also means that if there are several vanishing singular values, any linear combination of the corresponding right singular vectors is a valid solution. Analogously to the definition of a (right) null vector, a nonzero satisfying , with denoting the conjugate transpose of , is called a left null vector of .
[edit] Total least squares minimization
A total least squares problem refers to determining the vector which minimizes the 2norm of a vector under the constraint . The solution turns out to be the right singular vector of corresponding to the smallest singular value.
[edit] Range, null space and rank
Another application of the SVD is that it provides an explicit representation of the range and null space of a matrix M. The right singular vectors corresponding to vanishing singular values of M span the null space of M. E.g., the null space is spanned by the last two columns of in the above example. The left singular vectors corresponding to the nonzero singular values of M span the range of M. As a consequence, the rank of M equals the number of nonzero singular values which is the same as the number of nonzero diagonal elements in .
In numerical linear algebra the singular values can be used to determine the effective rank of a matrix, as rounding error may lead to small but nonzero singular values in a rank deficient matrix.
[edit] Lowrank matrix approximation
Some practical applications need to solve the problem of approximating a matrix with another matrix which has a specific rank . In the case that the approximation is based on minimizing the Frobenius norm of the difference between and under the constraint that it turns out that the solution is given by the SVD of , namely
where is the same matrix as except that it contains only the largest singular values (the other singular values are replaced by zero). This is known as the Eckart–Young theorem, as it was proved by those two authors in 1936 (although it was later found to have been known to earlier authors; see Stewart 1993).
[edit] Separable models
The SVD can be thought of as decomposing a matrix into a weighted, ordered sum of separable matrices. By separable, we mean that a matrix can be written as an outer product of two vectors , or, in coordinates, . Specifically, the matrix M can be decomposed as:
Here and are the i^{th} columns of the corresponding SVD matrices, are the ordered singular values, and each is separable. The SVD can be used to find the decomposition of an image processing filter into separable horizontal and vertical filters. Note that the number of nonzero is exactly the rank of the matrix.
Separable models often arise in biological systems, and the SVD decomposition is useful to analyze such systems. For example, some visual area V1 simple cells receptive fields can be well described^{[1]} by a Gabor filter in the space domain multiplied by a modulation function in the time domain. Thus, given a linear filter evaluated through, for example, reverse correlation, one can rearrange the two spatial dimensions into one dimension, thus yielding a two dimensional filter (space, time) which can be decomposed through SVD. The first column of U in the SVD decomposition is then a Gabor while the first column of V represents the time modulation (or viceversa). One may then define an index of separability, , which is the fraction of the power in the matrix M which is accounted for by the first separable matrix in the decomposition.^{[2]}
[edit] Nearest orthogonal matrix
It is possible to use the SVD of to determine the orthogonal matrix closest to . The closeness of fit is measured by the Frobenius norm of . The solution is the product .^{[3]} This intuitively makes sense because an orthogonal matrix would have the decomposition where is the identity matrix, so that if then the product amounts to replacing the singular values with ones.
A similar problem, with interesting applications in shape analysis, is the orthogonal Procrustes problem, which consists of finding an orthogonal matrix which most closely maps to . Specifically,
where denotes the Frobenius norm.
This problem is equivalent to finding the nearest orthogonal matrix to a given matrix .
[edit] The Kabsch Algorithm
The Kabsch algorithm (called Wahba's problem in other fields) uses SVD to compute the optimal rotation (with respect to leastsquares minimization) that will align a set of points with a corresponding set of points. It is used, among other applications, to compare the structures of molecules.
[edit] Other examples
The SVD is also applied extensively to the study of linear inverse problems, and is useful in the analysis of regularization methods such as that of Tikhonov. It is widely used in statistics where it is related to principal component analysis and to Correspondence analysis, and in signal processing and pattern recognition. It is also used in outputonly modal analysis, where the nonscaled mode shapes can be determined from the singular vectors. Yet another usage is latent semantic indexing in natural language text processing.
The SVD also plays a crucial role in the field of Quantum information, in a form often referred to as the Schmidt decomposition. Through it, states of two quantum systems are naturally decomposed, providing a necessary and sufficient condition for them to be entangled : if the rank of the matrix is larger than one.
One application of SVD to rather large matrices is in numerical weather prediction, where Lanczos methods are used to estimate the most linearly quickly growing few perturbations to the central numerical weather prediction over a given initial forward time period — i.e. the singular vectors corresponding to the largest singular values of the linearized propagator for the global weather over that time interval. The output singular vectors in this case are entire weather systems. These perturbations are then run through the full nonlinear model to generate an ensemble forecast, giving a handle on some of the uncertainty that should be allowed for around the current central prediction.
Another application of SVD for daily life is that point in perspective view can be unprojected in a photo using the calculated SVD matrix, this application leads to measuring length (a.k.a. the distance of two unprojected points in perspective photo) by marking out the 4 corner points of knownsize object in a single photo. PRuler is a demo to implement this application by taking a photo of a regular credit card
[edit] Relation to eigenvalue decomposition
The singular value decomposition is very general in the sense that it can be applied to any m × n matrix whereas eigenvalue decomposition can only be applied to certain classes of square matrices. Nevertheless, the two decompositions are related.
Given an SVD of M, as described above, the following two relations hold:
The right hand sides of these relations describe the eigenvalue decompositions of the left hand sides. Consequently:

 The columns of V (right singular vectors) are eigenvectors of
 The columns of U (left singular vectors) are eigenvectors of
 The nonzero elements of Σ (nonzero singular values) are the square roots of the nonzero eigenvalues of or
In the special case that M is a normal matrix, which by definition must be square, the spectral theorem says that it can be unitarily diagonalized using a basis of eigenvectors, so that it can be written for a unitary matrix U and a diagonal matrix D. When M is also positive semidefinite, the decomposition is also a singular value decomposition.
However, the eigenvalue decomposition and the singular value decomposition differ for all other matrices M: the eigenvalue decomposition is where U is not necessarily unitary and D is not necessarily positive semidefinite, while the SVD is where Σ is a diagonal positive semidefinite, and U and V are unitary matrices that are not necessarily related except through the matrix M.
[edit] Existence
An eigenvalue λ of a matrix is characterized by the algebraic relation M u = λ u. When M is Hermitian, a variational characterization is also available. Let M be a real n × n symmetric matrix. Define f :R^{n} → R by f(x) = x^{T} M x. By the extreme value theorem, this continuous function attains a maximum at some u when restricted to the closed unit sphere {x ≤ 1}. By the Lagrange multipliers theorem, u necessarily satisfies
where the nabla symbol, , is the del operator.
A short calculation shows the above leads to M u = λ u (symmetry of M is needed here). Therefore λ is the largest eigenvalue of M. The same calculation performed on the orthogonal complement of u gives the next largest eigenvalue and so on. The complex Hermitian case is similar; there f(x) = x* M x is a realvalued function of 2n real variables.
Singular values are similar in that they can be described algebraically or from variational principles. Although, unlike the eigenvalue case, Hermiticity, or symmetry, of M is no longer required.
This section gives these two arguments for existence of singular value decomposition.
[edit] Based on the spectral theorem
Let M be an mbyn matrix with complex entries. M*M is positive semidefinite and Hermitian. By the spectral theorem, there exists a unitary nbyn matrix V such that
where D is diagonal and positive definite. Partition V appropriately so we can write
Therefore V_{1}*M*MV_{1} = D and V_{2}*M*MV_{2} = 0. The latter means MV_{2} = 0.
Also, since V is unitary, V_{1}*V_{1} = I, V_{2}*V_{2} = I and V_{1}V_{1}* + V_{2}V_{2}* = I.
Define
Then
We see that this is almost the desired result, except that U_{1} and V_{1} are not unitary in general, but merely isometries. To finish the argument, one simply has to "fill out" these matrices to obtain unitaries. For example, one can choose U_{2} such that
is unitary.
Define
where extra zero rows are added or removed to make the number of zero rows equal the number of columns of U_{2}. Then
which is the desired result:
Notice the argument could begin with diagonalizing MM* rather than M*M (This shows directly that MM* and M*M have the same nonzero eigenvalues).
[edit] Based on variational characterization
The singular values can also be characterized as the maxima of u^{T}Mv, considered as a function of u and v, over particular subspaces. The singular vectors are the values of u and v where these maxima are attained.
Let M denote an m × n matrix with real entries. Let and denote the sets of unit 2norm vectors in R^{m} and R^{n} respectively. Define the function
for vectors u ∈ and v ∈ . Consider the function σ restricted to × . Since both and are compact sets, their product is also compact. Furthermore, since σ is continuous, it attains a largest value for at least one pair of vectors u ∈ and v ∈ . This largest value is denoted σ_{1} and the corresponding vectors are denoted u_{1} and v_{1}. Since is the largest value of it must be nonnegative. If it were negative, changing the sign of either u_{1} or v_{1} would make it positive and therefore larger.
Statement: u_{1}, v_{1} are left and right singular vectors of M with corresponding singular value σ_{1}.
Proof: Similar to the eigenvalues case, by assumption the two vectors satisfy the Lagrange multiplier equation:
After some algebra, this becomes
and
Multiplying the first equation from left by and the second equation from left by and taking u = v = 1 into account gives
So σ_{1} = 2 λ_{1} = 2 λ_{2}. By properties of the functional φ defined by
we have
Similarly,
This proves the statement.
More singular vectors and singular values can be found by maximizing σ(u, v) over normalized u, v which are orthogonal to u_{1} and v_{1}, respectively.
The passage from real to complex is similar to the eigenvalue case.
[edit] Geometric meaning
Because U and V are unitary, we know that the columns u_{1},...,u_{m} of U yield an orthonormal basis of K^{m} and the columns v_{1},...,v_{n} of V yield an orthonormal basis of K^{n} (with respect to the standard scalar products on these spaces).
The linear transformation T :K^{n} → K^{m} that takes a vector x to Mx has a particularly simple description with respect to these orthonormal bases: we have T(v_{i}) = σ_{i} u_{i}, for i = 1,...,min(m,n), where σ_{i} is the ith diagonal entry of Σ, and T(v_{i}) = 0 for i > min(m,n).
The geometric content of the SVD theorem can thus be summarized as follows: for every linear map T :K^{n} → K^{m} one can find orthonormal bases of K^{n} and K^{m} such that T maps the ith basis vector of K^{n} to a nonnegative multiple of the ith basis vector of K^{m}, and sends the leftover basis vectors to zero. With respect to these bases, the map T is therefore represented by a diagonal matrix with nonnegative real diagonal entries.
To get a more visual flavour of singular values and SVD decomposition —at least when working on real vector spaces— consider the sphere S of radius one in R^{n}. The linear map T maps this sphere onto an ellipsoid in R^{m}. Nonzero singular values are simply the lengths of the semiaxes of this ellipsoid. Especially when n=m, and all the singular values are distinct and nonzero, the SVD of the linear map T can be easily analysed as a succession of three consecutive moves : consider the ellipsoid T(S) and specifically its axes ; then consider the directions in R^{n} sent by T onto these axes. These directions happen to be mutually orthogonal. Apply first an isometry v* sending these directions to the coordinate axes of R^{n}. On a second move, apply an endomorphism d diagonalized along the coordinate axes and stretching or shrinking in each direction, using the semiaxes lengths of T(S) as stretching coefficients. The composition d o v* then sends the unitsphere onto an ellipsoid isometric to T(S). To define the third and last move u, apply an isometry to this ellipsoid so as to carry it over T(S). As can be easily checked, the composition u o d o v* coincides with T.
[edit] Calculating the SVD
[edit] Numerical Approach
The SVD of a matrix M is typically computed by a twostep procedure. In the first step, the matrix is reduced to a bidiagonal matrix. This takes O(mn^{2}) floatingpoint operations, assuming that m ≥ n (this formulation uses the big O notation). The second step is to compute the SVD of the bidiagonal matrix. This step can only be done with an iterative method (as with eigenvalue algorithms). However, in practice it suffices to compute the SVD up to a certain precision, like the machine epsilon. If this precision is considered constant, then the second step takes O(n) iterations, each costing O(n) flops. Thus, the first step is more expensive, and the overall cost is O(mn^{2}) flops (Trefethen & Bau III 1997, Lecture 31).
The first step can be done using Householder reflections for a cost of 4mn^{2} − 4n^{3}/3 flops, assuming that only the singular values are needed and not the singular vectors. If m is much larger than n then it is advantageous to first reduce the matrix M to a triangular matrix with the QR decomposition and then use Householder reflections to further reduce the matrix to bidiagonal form; the combined cost is 2mn^{2} + 2n^{3} flops (Trefethen & Bau III 1997, Lecture 31).
The second step can be done by a variant of the QR algorithm for the computation of eigenvalues, which was first described by Golub & Kahan (1965). The LAPACK subroutine DBDSQR^{[4]} implements this iterative method, with some modifications to cover the case where the singular values are very small (Demmel & Kahan 1990). Together with a first step using Householder reflections and, if appropriate, QR decomposition, this forms the DGESVD^{[5]} routine for the computation of the singular value decomposition.
The same algorithm is implemented in the GNU Scientific Library (GSL). The GSL also offers an alternative method, which uses a onesided Jacobi orthogonalization in step 2 (GSL Team 2007). This method computes the SVD of the bidiagonal matrix by solving a sequence of 2by2 SVD problems, similar to how the Jacobi eigenvalue algorithm solves a sequence of 2by2 eigenvalue methods (Golub & Van Loan 1996, §8.6.3). Yet another method for step 2 uses the idea of divideandconquer eigenvalue algorithms (Trefethen & Bau III 1997, Lecture 31).
[edit] Analytic Result of 2by2 SVD
The singular values of a 2by2 matrix can be found analytically. Let the matrix be where are complex numbers that parameterize the matrix, is the identity matrix, and denote the Pauli matrices. Then its two singular values are given by
[edit] Reduced SVDs
In applications it is quite unusual for the full SVD, including a full unitary decomposition of the nullspace of the matrix, to be required. Instead, it is often sufficient (as well as faster, and more economical for storage) to compute a reduced version of the SVD. The following can be distinguished for an m×n matrix M of rank r:
[edit] Thin SVD
Only the n column vectors of U corresponding to the row vectors of V* are calculated. The remaining column vectors of U are not calculated. This is significantly quicker and more economical than the full SVD if n<<m. The matrix U_{n} is thus m×n, Σ_{n} is n×n diagonal, and V is n×n.
The first stage in the calculation of a thin SVD will usually be a QR decomposition of M, which can make for a significantly quicker calculation if n<<m.
[edit] Compact SVD
Only the r column vectors of U and r row vectors of V* corresponding to the nonzero singular values Σ_{r} are calculated. The remaining vectors of U and V* are not calculated. This is quicker and more economical than the thin SVD if r<<n. The matrix U_{r} is thus m×r, Σ_{r} is r×r diagonal, and V_{r}* is r×n.
[edit] Truncated SVD
Only the t column vectors of U and t row vectors of V* corresponding to the t largest singular values Σ_{t} are calculated. The rest of the matrix is discarded. This can be much quicker and more economical than the compact SVD if t<<r. The matrix U_{t} is thus m×t, Σ_{t} is t×t diagonal, and V_{t}* is t×n'.
Of course the truncated SVD is no longer an exact decomposition of the original matrix M, but as discussed below, the approximate matrix is in a very useful sense the closest approximation to M that can be achieved by a matrix of rank t.
[edit] Norms
[edit] Ky Fan norms
The sum of the k largest singular values of M is a matrix norm, the Ky Fan knorm of M.
The first of the Ky Fan norms, the Ky Fan 1norm is the same as the operator norm of M as a linear operator with respect to the Euclidean norms of K^{m} and K^{n}. In other words, the Ky Fan 1norm is the operator norm induced by the standard l^{2} Euclidean inner product. For this reason, it is also called the operator 2norm. One can easily verify the relationship between the Ky Fan 1norm and singular values. It is true in general, for a bounded operator M on (possibly infinite dimensional) Hilbert spaces
But, in the matrix case, M*M^{½} is a normal matrix, so M* M^{½} is the largest eigenvalue of M* M^{½}, i.e. the largest singular value of M.
The last of the Ky Fan norms, the sum of all singular values, is the trace norm (also known as the 'nuclear norm'), defined by M = Tr[(M*M)^{½}] (the eigenvalues of M* M are the squares of the singular values).
[edit] Hilbert–Schmidt norm
The singular values are related to another norm on the space of operators. Consider the Hilbert–Schmidt inner product on the n × n matrices, defined by . So the induced norm is . Since trace is invariant under unitary equivalence, this shows
where are the singular values of M. This is called the Frobenius norm, Schatten 2norm, or Hilbert–Schmidt norm of M. Direct calculation shows that if
the Frobenius norm of M coincides with
[edit] Tensor SVD
Unfortunately, the problem of finding a low rank approximation to a tensor is illposed. In other words, there doesn't exist a best possible solution, but instead a sequence of better and better approximations that converge to infinitely large matrices. But in spite of this, there are several ways of attempting this decomposition. There exist two types of tensor decompositions which generalise SVD to multiway arrays. One decomposition decomposes a tensor into a sum of rank1 tensors, see CandecompPARAFAC (CP) algorithm. The CP algorithm should not be confused with a rankR decomposition but, for a given N, it decomposes a tensor into a sum of N rank1 tensors that optimally fit the original tensor. The second type of decomposition computes the orthonormal subspaces associated with the different axes or modes of a tensor (orthonormal row space, column space, fiber space, etc.). This decomposition is referred to in the literature as the Tucker3/TuckerM, Mmode SVD, multilinear SVD and sometimes referred to as a higherorder SVD (HOSVD). In addition, multilinear principal component analysis in multilinear subspace learning involves the same mathematical operations as Tucker decomposition, being used in a different context of dimensionality reduction.
[edit] Bounded operators on Hilbert spaces
The factorization can be extended to a bounded operator M on a separable Hilbert space H. Namely, for any bounded operator M, there exist a partial isometry U, a unitary V, a measure space (X, μ), and a nonnegative measurable f such that
where is the multiplication by f on L^{2}(X, μ).
This can be shown by mimicking the linear algebraic argument for the matricial case above. VT_{f} V* is the unique positive square root of M*M, as given by the Borel functional calculus for self adjoint operators. The reason why U need not be unitary is because, unlike the finite dimensional case, given an isometry U_{1} with non trivial kernel, a suitable U_{2} may not be found such that
is a unitary operator.
As for matrices, the singular value factorization is equivalent to the polar decomposition for operators: we can simply write
and notice that U V* is still a partial isometry while VT_{f} V* is positive.
[edit] Singular values and compact operators
To extend notion of singular values and left/rightsingular vectors to the operator case, one needs to restrict to compact operators. It is a general fact that compact operators on Banach spaces have only discrete spectrum. This is also true for compact operators on Hilbert spaces, since Hilbert spaces are a special case of Banach spaces. If T is compact, every nonzero λ in its spectrum is an eigenvalue. Furthermore, a compact self adjoint operator can be diagonalized by its eigenvectors. If M is compact, so is M*M. Applying the diagonalization result, the unitary image of its positive square root T_{f} has a set of orthonormal eigenvectors {e_{i}} corresponding to strictly positive eigenvalues {σ_{i}}. For any ψ ∈ H,
where the series converges in the norm topology on H. Notice how this resembles the expression from the finite dimensional case. The σ_{i} 's are called the singular values of M. {U e_{i}} and {V e_{i}} can be considered the left and rightsingular vectors of M respectively.
Compact operators on a Hilbert space are the closure of finiterank operators in the uniform operator topology. The above series expression gives an explicit such representation. An immediate consequence of this is:
Theorem M is compact if and only if M*M is compact.
[edit] History
The singular value decomposition was originally developed by differential geometers, who wished to determine whether a real bilinear form could be made equal to another by independent orthogonal transformations of the two spaces it acts on. Eugenio Beltrami and Camille Jordan discovered independently, in 1873 and 1874 respectively, that the singular values of the bilinear forms, represented as a matrix, form a complete set of invariants for bilinear forms under orthogonal substitutions. James Joseph Sylvester also arrived at the singular value decomposition for real square matrices in 1889, apparently independent of both Beltrami and Jordan. Sylvester called the singular values the canonical multipliers of the matrix A. The fourth mathematician to discover the singular value decomposition independently is Autonne in 1915, who arrived at it via the polar decomposition. The first proof of the singular value decomposition for rectangular and complex matrices seems to be by Carl Eckart and Gale Young in 1936;^{[6]} they saw it as a generalization of the principal axis transformation for Hermitian matrices.
In 1907, Erhard Schmidt defined an analog of singular values for integral operators (which are compact, under some weak technical assumptions); it seems he was unaware of the parallel work on singular values of finite matrices. This theory was further developed by Émile Picard in 1910, who is the first to call the numbers singular values (or rather, valeurs singulières).
Practical methods for computing the SVD date back to Kogbetliantz in 1954, 1955 and Hestenes in 1958.^{[7]} resembling closely the Jacobi eigenvalue algorithm, which uses plane rotations or Givens rotations. However, these were replaced by the method of Gene Golub and William Kahan published in 1965,^{[8]} which uses Householder transformations or reflections. In 1970, Golub and Christian Reinsch^{[9]} published a variant of the Golub/Kahan algorithm that is still the one mostused today.
[edit] See also
 Canonical correlation analysis (CCA)
 Canonical form
 Correspondence analysis (CA)
 Curse of dimensionality
 Digital signal processing
 Dimension reduction
 Eigendecomposition
 Empirical orthogonal functions (EOFs)
 Fourier analysis
 Fourierrelated transforms
 Generalized singular value decomposition
 Latent semantic analysis
 Latent semantic indexing
 Linear least squares
 Locality sensitive hashing
 Matrix decomposition
 Multilinear principal component analysis (MPCA)
 Nearest neighbor search
 Nonlinear iterative partial least squares
 Polar decomposition
 Principal components analysis (PCA)
 Singular value
 Time series
 Twodimensional singular value decomposition(2DSVD)
 von Neumann's trace inequality
 Wavelet compression
 Lowrank approximation
[edit] Notes
 ^ DeAngelis GC, Ohzawa I, Freeman RD (October 1995). "Receptivefield dynamics in the central visual pathways". Trends Neurosci. 18 (10): 451–8. doi:10.1016/01662236(95)94496R. PMID 8545912. http://linkinghub.elsevier.com/retrieve/pii/01662236(95)94496R.
 ^ Depireux DA, Simon JZ, Klein DJ, Shamma SA (March 2001). "Spectrotemporal response field characterization with dynamic ripples in ferret primary auditory cortex". J. Neurophysiol. 85 (3): 1220–34. PMID 11247991. http://jn.physiology.org/cgi/pmidlookup?view=long&pmid=11247991.
 ^ The Singular Value Decomposition in Symmetric (Lowdin) Orthogonalization and Data Compression
 ^ Netlib.org
 ^ Netlib.org
 ^ Eckart, C.; Young, G. (1936). "The approximation of one matrix by another of lower rank". Psychometrika 1 (3): 211–8. doi:10.1007/BF02288367.
 ^ Hestenes, M. R. (1958). "Inversion of Matrices by Biorthogonalization and Related Results". Journal of the Society for Industrial and Applied Mathematics 6 (1): 51–90. doi:10.1137/0106005. JSTOR 2098862. MR0092215.
 ^ Golub, G. H.; Kahan, W. (1965). "Calculating the singular values and pseudoinverse of a matrix". Journal of the Society for Industrial and Applied Mathematics: Series B, Numerical Analysis 2 (2): 205–224. doi:10.1137/0702016. JSTOR 2949777. MR0183105.
 ^ Golub, G. H.; Reinsch, C. (1970). "Singular value decomposition and least squares solutions". Numerische Mathematik 14 (5): 403–420. doi:10.1007/BF02163027. MR1553974.
[edit] References
 Trefethen, Lloyd N.; Bau III, David (1997). Numerical linear algebra. Philadelphia: Society for Industrial and Applied Mathematics. ISBN 9780898713619.
 Demmel, James; Kahan, William (1990). "Accurate singular values of bidiagonal matrices". Society for Industrial and Applied Mathematics. Journal on Scientific and Statistical Computing 11 (5): 873–912. doi:10.1137/0911052.
 Golub, Gene H.; Kahan, William (1965). "Calculating the singular values and pseudoinverse of a matrix". Journal of the Society for Industrial and Applied Mathematics: Series B, Numerical Analysis 2 (2): 205–224. doi:10.1137/0702016. JSTOR 2949777.
 Golub, Gene H.; Van Loan, Charles F. (1996). Matrix Computations (3rd ed.). Johns Hopkins. ISBN 9780801854149.
 GSL Team (2007). "§13.4 Singular Value Decomposition". GNU Scientific Library. Reference Manual. http://www.gnu.org/software/gsl/manual/html_node/SingularValueDecomposition.html.
 Halldor, Bjornsson and Venegas, Silvia A. (1997). "A manual for EOF and SVD analyses of climate data". McGill University, CCGCR Report No. 971, Montréal, Québec, 52pp.
 Hansen, P. C. (1987). "The truncated SVD as a method for regularization". BIT 27: 534–553. doi:10.1007/BF01937276.
 Horn, Roger A.; Johnson, Charles R. (1985). "Section 7.3". Matrix Analysis. Cambridge University Press. ISBN 0521386322.
 Horn, Roger A.; Johnson, Charles R. (1991). "Chapter 3". Topics in Matrix Analysis. Cambridge University Press. ISBN 0521467136.
 Samet, H. (2006). Foundations of Multidimensional and Metric Data Structures. Morgan Kaufmann. ISBN 0123694469.
 Strang G. (1998). "Section 6.7". Introduction to Linear Algebra (3rd ed.). WellesleyCambridge Press. ISBN 0961408855.
 Stewart, G. W. (1993). "On the Early History of the Singular Value Decomposition". SIAM Review 35 (4): 551–566. doi:10.1137/1035134. http://citeseer.ist.psu.edu/stewart92early.html.
 Wall, Michael E., Andreas Rechtsteiner, Luis M. Rocha (2003). "Singular value decomposition and principal component analysis". In D.P. Berrar, W. Dubitzky, M. Granzow. A Practical Approach to Microarray Data Analysis. Norwell, MA: Kluwer. pp. 91–109. http://public.lanl.gov/mewall/kluwer2002.html.
 Press, WH; Teukolsky, SA; Vetterling, WT; Flannery, BP (2007), "Section 2.6", Numerical Recipes: The Art of Scientific Computing (3rd ed.), New York: Cambridge University Press, ISBN 9780521880688, http://apps.nrbook.com/empanel/index.html?pg=65