Orthogonal Procrustes problem explained

and

\Omega

which most closely maps

.^[1] Specifically, the orthogonal Procrustes problem is an optimization problem given by

\| ⋅ \|_F

denotes the Frobenius norm. This is a special case of Wahba's problem (with identical weights; instead of considering two matrices, in Wahba's problem the columns of the matrices are considered as individual vectors). Another difference is that Wahba's problem tries to find a proper rotation matrix instead of just an orthogonal one.

The name Procrustes refers to a bandit from Greek mythology who made his victims fit his bed by either stretching their limbs or cutting them off.

Solution

This problem was originally solved by Peter Schönemann in a 1964 thesis, and shortly after appeared in the journal Psychometrika.

This problem is equivalent to finding the nearest orthogonal matrix to a given matrix

M=BA^T

, i.e. solving the closest orthogonal approximation problem

min_R\|R-M\|_F subject to R^TR=I

. To find matrix

, one uses the singular value decomposition (for which the entries of

\Sigma

are non-negative)

M=U\SigmaV^T

to write

R=UV^T.

Proof of Solution

One proof depends on the basic properties of the Frobenius inner product that induces the Frobenius norm:

\begin{align} R&=\argmin_\Omega||\Omega

	2
A-B\\|
	F

\\ &=\argmin_\Omega\langle\OmegaA-B,\OmegaA-B\rangle_F\\ &=\argmin_\Omega\|\Omega

	2
A\\|
	F

	2
\\|B\\|
	F

-2\langle\OmegaA,B\rangle_F\\ &=\argmin_\Omega

	2
\\|A\\|
	F

	2
\\|B\\|
	F

-2\langle\OmegaA,B\rangle_F\\ &=\argmax_\Omega\langle\OmegaA,B\rangle_F\\ &=\argmax_\Omega\langle\Omega,BA^T\rangle_F\\ &=\argmax_\Omega\langle\Omega,U\SigmaV^T\rangle_F\\ &=\argmax_\Omega\langleU^T\OmegaV,\Sigma\rangle_F\\ &=\argmax_\Omega\langleS,\Sigma\rangle_F whereS=U^T\OmegaV\\ \end{align}

This quantity

is an orthogonal matrix (as it is a product of orthogonal matrices) and thus the expression is maximised when

equals the identity matrix

. Thus

\begin{align} I&=U^TRV\\ R&=UV^T\\ \end{align}

where

is the solution for the optimal value of

\Omega

that minimizes the norm squared

||\Omega

	2
A-B\\|
	F

Generalized/constrained Procrustes problems

There are a number of related problems to the classical orthogonal Procrustes problem. One might generalize it by seeking the closest matrix in which the columns are orthogonal, but not necessarily orthonormal.

Alternately, one might constrain it by only allowing rotation matrices (i.e. orthogonal matrices with determinant 1, also known as special orthogonal matrices). In this case, one can write (using the above decomposition

M=U\SigmaV^T

)

R=U\Sigma'V^T,

where

\Sigma'

is a modified

\Sigma

, with the smallest singular value replaced by

\det(UV^T)

(+1 or -1), and the other singular values replaced by 1, so that the determinant of R is guaranteed to be positive. For more information, see the Kabsch algorithm.

The unbalanced Procrustes problem concerns minimizing the norm of

AU-B

, where

A\inR^{m x},U\inR^\ell

, and

B\inR^{m x}

, with

m>\ell\gen

, or alternately with complex valued matrices. This is a problem over the Stiefel manifold

U\inU(m,\ell)

, and has no currently known closed form. To distinguish, the standard Procrustes problem (

A\inR^{m x}

) is referred to as the balanced problem in these contexts.

References

Book: Golub. G.H. . Van Loan . C. . 2013. Matrix Computations . JHU Press . 327 . 978-1421407944 . 4.

Orthogonal Procrustes problem explained

Solution

Proof of Solution

Generalized/constrained Procrustes problems

See also

References