Wolfram Language & System Documentation Center

OrthogonalMatrixQ

gives True if m is an explicitly orthogonal matrix, and False otherwise.

Details and Options

A p×q matrix m is orthogonal if p≥q and Transpose[m].m is the q×q identity matrix, or p≤q and m.Transpose[m] is the p×p identity matrix.
OrthogonalMatrixQ works for symbolic as well as numerical matrices.
The following options can be given:

Normalized	True	test if matrix columns are normalized
SameTest	Automatic	function to test equality of expressions
Tolerance	Automatic	tolerance for approximate numbers

For exact and symbolic matrices, the option SameTest->f indicates that two entries a_ij and b_ij are taken to be equal if f[a_ij,b_ij] gives True.
For approximate matrices, the option Tolerance->t can be used to indicate that the norm γ=m.m^T-I_n_∞ satisfying γ≤t is taken to be zero where I_n is the identity matrix.

Examples

open all close all

Basic Examples (2)

Test if a 2×2 numeric matrix is orthogonal:

Test if a 3×3 symbolic matrix is orthogonal:

Verify the condition by hand:

Scope (14)

Basic Uses (6)

Test if a real matrix is orthogonal:

A real orthogonal matrix is also unitary:

Test if a complex matrix is orthogonal:

This matrix satisfies :

A complex-valued orthogonal matrix is not unitary:

Test if an exact matrix is orthogonal:

Make the matrix orthogonal:

Use OrthogonalMatrixQ with arbitrary-precision matrix:

A random matrix is typically not orthogonal:

Use OrthogonalMatrixQ with a symbolic matrix:

The matrix becomes orthogonal when and :

OrthogonalMatrixQ works efficiently with large numerical matrices:

Special Matrices (4)

Use OrthogonalMatrixQ with sparse matrices:

Use OrthogonalMatrixQ with structured matrices:

The identity matrix is orthogonal:

HilbertMatrix is not orthogonal:

Rectangular Semi-orthogonal Matrices (4)

Test if a rectangular matrix is semi-orthogonal:

As there are more columns than rows, this indicates that the rows are orthonormal:

The columns are not orthonormal:

Test a matrix with more rows than columns:

The columns of the matrix are orthonormal:

The rows are not orthonormal:

Generate a random orthogonal matrix:

Any subset of its rows forms a rectangular semi-orthogonal matrix:

As does any subset of its columns:

Rectangular identity matrices are semi-orthogonal:

Options (4)

Normalized (2)

Symbolic orthogonal matrix columns are often not normalized to 1:

Avoid testing if the columns are normalized:

Multiply the second column of an orthogonal matrix by 2:

OrthogonalMatrixQ with NormalizedFalse will still give True for m:

However, it will not give true for Transpose[m]:

This is because is a diagonal matrix, but is not:

SameTest (1)

This matrix is orthogonal for a positive real , but OrthogonalMatrixQ gives False:

Use the option SameTest to get the correct answer:

Tolerance (1)

Generate an orthogonal real-valued matrix with some random perturbation of order 10^-13:

q.q^ is not exactly zero outside the main diagonal:

Adjust the option Tolerance for accepting the matrix as orthogonal:

Tolerance is applied to the following value:

Applications (10)

Sources of Orthogonal Matrices (5)

Any orthonormal basis for $TemplateBox[{}, Reals]^n$ forms an orthogonal matrix:

The basis is orthonormal:

Putting the basis vectors in rows of a matrix forms an orthogonal matrix:

Putting them in columns also gives an orthogonal matrix:

Orthogonalize applied to real, linearly independent vectors generates an orthogonal matrix:

The matrix does not need to be square, in which case the resulting matrix is semi-orthogonal:

But the starting matrix must have full rank:

Any rotation matrix is orthogonal:

Any permutation matrix is orthogonal:

Matrices drawn from CircularRealMatrixDistribution are orthogonal:

Uses of Orthogonal Matrices (5)

Orthogonal matrices preserve the standard inner product on $TemplateBox[{}, Reals]^n$ . In other words, if is orthogonal and and are vectors, then :

This means the angles between the vectors are unchanged:

Since the norm is derived from the inner product, norms are preserved as well:

Any orthogonal matrix represents a rotation and/or reflection. If the matrix has determinant , it is a pure rotation. If it the determinant is , the matrix includes a reflection. Consider the following matrix:

It is orthogonal and has determinant :

Thus, it is a pure rotation; the Cartesian unit vectors and maintain their relative positions:

The following matrix is orthogonal but has determinant :

Thus, it includes a reflection; the Cartesian unit vectors and reverse their relative positions:

Orthogonal matrices play an important role in many matrix decompositions:

The matrix is always orthogonal for any nonzero real vector :

is called a Householder reflection; as a reflection, its determinant is :

It represents a reflection through a plane perpendicular to , sending to :

Any vector perpendicular to is unchanged by :

In matrix computations, is used to set to zero selected components of a given column vector :

Find the function satisfying the following differential equation:

Represent the cross-product with by means of multiplication by the antisymmetric matrix :

Compute the exponential and use it to define a solution to the equation:

Verify that satisfies the differential equation and initial condition:

The matrix is orthogonal for all values of :

Thus, the orbit of the solution is at a constant distance from the origin, in this case a circle:

Properties & Relations (14)

A matrix is orthogonal if m.Transpose[m]IdentityMatrix[n]:

For an approximate matrix, the identity is approximately true:

The inverse of an orthogonal matrix is its transpose:

Thus, the inverse and transpose are orthogonal matrices as well:

A real orthogonal matrix preserves the standard inner product of vectors in $TemplateBox[{}, Reals]^n$ :

As a consequence, real orthogonal matrices preserve norms as well:

Any real-valued orthogonal matrix is unitary:

But a complex unitary matrix is typically not orthogonal:

Products of orthogonal matrices are orthogonal:

A real-valued orthogonal matrix is normal:

A complex-valued orthogonal matrix need not be normal:

Real-valued orthogonal matrices have eigenvalues that lie on the unit circle:

Use Eigenvalues to find eigenvalues:

Verify they lie on the unit circle:

This does not apply to complex-valued orthogonal matrices:

Real orthogonal matrices have a complete set of eigenvectors:

As a consequence, they must be diagonalizable:

Use Eigenvectors to find eigenvectors:

A complex orthogonal matrix can fail to be diagonalizable:

The singular values are all 1 for a real orthogonal matrix:

This need not be true for a complex orthogonal matrix:

The determinant of an orthogonal matrix is 1 or :

The 2-norm of a real orthogonal matrix is always 1:

This need not be true for complex orthogonal matrices:

Integer powers of orthogonal matrices are orthogonal:

MatrixExp[m] for real antisymmetric m is both orthogonal and unitary:

For complex antisymmetric m, the exponential is orthogonal but not, in general, unitary:

OrthogonalMatrix can be used to explicitly construct orthogonal matrices:

These satisfy OrthogonalMatrixQ:

Possible Issues (1)

OrthogonalMatrixQ uses the definition $TemplateBox[{m}, Transpose].m=I_n$ for both real- and complex-valued matrices:

These complex matrices need not be normal or possess many properties of real orthogonal matrices:

UnitaryMatrixQ tests the more common definition $TemplateBox[{m}, ConjugateTranspose].m=I_n$ that ensures a complex matrix is normal:

Alternatively, test if the entries are real to restrict to real orthogonal matrices:

Neat Examples (1)

Rotation matrices are orthogonal:

Top

More Learning

Tech Support

Wolfram Solutions

Wolfram Solutions For Education

Get Started

Grow Your Skills

Work with Us

Educational Programs for Adults

Educational Programs for Youth

Read

OrthogonalMatrixQ

Details and Options

Examples

Basic Examples (2)