SO Groups

The special orthogonal group of dimension , denoted , is the group of linear transformations of that preserve lengths and orientations; it is the rotation group in dimensions, formally defined as

Typically seen, is a 3-dimensional group and is fundamental in rigid-body mechanics, robotics, and computer vision.

As a group under matrix multiplication, satisfies:

  • Closure: if , then .
  • Associativity: .
  • Identity: .
  • Inverse: .

is also a Lie group: a smooth manifold with a smooth group operation. The tangent space at the identity is the Lie algebra , consisting of skew-symmetric matrices (). For , the Lie algebra is 3-dimensional and is connected to the group via the exponential map.

Intuition

The two defining constraints and encode geometry directly. Orthogonality means the columns of form an orthonormal basis— preserves lengths and angles. The determinant constraint excludes reflections (which have ), keeping only orientation-preserving transformations.

The group is a 3-dimensional manifold embedded in : although a rotation matrix has 9 entries, the 6 orthogonality constraints ( gives 6 independent equations) leave exactly 3 degrees of freedom. These can be parameterized by Euler angles, exponential coordinates, or quaternions, each with different trade-offs.

Topologically, (real projective 3-space), which means it cannot be smoothly covered by a single 3-parameter chart without singularities. This is the deep reason behind gimbal lock: any 3-parameter representation must have at least one singular configuration.

Properties

Useful Identities

(Inverse is Transpose) For , . This follows directly from the defining property and makes inversion computationally free.

(Column/Row Orthonormality) The columns of satisfy and form a right-handed orthonormal frame: (and cyclic permutations). The rows have the same property.

(Rotation Preserves Cross Products) For and : . This is equivalent to and is the reason rotations preserve orientation.

(Hat Conjugation) For and : . This identity is fundamental in velocity kinematics and Jacobian derivations—it describes how skew-symmetric matrices transform under rotation.

(Dimension of ) . For : . For : (a single angle).

: Planar Rotations

Every has the form . The group is abelian () and topologically a circle .

: Spatial Rotations

is non-abelian: the order of rotations matters. Its Lie algebra consists of skew-symmetric matrices, which are in bijection with via the hat map . The Lie bracket is the matrix commutator , which corresponds to the cross product in .

Examples

Verifying Group Properties

Let and .