Conformal rotation vector¶

Represent a three-dimensional orientation by stereographically projecting its unit quaternion to a three-component modified Rodrigues vector, with an explicit shadow-set switch around the chart singularity.

Version: v3 · 2026-09-19 · History
Domain-specific #: 1536
Domain group: Natural Sciences
Origin domain: Physics
Subdomains: Attitude Kinematics, Three Dimensional Rotation Parameterizations → Physics

Core Idea¶

For a unit quaternion \(q=(q_0,\mathbf q)\), one common conformal-rotation-vector convention uses the modified Rodrigues vector \(\boldsymbol\sigma=\mathbf q/(1+q_0)=\hat{\mathbf e}\tan(\phi/4)\), the stereographic image of (q) in the pure-imaginary quaternion hyperplane. Stereographic projection replaces the four quaternion components plus a unit-norm constraint with three rational coordinates; because the antipodal quaternions encode the same physical rotation, a shadow vector can replace a large principal vector before the chosen projection pole is reached.

How would you explain it like I'm…

No faithful explanation at this level. All three generators agree that any five-year-old picture collapses into 'three numbers that say how something is turned', the generic rotation triple the concept explicitly excludes; its identity is the stereographic quarter-angle chart on the double-covering quaternion sphere with shadow switching.

Flat Map of Every Turn

Engineers who steer spaceships and robots need a way to write down exactly how something is turned. One good method writes every possible turn as a point on a special ball that lives in more dimensions than we can picture, and every turn shows up twice on that ball, at opposite points. A Conformal rotation vector is like making a flat map of that ball, the way you can flatten a globe by shining a light from one pole onto a sheet. The map gives just three numbers for each turn. Close to the light's pole the map stretches out to infinity, so before you get there you switch to the turn's twin point on the other side, which describes the very same turn.

Stereographic Quarter-Angle Rotation Vector

A unit quaternion is a set of four numbers, with total length one, that encodes a 3D rotation; q and -q always encode the same rotation, so the unit-quaternion sphere covers the rotations twice. The conformal rotation vector, in one common convention called the modified Rodrigues vector, takes the three-number vector part of q and divides it by one plus the scalar part, giving sigma = q_vec/(1+q_0). Geometrically this is stereographic projection of the quaternion sphere onto a flat three-dimensional space, and its length equals the tangent of one quarter of the rotation angle, along the rotation axis. It swaps four constrained numbers for three unconstrained ones, with an exact rational formula to get back. The projection blows up near the projection pole, but because -q gives the same rotation, you can switch to the 'shadow' vector before reaching it. This differs from plain axis-angle vectors and from the classical Rodrigues (Gibbs) parameters, which use the tangent of half the angle.

For a unit quaternion q = (q_0, q), a common conformal-rotation-vector convention is the modified Rodrigues vector sigma = q/(1+q_0) = e tan(phi/4), where e is the rotation axis and phi the rotation angle. This is the stereographic projection of the unit 3-sphere of quaternions onto the pure-imaginary hyperplane, replacing four components plus a unit-norm constraint with three rational coordinates and admitting a rational inverse map. Because the quaternion sphere double-covers the rotation group, antipodal quaternions q and -q represent the same attitude; their images are a principal vector and a shadow vector, and switching to the shadow before the projection pole is approached manages the singularity. The object's identity is this specific chart with quarter-angle coordinates, antipodal equivalence, rational inverse, and shadow-set handling, not the generic use of three numbers for a rotation. It is distinct from axis-angle vectors, from Gibbs or classical Rodrigues parameters (half-angle tangent), from Euler angles, and from an unconstrained quaternion vector part. Verifying it requires fixing the quaternion ordering and the active or passive convention, checking unit norm, computing the forward and inverse maps, confirming both reproduce the same rotation matrix, locating the singularity, and confirming the shadow switch preserves attitude. Once established it supports minimal-coordinate attitude propagation, estimation and filtering, spacecraft guidance and control, robot orientation, and camera pose.

Scope of Application¶

Conformal rotation vector applies when the analyst can specify an orientation in the rotation group SO(3), represented first by one of the two unit quaternions on the three-sphere that encode it and then by a vector in a stereographic coordinate chart and establish that the three coordinates arise from a declared stereographic projection of a unit quaternion, reconstruct the same SO(3) attitude under the declared convention, and are managed with the corresponding singularity or shadow-set rule. The identity applies only after the quaternion, angle, frame, and projection conventions are fixed; formulas copied between scalar-first and scalar-last, active and passive, or opposite-pole conventions can represent the inverse or a different attitude.

Clarity¶

A clear claim names the carrier, governing rule, assumptions, and recognition test. This matters because modified Rodrigues parameters, conformal rotation vector, and Wiener–Milenkovic parameters are overlapping names, while rotation vector and Rodrigues parameters often denote two different coordinate maps. The disciplined statement is that the object counts as Conformal rotation vector exactly when the three coordinates arise from a declared stereographic projection of a unit quaternion, reconstruct the same SO(3) attitude under the declared convention, and are managed with the corresponding singularity or shadow-set rule

Manages Complexity¶

The abstraction compresses principal and shadow modified Rodrigues parameters, Wiener–Milenkovic terminology, attitude and attitude-error coordinates, continuous and sampled propagation, spacecraft, robotics and computer-vision conventions, and alternative projection-pole choices into a stable carrier, rule, invariant, and failure boundary. It makes comparison tractable while retaining the variables that control validity.

Abstract Reasoning¶

Type the carrier. Establish an orientation in the rotation group SO(3), represented first by one of the two unit quaternions on the three-sphere that encode it and then by a vector in a stereographic coordinate chart and reject examples from a different problem. 2.

Knowledge Transfer¶

Transfer within attitude kinematics is strong when new cases preserve the same carrier, mechanism, and diagnostic. The move from A spacecraft attitude estimator propagates modified Rodrigues parameters while their norm is moderate and replaces them by the shadow vector when the chosen chart approaches its 360-degree singular orientation to A vision pipeline encodes camera orientations with modified Rodrigues parameters so local optimization uses three variables and rational quaternion conversion rather than enforcing a four-component unit constraint at every iterate demonstrates that continuity.

Relationships to Other Abstractions¶

Current abstraction Conformal rotation vector Domain-specific

Parents (1) — outgoing typed relations

Conformal rotation vector is a kind of Representation Prime

The construction is literally a representation: a target orientation is mapped into a coordinate medium under explicit faithfulness, interpretation, and operational-use conventions.

Hierarchy path (1) — routes to 1 parentless root

Conformal rotation vector → is a kind of Representation → presupposes Abstraction

Neighborhood in Abstraction Space¶

Conformal rotation vector sits in a moderately populated region (55^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25