[MATH-846] Rotation constructor does wrong calculation

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.0
• Fix Version/s: 3.1

Environment

Windows java 1.7r5

• Assignee: Unassigned
• Reporter: Frank Hess
• Created: 15/Aug/12 23:07
• Updated: 04/Mar/13 18:53
• Resolved: 16/Aug/12 16:11

Description

The following code produces the wrong result. The resulting Rotation is does not even hold a normalized quaternion:

final Vector3D u1 = new Vector3D(1.0, 0.0, 0.0);
final Vector3D u2 = new Vector3D(1.0, -1.0, 0.0);
final Vector3D v1 = new Vector3D(0.9999999, 0., 0.0);
final Vector3D v2 = new Vector3D(0., 1., 0.0);
final Rotation rot = new Rotation(u1, u2, v1, v2);
System.err.println("rot quaternion: " + rot.getQ0() + " " + rot.getQ1() + " " + rot.getQ2() + " " + rot.getQ3());

For me it outputs:

rot quaternion: 0.0 0.0 0.0 -7.450580596923828E-9

The correct output should have been:

rot quaternion: 0.0 1.0 0.0 0.0

The constructor seems to be hitting some kind of numerical instability.