Description
Either the default behavior of equiv_orientations in AngularSeparation must be updated to include the degenerate identity quaternion (-1,0,0,0) or the function behavior should be updated to not force the user to supply the degenerate negative quaternions.
Motivation and Context
I believe there is some unintentional behavior in freud in AngularSeparation. AngularSeparation takes in a parameter called equiv_orientations, which allows you to give quaternions that map a shape to itself to account for shape symmetry.
However, if you don't give any value, this parameter defaults to (1,0,0,0), According to the documentation, you are meant to include the degenerate negative quaternions (-q for each q). If you explicitly give equiv_orientations = [(1,0,0,0),(-1,0,0,0)], the result is different.
I'm not marking this as a bug, because the documentation correctly describes the behavior, but the default behavior in not physically relevant.
Description
Either the default behavior of equiv_orientations in AngularSeparation must be updated to include the degenerate identity quaternion (-1,0,0,0) or the function behavior should be updated to not force the user to supply the degenerate negative quaternions.
Motivation and Context
I believe there is some unintentional behavior in freud in AngularSeparation. AngularSeparation takes in a parameter called equiv_orientations, which allows you to give quaternions that map a shape to itself to account for shape symmetry.
However, if you don't give any value, this parameter defaults to (1,0,0,0), According to the documentation, you are meant to include the degenerate negative quaternions (-q for each q). If you explicitly give equiv_orientations = [(1,0,0,0),(-1,0,0,0)], the result is different.
I'm not marking this as a bug, because the documentation correctly describes the behavior, but the default behavior in not physically relevant.