Example. The split three-dimensional Spin action is symmetric matrix congruence [fcap-003X]
Example. The split three-dimensional Spin action is symmetric matrix congruence [fcap-003X]
For \(Q_{2,1}(v)=v_0^2+v_1^2-v_2^2\), keep the algebra map
\(\Phi :\mathcal {C}\kern -2pt\ell ^+(Q_{2,1})\simeq M_2(\mathbb R)\) of Example [fcap-002I].
Multiplication by the last generator gives symmetric matrix coordinates
\[X(v)=\Phi (ve_2)=
\begin {pmatrix}-v_2-v_1&v_0\\v_0&-v_2+v_1\end {pmatrix},
\qquad \det X(v)=-Q_{2,1}(v).\]
The inverse sends \(\begin {pmatrix}a&b\\b&d\end {pmatrix}\) to
\((b,(d-a)/2,-(a+d)/2)\). Thus \(X\) is a real linear equivalence onto
the symmetric matrices, with the stated negative determinant form
TauCeti.coe_realCliffordTwoOneVectorEquivSymmetric_apply
TauCeti.realCliffordTwoOneVectorEquivSymmetric_det.
For \(s\in \operatorname {Spin}(Q_{2,1})\) and \(M=\Phi (s)\in SL_2(\mathbb R)\),
the vector action becomes
\[X(svs^{-1})=M X(v)M^{\mathsf T}.\]
To prove the formula, write \(A=e_0e_1\), \(B=e_0e_2\), and
\(C=e_1e_2\). Since \(e_2^2=-1\), conjugation by \(e_2\) is
\(\gamma (x)=-e_2xe_2\). It fixes \(A\) and negates \(B,C\).
On their three matrices in Example [fcap-002I], and on the identity, this gives
\[\Phi (\gamma (x))=\operatorname {adj}(\Phi (x))^{\mathsf T}.\]
Both sides are linear in \(x\), so the basis check proves the equality
on the whole even algebra. Moreover,
\[(svs^{-1})e_2=s(ve_2)\gamma (s^{-1}).\]
The determinant-one equation gives
\(\operatorname {adj}(M^{-1})=M\); applying \(\Phi \) proves the action.
TauCeti's bilinear-generator formula sends \(A,B,C\) to exactly these
three matrices, so its group map and action use this same \(\Phi \)
TauCeti.realCliffordTwoOneEvenEquivMatrix_ι
TauCeti.realSpinTwoOneEquivSpecialLinear_action.
For example, let \(M=\operatorname {diag}(a,a^{-1})\) with \(a>0\), and choose its Spin preimage using Example [fcap-002I]. Put \(c=(a^2+a^{-2})/2\) and \(h=(a^2-a^{-2})/2\). The congruence action is \[v_0'=v_0,\qquad v_1'=cv_1+hv_2,\qquad v_2'=hv_1+cv_2.\] Here \(c^2-h^2=1\). The two null coordinates \(v_1+v_2\) and \(v_1-v_2\) are multiplied by \(a^2\) and \(a^{-2}\), respectively. This exhibits the action on both null directions in the chosen coordinates.
For comparison, the compact quaternion action of Example [fcap-003J]
has a complex matrix realization. Distinguish quaternion units
\(\mathbf i,\mathbf j,\mathbf k\) from the complex scalar \(i\). The map
\[\iota (a+b\mathbf i+c\mathbf j+d\mathbf k)=
\begin {pmatrix}a+ib&c+id\\-c+id&a-ib\end {pmatrix}\]
preserves multiplication and sends quaternion conjugation to conjugate
transpose. Its determinant is \(a^2+b^2+c^2+d^2\), so it identifies the
unit quaternions with \(SU(2)\). Indeed a special unitary matrix has the
form \(\begin {pmatrix}z&w\\-\bar w&\bar z\end {pmatrix}\) with
\(|z|^2+|w|^2=1\), which recovers the four real coordinates. These are
Gallier's axes [gallier2014clifford, Section 1.1, pp. 8--9], and
the exact embedding and group equivalence in TauCeti
Quaternion.toComplexMatrix
Quaternion.unitaryEquivSpecialUnitaryGroup.
Keep \(\eta _3(v)=-v_1\mathbf i+v_0\mathbf j-v_2\mathbf k\)
from Example [fcap-003J], and put \(K(v)=i\iota (\eta _3(v))\).
It is a traceless Hermitian matrix. Applying \(i\iota \) to the already
proved quaternion action gives
\[K(svs^{-1})=U K(v)U^*,\qquad U=\iota (\theta (s))\in SU(2),\]
where \(U^*\) is conjugate transpose. The composite Spin-to-\(SU(2)\)
isomorphism is available in TauCeti
TauCeti.realSpinThreeEquivSpecialUnitary.
Its action theorem uses the unscaled matrix \(\iota (\eta _3(v))\)
TauCeti.realSpinThreeEquivSpecialUnitary_action;
multiplication by the central scalar \(i\) gives the displayed \(K\) formula.
Thus neither quaternion nor vector axes change. The split formula uses
real symmetric matrices and \(SL_2(\mathbb R)\) instead. The
fixed-vector transport underlying these formulas is illustrated by
[porteous1995clifford, Chapter 16, Proposition 16.9, pp. 143--144].
For the indefinite vector-action image, retain the spinor-norm kernel
of Theorem [fcap-0023]; no component or covering-space assertion is needed.