Split-Biquaternion Anti-Hermitian Subspace
Introduction
The split biquaternion algebra $\mathbb{B}_{\mathrm{spl}}$ carries four linear involutions, and four of the resulting fixed spaces are its distinguished real subspaces. This article treats the anti-Hermitian subspace $\mathbb{B}_{\Phi}$, the fixed space of anti-Hermitian conjugation: its definition, its basis, the failure of closure under multiplication, the symmetrized product that carries it into $\mathbb{B}_{\Psi}$, the Lie structure it acquires under the commutator, the forms on it, its roots of $-1$, the action of the four involutions and its intersections with the other subspaces. Its complementary partner $\mathbb{B}_{\Psi}$ is treated in Split-Biquaternion Hermitian Subspace; the comparative tables are in Split-Biquaternion Relations Between Subspaces and Split-Biquaternion Involution Lattice.
The treatment is purely mathematical. No physics is invoked. The split biquaternion algebra is assumed from the basic algebra article and the quaternion algebra $\mathbb{H}$ from the article on quaternion algebra. Throughout, elements are written $\tilde{Q} = Q_t e_t + Q_x \vec e_x + Q_y \vec e_y + Q_z \vec e_z$ with $Q_\mu = q_\mu + j q'_\mu \in \mathbb{C}_{\mathrm{spl}}$, and the conjugations are ${}^{\natural}$ (quaternion), $\bar{\cdot}$ (split complex), ${}^{*} = \bar{\cdot}\circ{}^{\natural}$ (Hermitian) and ${}^{\flat} = -{}^{*}$ (anti-Hermitian). The split-biquaternion norm is $\langle\tilde{Q}\rangle_{\natural} = \tilde{Q}\tilde{Q}^{\natural} = \sum_\mu Q_\mu^2$, and the Hermitian norm is $\langle\tilde{Q}\rangle_{*} = \sum_\mu (q_\mu^2 - q'^2_\mu)$.
Definition and Basis
Definition. The anti-Hermitian subspace is the fixed space of anti-Hermitian conjugation,
$$ \mathbb{B}_{\Phi} = \left\{ \tilde{Q} \in \mathbb{B}_{\mathrm{spl}} : \tilde{Q}^{\flat} = \tilde{Q} \right\}, \qquad \tilde{Q}^\flat = -\tilde{Q}^{*} = -\overline{\tilde{Q}^{\natural}}. $$
The Condition in Coordinates
Comparing $-\overline{\tilde{Q}^{\natural}} = \tilde{Q}$ coefficient by coefficient:
- the coefficient of $e_t$ gives $-\bar{Q_t} = Q_t$, that is $\bar{Q_t} = -Q_t$, so $Q_t = j r_0$ is purely split-imaginary;
- the coefficient of $\vec e_k$ gives $\bar{Q_k} = Q_k$, so $Q_k = q_k$ is real.
The subspace is therefore the set of elements with purely split-imaginary scalar part and real vector part,
$$ \tilde{Q} = j r_0 e_t + q_x \vec e_x + q_y \vec e_y + q_z \vec e_z, \qquad r_0, q_x, q_y, q_z \in \mathbb{R}. $$
Basis and Dimension
Proposition. The anti-Hermitian subspace is a real vector space of dimension $4$, with basis $je_t, \vec e_x, \vec e_y, \vec e_z$.
Proof. The condition removes the four real scalar parameters and the four split-imaginary vector parameters, leaving the four real parameters $r_0, q_x, q_y, q_z$; the four basis elements are linearly independent and span the set.
Algebra Structure
It Is Not a Subalgebra
Proposition. $\mathbb{B}_{\Phi}$ is not closed under multiplication, and is therefore not a subalgebra of $\mathbb{B}_{\mathrm{spl}}$.
Proof. The element $je_t$ lies in the subspace, but its square is
$$ (je_t)^2 = j^2 = e_t, $$
which has real scalar part $1$ and so is not in $\mathbb{B}_{\Phi}$.
The obstruction is that a product of two anti-Hermitian elements has a real scalar part, which belongs to $\mathbb{B}_{\Psi}$; the product leaves $\mathbb{B}_{\Phi}$ in the Hermitian direction.
The Symmetrized Product Lands in the Hermitian Subspace
Proposition. For $\tilde{Q}_1 = j q'_{1t} e_t + \vec q_1$ and $\tilde{Q}_2 = j q'_{2t} e_t + \vec q_2$ in $\mathbb{B}_{\Phi}$, with $\vec q_1, \vec q_2$ real pure quaternions,
$$ \tilde{Q}_1 \bullet \tilde{Q}_2 = \tfrac{1}{2}(\tilde{Q}_1\tilde{Q}_2 + \tilde{Q}_2\tilde{Q}_1) = \left(q'_{1t} q'_{2t} - \vec q_1\cdot\vec q_2\right) e_t + j\left(q'_{1t} \vec q_2 + q'_{2t} \vec q_1\right) \in \mathbb{B}_{\Psi}. $$
Proof. Expanding and using $\vec q_1\vec q_2 = -\vec q_1\cdot\vec q_2 + \vec q_1\times\vec q_2$, the products $\tilde{Q}_1\tilde{Q}_2$ and $\tilde{Q}_2\tilde{Q}_1$ differ only in the sign of the cross term $\vec q_1\times\vec q_2$, which therefore cancels under symmetrization; the remaining scalar part is real and the remaining vector part is purely split-imaginary.
So the symmetrized product of two anti-Hermitian elements is Hermitian, the mirror image of the fact that the commutator of two Hermitian elements is anti-Hermitian.
It Is a Lie Subalgebra
Proposition. Under the commutator, $\mathbb{B}_{\Phi}$ is a Lie subalgebra of $\mathbb{B}_{\mathrm{spl}}$:
$$ \boxed{\; [\tilde{Q}_1, \tilde{Q}_2] = \tilde{Q}_1\tilde{Q}_2 - \tilde{Q}_2\tilde{Q}_1 = 2\,\vec q_1 \times \vec q_2 \in \mathbb{B}_{\Phi} \;}. $$
Proof. With the same expansion, the commutator retains only the cross term $2\vec q_1\times\vec q_2$, which is a pure real quaternion; this lies in $\mathbb{B}_{\Phi}$ because its scalar part is $0$ and its vector part is real. The Jacobi identity is inherited from the associativity of the product.
The bracket image $[\mathbb{B}_{\Phi}, \mathbb{B}_{\Phi}]$ is the three-dimensional space of pure real quaternions, the same three-dimensional Lie algebra $\mathrm{SO}(3)$ that is the commutator subalgebra of the quaternion subspace; the bracket of anti-Hermitian elements generates $\mathrm{SO}(3)$, developed in Split-Biquaternion Exponential and Lie Group Structure.
The Correspondence with the Hermitian Subspace
Proposition. Multiplication by $j$ is a linear isomorphism $\mathbb{B}_{\Phi} \to \mathbb{B}_{\Psi}$ with inverse multiplication by $j$:
$$ j\left(j r_0 e_t + \vec q\right) = r_0 e_t + j \vec q \in \mathbb{B}_{\Psi}, \qquad j\, \mathbb{B}_{\Phi} = \mathbb{B}_{\Psi}, \qquad j\, \mathbb{B}_{\Psi} = \mathbb{B}_{\Phi}. $$
Proof. Since $j$ is central with $j^2 = 1$, multiplying an element with purely split-imaginary scalar part and real vector part by $j$ exchanges the two parts, producing a real scalar part and a purely split-imaginary vector part, which is $\mathbb{B}_{\Psi}$; applying $j$ twice is the identity.
The map swaps the roles of the forms: the split-biquaternion norm on $\mathbb{B}_{\Phi}$ is the split-biquaternion norm on $\mathbb{B}_{\Psi}$ pulled back, and the Hermitian form changes sign. In the biquaternion algebra the corresponding map is multiplication by the central unit $i$.
The Square
For $\tilde{Q} = j r_0 e_t + \vec q$ the square is
$$ \tilde{Q}^2 = \left(r_0^2 - (\vec q, \vec q)\right) e_t + 2 j r_0 \vec q \in \mathbb{B}_{\Psi}, $$
the square of an anti-Hermitian element being Hermitian. In particular, for a real vector $\vec q$,
$$ \vec q^2 = -(\vec q,\vec q)\, e_t, $$
a negative real scalar. The square of $je_t$ is $e_t$, so $je_t$ itself is an involution of the algebra that is not the identity.
The Split-Biquaternion Norm and the Hermitian Form
The Split-Biquaternion Norm and Its Inertia
Theorem. On the anti-Hermitian subspace the split-biquaternion norm is the positive definite real quadratic form
$$ \boxed{\; \langle\tilde{Q}\rangle_{\natural} = r_0^2 + q_x^2 + q_y^2 + q_z^2, \qquad \tilde{Q} = j r_0 e_t + q_x \vec e_x + q_y \vec e_y + q_z \vec e_z \;}, $$
of inertia $(4,0)$.
Proof. Substituting $Q_t = j r_0$ and $Q_k = q_k$ in $\langle\tilde{Q}\rangle_{\natural} = \sum_\mu Q_\mu^2$ gives $r_0^2 + \sum_k q_k^2$, because $(j r_0)^2 = j^2 r_0^2 = +r_0^2$.
Here the split-biquaternion norm is definite exactly as on $\mathbb{B}_{\Psi}$; the difference from the Hermitian subspace appears in the Hermitian norm, not the split-biquaternion norm.
The Hermitian Norm and Its Isotropic Cone
The Hermitian norm $\langle\tilde Q\rangle_{*} = \sum_\mu (q_\mu^2 - q'^2_\mu)$, which on $\mathbb{B}_{\Phi}$ is
$$ \sum_{\mu\in\{t,x,y,z\}} (q_\mu^2 - q'^2_\mu) = -r_0^2 + q_x^2 + q_y^2 + q_z^2, $$
of inertia $(3,1)$. Its isotropic cone is the three-dimensional cone
$$ r_0^2 = q_x^2 + q_y^2 + q_z^2. $$
The inertias of the Hermitian form on the two sectors are complementary, $(1,3)$ on $\mathbb{B}_{\Psi}$ and $(3,1)$ on $\mathbb{B}_{\Phi}$; the map $j$ of the correspondence above exchanges the two.
Invertibles, Zero Divisors and Nilpotents
Theorem. For $\tilde{Q} \in \mathbb{B}_{\Phi}$ the following are equivalent: $\tilde{Q}$ is invertible; $\langle\tilde{Q}\rangle_{\natural} \neq 0$; $\tilde{Q} \neq 0$. Hence $\mathbb{B}_{\Phi}$ contains no zero divisor and no nilpotent.
Proof. The split-biquaternion norm is positive definite, vanishing only at the origin, so every nonzero element is invertible, with inverse $\tilde{Q}^{-1} = \tilde{Q}^{\natural}/\langle\tilde{Q}\rangle_{\natural}$. A nilpotent $\tilde{Q} \neq 0$ would require $\tilde{Q}^2 = 0$, that is $r_0^2 = |\vec q|^2$ and $r_0\vec q = 0$, forcing $\tilde{Q} = 0$.
As on $\mathbb{B}_{\Psi}$, the isotropic cone of the Hermitian form consists of invertibles, not zero divisors: the two sectors together contain no zero divisor at all, and the zero divisors of $\mathbb{B}_{\mathrm{spl}}$ are confined to the two four-dimensional subspaces $Z_+$ and $Z_-$ of Split-Biquaternion Zero Divisors.
The Roots of Minus One
Theorem. The roots of $-1$ in $\mathbb{B}_{\Phi}$ are the unit pure real quaternions,
$$ \boxed{\; \zeta = q_x \vec e_x + q_y \vec e_y + q_z \vec e_z, \qquad q_x^2 + q_y^2 + q_z^2 = 1 \;}, $$
which form a two-sphere $S^2$.
Proof. For $\tilde{Q} = j r_0 e_t + \vec q$ the equation $\tilde{Q}^2 = -1$ reads, by the square formula, $r_0^2 - |\vec q|^2 = -1$ and $2 r_0 \vec q = 0$. The second equation gives $r_0 = 0$ or $\vec q = 0$; $\vec q = 0$ would give $r_0^2 = -1$, impossible, so $r_0 = 0$ and $|\vec q|^2 = 1$.
The three imaginary units $\vec e_x, \vec e_y, \vec e_z$ lie in $\mathbb{B}_{\Phi}$, so the subspace contains the quaternion imaginary sphere; the roots of $-1$ in $\mathbb{B}_{\Phi}$ are exactly the roots of $-1$ in the quaternion subspace. The roots of $-1$ in $\mathbb{B}_{\Psi}$ are instead the elements $j\vec q$ with $\vec q$ a unit pure real quaternion, the sphere $jS^2$; the two spheres are exchanged by multiplication by $j$.
The Image in the Two Halves
Under the idempotent-decomposition isomorphism $\varphi : \mathbb{B}_{\mathrm{spl}} \to \mathbb{H} \oplus \mathbb{H}$, an anti-Hermitian element $\tilde{Q} = j r_0 e_t + \vec q$ has components
$$ \tilde{Q}_+ = r_0 + \vec q, \qquad \tilde{Q}_- = -r_0 + \vec q = -(\tilde{Q}_+)^{\natural}. $$
The image of $\mathbb{B}_{\Phi}$ is therefore the set of anti-conjugate pairs
$$ \varphi(\mathbb{B}_{\Phi}) = \left\{ (h, -h^{\natural}) : h \in \mathbb{H} \right\} \cong \mathbb{H}, $$
a real four-dimensional subspace of $\mathbb{H} \oplus \mathbb{H}$. Together with the image $(h, h^{\natural})$ of $\mathbb{B}_{\Psi}$ it exhibits the Hermitian decomposition $\mathbb{B}_{\mathrm{spl}} = \mathbb{B}_{\Psi} \oplus \mathbb{B}_{\Phi}$ as the decomposition of a pair $(h_1, h_2)$ into its conjugate-symmetric and conjugate-antisymmetric parts.
The Four Involutions on It
In the basis $je_t, \vec e_x, \vec e_y, \vec e_z$ the four involutions act diagonally:
| involution | matrix | effect |
|---|---|---|
| ${}^{\natural}$ | $\operatorname{diag}(1, -1, -1, -1)$ | negates the vector part |
| $\bar{\cdot}$ | $\operatorname{diag}(-1, 1, 1, 1)$ | negates the scalar part |
| ${}^{*}$ | $-\mathrm{id}$ | minus the identity |
| ${}^{\flat}$ | $+\mathrm{id}$ | the identity, by definition |
The subspace is invariant under all four. Quaternion conjugation negates the vector part and fixes the split-imaginary scalar $je_t$; split complex conjugation does the opposite; their composite Hermitian conjugation negates the whole subspace. Inside $\mathbb{B}_{\Phi}$ the fixed space of quaternion conjugation is the line $j\mathbb{R}$ of purely split-imaginary scalars, and the fixed space of split complex conjugation is the three-dimensional space $\mathrm{span}\{\vec e_x,\vec e_y,\vec e_z\}$ of real vectors.
Relations to the Other Subspaces
The intersections of $\mathbb{B}_{\Phi}$ with the other distinguished subspaces are, with dimensions:
| pair | intersection | $\dim_{\mathbb{R}}$ |
|---|---|---|
| $\mathbb{B}_{\Phi} \cap \mathbb{C}_{\mathrm{spl},\mathbb{B}_{\mathrm{spl}}}$ | $j\mathbb{R}$ | $1$ |
| $\mathbb{B}_{\Phi} \cap \mathbb{H}_{\mathbb{B}_{\mathrm{spl}}}$ | $\mathrm{span}\{\vec e_x, \vec e_y, \vec e_z\}$ | $3$ |
| $\mathbb{B}_{\Phi} \cap \mathbb{B}_{\Psi}$ | $\{0\}$ | $0$ |
| $\mathbb{B}_{\Phi} \cap \mathrm{Vect}(\mathbb{B}_{\mathrm{spl}})$ | $\mathrm{span}\{\vec e_x, \vec e_y, \vec e_z\}$ | $3$ |
The subspace is complementary to $\mathbb{B}_{\Psi}$, giving the Hermitian decomposition $\mathbb{B}_{\mathrm{spl}} = \mathbb{B}_{\Psi} \oplus \mathbb{B}_{\Phi}$. Its coordinate-block decomposition is $\mathbb{B}_{\Phi} = j\mathbb{R} \oplus \mathrm{span}\{\vec e_x, \vec e_y, \vec e_z\}$: the line $j\mathbb{R}$ is shared with the split complex subspace, and the real triple with the quaternion subspace and with the vector subspace. Its sums with the centre and with $\mathbb{C}_{\mathrm{spl},\mathbb{B}_{\mathrm{spl}}}$-complements have dimension $5$; only the pair with $\mathbb{B}_{\Psi}$ spans the algebra. The full tables are in Split-Biquaternion Relations Between Subspaces.
The Analysis on the Anti-Hermitian Subspace
The anti-Hermitian subspace is not a subalgebra, so, like the Hermitian subspace, it carries no intrinsic multiplicative function theory; the analysis on it is the restriction of the ambient analysis of Split-Biquaternion Analysis to a four-dimensional subspace. The symmetrized product of two elements of $\mathbb{B}_{\Phi}$ lies in $\mathbb{B}_{\Psi}$, so the functions on $\mathbb{B}_{\Phi}$ acquire a bracket, the commutator, rather than a pointwise product, and the intrinsic differential structure is the Lie-theoretic one of Split-Biquaternion Anti-Hermitian Subspace. The second-order operator obtained from the restriction of the ambient operators is the wave operator of the restricted Lorentzian form $-r_0^2 + \sum_k q_k^2$ of inertia $(3,1)$.
The Geometry of the Anti-Hermitian Subspace
The anti-Hermitian subspace is the negative eigenspace of the involution ${}^{*}$ of the algebra, and geometrically the invariant four-plane of the corresponding linear involution of $\mathbb{R}^8$. It is a Lorentzian subspace of the Hermitian scalar form: $g$ restricts to $-r_0^2 + \sum_k q_k^2$ of inertia $(3,1)$, while the split-biquaternion norm restricts to the positive definite form $r_0^2 + \sum_k q_k^2$ of inertia $(4,0)$ recorded in the notation table above. The null cone of $g$ is again cut out by $r_0^2 = \sum_k q_k^2$; the restricted form being indefinite of inertia $(3,1)$, it has isotropic lines, and their projectivisation in $\mathbb{P}(\mathbb{B}_{\Phi})$ is a two-sphere. The motions it carries are the same Lorentz group $SO(1,3)$, realised on $\mathbb{B}_{\Phi}$ rather than on $\mathbb{B}_{\Psi}$; the two subspaces are interchanged by multiplication by $j$, since $(j\tilde Q)^{*} = -j\tilde{Q}^{*}$.
Summary
The anti-Hermitian subspace $\mathbb{B}_{\Phi}$ is the fixed space of anti-Hermitian conjugation, the set of elements $\tilde{Q} = j r_0 e_t + q_x \vec e_x + q_y \vec e_y + q_z \vec e_z$ with purely split-imaginary scalar part and real vector part; it is a real vector space of dimension $4$ with basis $je_t, \vec e_x, \vec e_y, \vec e_z$. It is not a subalgebra — $(je_t)^2 = e_t$ leaves it — but the symmetrized product of two of its elements is Hermitian, the commutator is $2\,\vec q_1\times\vec q_2$, a pure real quaternion, and under that bracket $\mathbb{B}_{\Phi}$ is a Lie subalgebra with bracket image the three-dimensional $\mathrm{SO}(3)$. Multiplication by $j$ is a linear isomorphism $\mathbb{B}_{\Phi} \to \mathbb{B}_{\Psi}$, and the square of an anti-Hermitian element is Hermitian. The split-biquaternion norm restricts to the positive definite form $N = r_0^2 + q_x^2 + q_y^2 + q_z^2$ of inertia $(4,0)$, so $\mathbb{B}_{\Phi}$ contains no zero divisor and no nilpotent; the indefinite form is the Hermitian form, of inertia $(3,1)$ and isotropic cone $r_0^2 = q_x^2 + q_y^2 + q_z^2$, whose nonzero points are all units. The roots of $-1$ in the subspace are the unit pure real quaternions, a two-sphere $S^2$, the same roots as in the quaternion subspace. The image under $\varphi$ is the set of anti-conjugate pairs $(h, -h^{\natural})$ in $\mathbb{H} \oplus \mathbb{H}$. Of the four involutions, anti-Hermitian conjugation fixes the subspace pointwise, quaternion and split complex conjugations negate the vector and scalar parts respectively, and Hermitian conjugation is minus the identity; the subspace is complementary to $\mathbb{B}_{\Psi}$.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\mathbb{B}_{\mathrm{spl}} = \mathbb{C}_{\mathrm{spl}} \otimes_{\mathbb{R}} \mathbb{H}$ | Split biquaternion algebra, real dimension $8$ |
| $\tilde{Q} = j r_0 e_t + q_x \vec e_x + q_y \vec e_y + q_z \vec e_z$ | Element of the anti-Hermitian subspace, coefficients real |
| $\mathbb{B}_{\Phi}$ | Anti-Hermitian subspace, fixed space of ${}^{\flat}$ |
| $\mathbb{B}_{\Psi}$ | Hermitian subspace, fixed space of ${}^{*}$ |
| ${}^{\natural}, \bar{\cdot}, {}^{*}, {}^{\flat}$ | The four conjugations |
| $\langle\tilde{Q}\rangle_{\natural} = r_0^2 + \lvert\vec q\rvert^2$ | Norm on $\mathbb{B}_{\Phi}$, inertia $(4,0)$ |
| $\langle\tilde Q\rangle_{*} = \sum_\mu (q_\mu^2 - q'^2_\mu)$ | Hermitian norm, inertia $(3,1)$ on $\mathbb{B}_{\Phi}$ |
| $\tilde{Q}_1 \bullet \tilde{Q}_2 = \tfrac{1}{2}(\tilde{Q}_1\tilde{Q}_2 + \tilde{Q}_2\tilde{Q}_1)$ | Symmetrized product, landing in $\mathbb{B}_{\Psi}$ |
| $[\tilde{Q}_1, \tilde{Q}_2] = 2\,\vec q_1 \times \vec q_2$ | Commutator, a pure real quaternion |
| $j : \mathbb{B}_{\Phi} \to \mathbb{B}_{\Psi}$ | Isomorphism by multiplication by $j$ |
| $S^2$ | The roots of $-1$ in $\mathbb{B}_{\Phi}$, unit pure real quaternions |
| $r_0^2 = q_x^2 + q_y^2 + q_z^2$ | Isotropic cone of the Hermitian norm |
| $SO(1,3)$ | Lorentz group acting on the anti-Hermitian subspace |
| $(j\tilde{Q})^{*} = -j\tilde{Q}^{*}$ | Multiplication by $j$ swaps $\mathbb{B}_{\Psi}$ and $\mathbb{B}_{\Phi}$ |
Further Reading
- I. L. Kantor and A. S. Solodovnikov, Hypercomplex Numbers: An Elementary Introduction to Algebras (Springer, 1989), for anti-Hermitian elements and the Lie structure of algebras with a conjugation.
- J. P. Ward, Quaternions and Cayley Numbers: Algebra and Applications (Kluwer, 1997), for the Hermitian and anti-Hermitian sectors of the split biquaternion algebra.
- Pertti Lounesto, Clifford Algebras and Spinors (Cambridge, 2001), for the Lie algebra $\mathrm{SO}(3)$ generated by the pure imaginary units and the roots of $-1$.
- Nathan Jacobson, Lie Algebras (Dover, 1979), for the general theory of Lie subalgebras and the bracket relations of the classical algebras.