Split-Quaternion Relations Between Subspaces

Introduction

The four distinguished subspaces of $\mathbb{H}_{\mathrm{spl}}$ — the scalar line $S$, the vector subspace $V$ and the two split-complex planes $\mathbb{C}_{\mathrm{spl},2}, \mathbb{C}_{\mathrm{spl},3}$ — are defined by different conditions, but they are not independent: they contain one another, they intersect, they sum to subspaces of $\mathbb{H}_{\mathrm{spl}}$, and they are cut out of the four real coordinates by the involutions. This article collects those relations in one place, after the individual articles Split-Quaternion Scalar and Vector Subspaces and Split-Quaternion Split-Complex Subspaces, and before the structural article Split-Quaternion Involution Lattice, which treats the involutions themselves.

Every number below — dimensions of intersections, of sums, and the entries of the tables — is derived from the definitions by comparison of coefficients, so the article doubles as an index of the four subspaces and of the coordinate blocks.

The Three Decompositions

Two of the three involutions split $\mathbb{H}_{\mathrm{spl}}$ into distinguished subspaces, and the third splits it into two further subspaces:

involution fixed space anti-fixed space decomposition
${}^{\natural}$ $S = \mathbb{R}\cdot 1$ $V = \operatorname{span}\{\vec e_x,\vec e_y,\vec e_z\}$ $\mathbb{H}_{\mathrm{spl}} = S \oplus V$
$\alpha$ $\mathbb{C}_{\mathrm{spl},3} = \operatorname{span}\{1,\vec e_z\}$ $\operatorname{span}\{\vec e_x,\vec e_y\}$ $\mathbb{H}_{\mathrm{spl}} = \mathbb{C}_{\mathrm{spl},3} \oplus \operatorname{span}\{\vec e_x,\vec e_y\}$
$\rho$ $\operatorname{span}\{1,\vec e_x,\vec e_y\}$ $\mathbb{R} \vec e_z$ $\mathbb{H}_{\mathrm{spl}} = \operatorname{span}\{1,\vec e_x,\vec e_y\} \oplus \mathbb{R} \vec e_z$

The three decompositions are the scalar–vector decomposition, and the two decompositions associated with the principal involution and the reversal. Every element has three readings, one per row. The finest common refinement of the three is the coordinate-block decomposition of the next section.

The Four Subspaces at a Glance

subspace defining condition real basis $\dim_{\mathbb{R}}$ subalgebra norm
$S$ $\tilde{q}^{\natural} = \tilde q$ $1$ $1$ yes, $\cong \mathbb{R}$ $q_t^2$, positive definite
$V$ $\tilde{q}^{\natural} = -\tilde q$ $\vec e_x,\vec e_y,\vec e_z$ $3$ no (Lie) $q_x^2-q_y^2-q_z^2$, inertia $(2,1)$
$\mathbb{C}_{\mathrm{spl},2}$ closed under products of $1, \vec e_y$ $1,\vec e_y$ $2$ yes, $\cong \mathbb{C}_{\mathrm{spl}}$ $q_t^2-q_y^2$, inertia $(1,1)$
$\mathbb{C}_{\mathrm{spl},3}$ $\alpha(\tilde q) = \tilde q$ $1,\vec e_z$ $2$ yes, $\cong \mathbb{C}_{\mathrm{spl}}$ $q_t^2-q_z^2$, inertia $(1,1)$

Only three of the four are closed under multiplication: the scalar line and the two split-complex planes. The vector subspace is not a subalgebra but is closed under the commutator, and it is the only one of the four that carries a nonzero bracket — the other three are commutative. The norms are the restrictions of $\langle\cdot\rangle_{\natural}$; all four are non-degenerate, the first definite and the other three indefinite.

The Four Coordinate Blocks

The four real coordinates of a split-quaternion are $(q_t,q_x,q_y,q_z)$ in the basis $1, \vec e_x, \vec e_y, \vec e_z$. They group into four one-dimensional coordinate blocks,

$$ \mathbb{H}_{\mathrm{spl}} = \langle 1 \rangle \oplus \langle \vec e_x \rangle \oplus \langle \vec e_y \rangle \oplus \langle \vec e_z \rangle, $$

where $\langle \cdot\rangle$ denotes the real span of the displayed elements. Each of the blocks $\langle 1\rangle$, $\langle \vec e_y\rangle$ and $\langle \vec e_z\rangle$ is the intersection of the distinguished subspaces containing it, and $\langle \vec e_x\rangle$ is the complement of $\mathbb{C}_{\mathrm{spl},2} + \mathbb{C}_{\mathrm{spl},3}$ in $\mathbb{H}_{\mathrm{spl}}$; each of the four distinguished subspaces is a sum of blocks:

$$ S = \langle 1 \rangle, \qquad V = \langle \vec e_x \rangle \oplus \langle \vec e_y \rangle \oplus \langle \vec e_z \rangle, $$

$$ \mathbb{C}_{\mathrm{spl},2} = \langle 1 \rangle \oplus \langle \vec e_y \rangle, \qquad \mathbb{C}_{\mathrm{spl},3} = \langle 1 \rangle \oplus \langle \vec e_z \rangle . $$

The blocks are the building units of the article: every intersection below is a sum of blocks, which is why the dimension arithmetic is so simple.

The Intersections

The ten pairwise intersections of the four subspaces, with the dimensions (the diagonal carries the dimensions of the subspaces themselves):

$S$ $V$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{C}_{\mathrm{spl},3}$
$S$ $1$ $0$ $1$ $1$
$V$ $0$ $3$ $1$ $1$
$\mathbb{C}_{\mathrm{spl},2}$ $1$ $1$ $2$ $1$
$\mathbb{C}_{\mathrm{spl},3}$ $1$ $1$ $1$ $2$

Explicitly,

$$ \boxed{\; S \cap V = \{0\}, \quad S \cap \mathbb{C}_{\mathrm{spl},2} = S, \quad S \cap \mathbb{C}_{\mathrm{spl},3} = S, \quad V \cap \mathbb{C}_{\mathrm{spl},2} = \mathbb{R} \vec e_y \;}, $$

$$ \boxed{\; V \cap \mathbb{C}_{\mathrm{spl},3} = \mathbb{R} \vec e_z, \quad \mathbb{C}_{\mathrm{spl},2} \cap \mathbb{C}_{\mathrm{spl},3} = S \;}. $$

Three features are worth isolating:

  • The scalar line is contained in both split-complex planes, so it meets them in dimension $1$; it meets the vector subspace only in the origin, being the complement of $V$ in the scalar–vector decomposition.
  • The vector subspace meets each split-complex plane in a single line, spanned by the common basis element $\vec e_y$ or $\vec e_z$, and the two planes meet each other in the scalar line.
  • No two of the four subspaces are equal, and no pair except the obvious containment pairs is nested: $S \subset \mathbb{C}_{\mathrm{spl},2}$, $S \subset \mathbb{C}_{\mathrm{spl},3}$, and $S \subset S \oplus V$ but $S \not\subseteq V$, $\mathbb{C}_{\mathrm{spl},2} \not\subseteq \mathbb{C}_{\mathrm{spl},3}$.

The Sums

The sums of the pairs, with the same ordering:

$S$ $V$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{C}_{\mathrm{spl},3}$
$S$ $1$ $4$ $2$ $2$
$V$ $4$ $3$ $4$ $4$
$\mathbb{C}_{\mathrm{spl},2}$ $2$ $4$ $2$ $3$
$\mathbb{C}_{\mathrm{spl},3}$ $2$ $4$ $3$ $2$

The dimension of a sum is the sum of the dimensions minus the dimension of the intersection. The pairs that span the whole algebra are $S + V = \mathbb{H}_{\mathrm{spl}}$, $V + \mathbb{C}_{\mathrm{spl},2} = \mathbb{H}_{\mathrm{spl}}$, and $V + \mathbb{C}_{\mathrm{spl},3} = \mathbb{H}_{\mathrm{spl}}$; the pair $\mathbb{C}_{\mathrm{spl},2} + \mathbb{C}_{\mathrm{spl},3} = \operatorname{span}\{1,\vec e_y,\vec e_z\}$ has dimension $3$ and misses the line $\mathbb{R} \vec e_x$; and $S + \mathbb{C}_{\mathrm{spl},k} = \mathbb{C}_{\mathrm{spl},k}$ because $S$ lies inside each split-complex plane. The sum $V + \mathbb{C}_{\mathrm{spl},k}$ spans the algebra because $V$ contributes $\vec e_x$ and $\mathbb{C}_{\mathrm{spl},k}$ contributes $1$, and together with $\vec e_y$ or $\vec e_z$ they exhaust the four coordinates.

The Involutions as Sign Patterns

Each involution preserves each of the four subspaces, and its restriction to a subspace is diagonal in the displayed bases with signs $\pm 1$. The following table records the multiplicity of the sign $-1$ on each subspace, in the basis of the table above:

involution $S$ $V$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{C}_{\mathrm{spl},3}$
${}^{\natural}$ $0$ of $1$ $3$ of $3$ $1$ of $2$ $1$ of $2$
$\alpha$ $0$ of $1$ $2$ of $3$ $1$ of $2$ $0$ of $2$
$\rho$ $0$ of $1$ $1$ of $3$ $0$ of $2$ $1$ of $2$

The vanishing rows are the definitions: ${}^{\natural}$ acts as the identity exactly on $S$, $\alpha$ exactly on $\mathbb{C}_{\mathrm{spl},3}$, and $\rho$ exactly on $\operatorname{span}\{1,\vec e_x,\vec e_y\} \supset \mathbb{C}_{\mathrm{spl},2}$. The full multiplicities give the anti-fixed spaces: ${}^{\natural}$ negates all of $V$, $\alpha$ negates the plane $\operatorname{span}\{\vec e_x,\vec e_y\}$, and $\rho$ negates the line $\mathbb{R} \vec e_z$. The remaining entries are mixed restrictions. The composition rule ${}^{\natural} = \alpha\rho$ is visible row by row: the coordinates negated by ${}^{\natural}$ are exactly those negated by one of $\alpha, \rho$ but not both, so the multiplicities compose by symmetric difference.

How the Product and the Commutator Act

The Scalar Line

$S$ is the centre, so it acts on every subspace by scalar multiplication: $S \cdot V \subseteq V$, $S \cdot \mathbb{C}_{\mathrm{spl},k} \subseteq \mathbb{C}_{\mathrm{spl},k}$, and $[S, \tilde q] = 0$ for all $\tilde q$. As a product, $S \times S \to S$ is closed.

The Vector Subspace

The product of two vectors is not a vector: it splits as

$$ \boxed{\; \vec q\vec q_1 = -B(\vec q,\vec q_1)\cdot 1 + \tfrac{1}{2}[\vec q,\vec q_1] \;}, $$

with scalar part $-B(\vec q,\vec q_1) \in S$ and the remainder $\tfrac{1}{2}[\vec q,\vec q_1] \in V$; so $V \cdot V \subseteq S \oplus V = \mathbb{H}_{\mathrm{spl}}$, with equality. The commutator stays in $V$, and in the basis $\vec e_x, \vec e_y, \vec e_z$,

$$ \boxed{\; [\vec e_x,\vec e_y] = 2 \vec e_z, \qquad [\vec e_y,\vec e_z] = -2 \vec e_x, \qquad [\vec e_z,\vec e_x] = 2 \vec e_y \;}, $$

so $[V,V] \subseteq V$ and $V \cong \mathrm{SL}_2(\mathbb{R})$. The bracket is alternating and satisfies the Jacobi identity. It is the antisymmetrisation of the product on the vector subspace.

The Split-Complex Planes

Each split-complex plane is a commutative subalgebra, so $\mathbb{C}_{\mathrm{spl},k} \cdot \mathbb{C}_{\mathrm{spl},k} \subseteq \mathbb{C}_{\mathrm{spl},k}$ and $[\mathbb{C}_{\mathrm{spl},k}, \mathbb{C}_{\mathrm{spl},k}] = 0$. Across the planes the product leaves both: since $\vec e_y \vec e_z = -\vec e_x$ and $\vec e_z \vec e_y = \vec e_x$, one has $\mathbb{C}_{\mathrm{spl},2} \cdot \mathbb{C}_{\mathrm{spl},3} = \mathbb{H}_{\mathrm{spl}}$ and $\mathbb{C}_{\mathrm{spl},2} \cdot \mathbb{C}_{\mathrm{spl},3} \not\subseteq \mathbb{C}_{\mathrm{spl},2} \cup \mathbb{C}_{\mathrm{spl},3}$. The product of a scalar with a plane stays in the plane, and the two planes share only the scalar line.

The Operation Table

product $S$ $V$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{C}_{\mathrm{spl},3}$
$S \cdot$ $S$ $V$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{C}_{\mathrm{spl},3}$
$V \cdot$ $V$ $\mathbb{H}_{\mathrm{spl}}$ $\mathbb{H}_{\mathrm{spl}}$ $\mathbb{H}_{\mathrm{spl}}$
$\mathbb{C}_{\mathrm{spl},2} \cdot$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{H}_{\mathrm{spl}}$ $\mathbb{C}_{\mathrm{spl},2}$ $\mathbb{H}_{\mathrm{spl}}$
$\mathbb{C}_{\mathrm{spl},3} \cdot$ $\mathbb{C}_{\mathrm{spl},3}$ $\mathbb{H}_{\mathrm{spl}}$ $\mathbb{H}_{\mathrm{spl}}$ $\mathbb{C}_{\mathrm{spl},3}$

The commutator table is the antisymmetrisation of this one: it vanishes on the diagonal pairs among $S, \mathbb{C}_{\mathrm{spl},2}, \mathbb{C}_{\mathrm{spl},3}$ (all commutative), equals the bracket of $V$ restricted to the vector–vector entry, and is nonzero only where the product table has an off-diagonal entry that is not symmetric.

Worked Verifications

Example (an intersection of dimension one). Take $V \cap \mathbb{C}_{\mathrm{spl},2}$. An element is $q_x \vec e_x + q_y \vec e_y + q_z \vec e_z = q_t + q_y \vec e_y$; comparing coefficients gives $q_x = 0$, $q_z = 0$ and $q_t = 0$, so the intersection is $\mathbb{R} \vec e_y$, of dimension $1$, as tabulated.

Example (a pair that spans the algebra). The four elements $\vec e_x, \vec e_y, \vec e_z$ span $V$ and $1, \vec e_y$ span $\mathbb{C}_{\mathrm{spl},2}$; together they include $1, \vec e_x, \vec e_y, \vec e_z$, a basis, so $V + \mathbb{C}_{\mathrm{spl},2} = \mathbb{H}_{\mathrm{spl}}$, in agreement with $\dim V + \dim \mathbb{C}_{\mathrm{spl},2} - \dim(V \cap \mathbb{C}_{\mathrm{spl},2}) = 3 + 2 - 1 = 4$.

Example (a product that leaves its subspace). $\vec e_y \in \mathbb{C}_{\mathrm{spl},2}$ and $\vec e_z \in V$; their product is $\vec e_y \vec e_z = -\vec e_x \notin \mathbb{C}_{\mathrm{spl},2}$ and $\notin \mathbb{C}_{\mathrm{spl},3}$ (indeed $\vec e_x \in V$), consistent with the entry $\mathbb{H}_{\mathrm{spl}}$ in the product table.

Example (a mixed element and its blocks). For $\tilde q = 2 + 3\vec e_x - \vec e_y + 4\vec e_z$ the scalar block is $2$, the vector block is $3\vec e_x - \vec e_y + 4\vec e_z$, and the split-complex readings are $2 - \vec e_y + 4\vec e_z$ in $\mathbb{C}_{\mathrm{spl},2}$ and $2 + 3\vec e_x + 4\vec e_z$ modulo the complementary block; the involution images are $\tilde{q}^{\natural} = 2 - 3\vec e_x + \vec e_y - 4\vec e_z$, $\alpha(\tilde q) = 2 - 3\vec e_x + \vec e_y + 4\vec e_z$, $\rho(\tilde q) = 2 + 3\vec e_x - \vec e_y - 4\vec e_z$.

The Relation to the Corresponding Article for $\mathbb{B}$

The biquaternion article Subspaces and Decompositions performs the same task for the remarkable subspaces of $\mathbb{B}$, cut out by the four $\mathbb{C}$-linear and conjugate-linear involutions. The bookkeeping is analogous — decompositions, a glance table, coordinate blocks, tables of intersections, sums, sign patterns and operations — but the geometry of the tables differs, because $\mathbb{B}$ has eight real coordinates and remarkable subspaces, whereas $\mathbb{H}_{\mathrm{spl}}$ has four coordinates and four subspaces; in $\mathbb{B}$ the blocks have dimensions $1$ and $3$, while here all four blocks have dimension $1$, which is why the intersections here take only the values $0, 1, 2, 3$. The Biquaternions category is the structural model; nothing biquaternion-specific — the central unit $i$, the quaternionic half-algebra, the Hermitian and anti-Hermitian pairing — is imported here.

Summary

The four distinguished subspaces of $\mathbb{H}_{\mathrm{spl}}$ are the scalar line $S = \langle 1\rangle$, the vector subspace $V = \langle \vec e_x,\vec e_y,\vec e_z\rangle$, and the two split-complex planes $\mathbb{C}_{\mathrm{spl},2} = \langle 1,\vec e_y\rangle$ and $\mathbb{C}_{\mathrm{spl},3} = \langle 1,\vec e_z\rangle$. They are the eigenspaces of the involutions, with ${}^{\natural}$ giving $S \oplus V$, $\alpha$ giving $\mathbb{C}_{\mathrm{spl},3} \oplus \operatorname{span}\{\vec e_x,\vec e_y\}$, and $\rho$ giving $\operatorname{span}\{1,\vec e_x,\vec e_y\} \oplus \mathbb{R} \vec e_z$; the finest common decomposition is the coordinate-block decomposition $\mathbb{H}_{\mathrm{spl}} = \langle 1\rangle \oplus \langle \vec e_x\rangle \oplus \langle \vec e_y\rangle \oplus \langle \vec e_z\rangle$.

The intersections are $S \cap V = 0$, $V \cap \mathbb{C}_{\mathrm{spl},k} = \mathbb{R} \vec e_k$, and $\mathbb{C}_{\mathrm{spl},2} \cap \mathbb{C}_{\mathrm{spl},3} = S$, with $S$ inside both planes. The sums show that $V$ together with either split-complex plane spans the algebra, while the two planes together span only $\operatorname{span}\{1,\vec e_y,\vec e_z\}$. The involutions act on each subspace by the sign patterns tabulated above, and the composition rule ${}^{\natural} = \alpha\rho$ composes the multiplicities by symmetric difference. The product is closed on $S, \mathbb{C}_{\mathrm{spl},2}, \mathbb{C}_{\mathrm{spl},3}$, sends $V \cdot V$ into $S \oplus V$, and sends the products across the split-complex planes into the whole algebra; the commutator vanishes on the commutative pieces and makes $V$ the Lie algebra $\mathrm{SL}_2(\mathbb{R})$.

Summary of Notation

Symbol Meaning Article
$\mathbb{H}_{\mathrm{spl}}$ the split-quaternion algebra Split-Quaternion Algebra
$S = \langle 1\rangle$ the scalar line Split-Quaternion Scalar and Vector Subspaces
$V = \langle \vec e_x,\vec e_y,\vec e_z\rangle$ the vector subspace Split-Quaternion Scalar and Vector Subspaces
$\mathbb{C}_{\mathrm{spl},2} = \langle 1,\vec e_y\rangle$, $\mathbb{C}_{\mathrm{spl},3} = \langle 1,\vec e_z\rangle$ the split-complex planes Split-Quaternion Split-Complex Subspaces
$\langle \cdot \rangle$ real span of the displayed elements this article
${}^{\natural}$, $\alpha$, $\rho$ conjugation, principal involution, reversal Split-Quaternion Subspaces and the Involutions
$B$, $\langle\cdot\rangle_{\natural}$ the polarised form and the split-quaternion norm Split-Quaternion Norm and Invertibility
$[\tilde q_1,\tilde q_2] = \tilde q_1 \tilde q_2-\tilde q_2\tilde q_1$ the commutator Split-Quaternion Scalar and Vector Subspaces

Further Reading

  • Richard S. Pierce, Associative Algebras (Springer, 1982), for the systematic bookkeeping of subspaces and their intersections in an algebra with idempotents.
  • Ian R. Porteous, Clifford Algebras and the Classical Groups (Cambridge University Press, 1995), for the coordinate blocks and the involutions of a Clifford algebra.
  • Pertti Lounesto, Clifford Algebras and Spinors, 2nd ed. (Cambridge University Press, 2001), for the lattice of subspaces fixed by the involutions of the low-dimensional Clifford algebras.