Algebraic realisation of three fermion generations with $S_{3}$ family and unbroken gauge symmetry from $\mathbb{C}\ell(8)$

Abstract

Building on previous work, we extend an algebraic realisation of three fermion generations within the complex Clifford algebra $\mathbb{C}\ell(8)$, by incorporating an unbroken $U(1)_{em}$ gauge symmetry. The algebra $\mathbb{C}\ell(8)$ is the multiplication algebra of the complexification of the Cayley-Dickson algebra of sedenions, $\mathbb{S}$. Previous work represented three generations of fermions with $SU(3)_{C}$ colour symmetry, permuted by an $S_{3}$ symmetry of order-three. However, it proved difficult to find a $U(1)$ generator that assigns the correct electric charge to all states. By generalising the embedding of the order-three $S_{3}$ automorphism of $\mathbb{S}$ into $\mathbb{C}\ell(8)$, we include an $S_{3}$-invariant $U(1)$ that correctly assigns electric charge. First generation states are represented in terms of two even $\mathbb{C}\ell(8)$ semi-spinors, obtained from two minimal left ideals, related to each other via the order-two $S_{3}$ symmetry. The remaining two generations are obtained by applying the $S_{3}$ symmetry of order-three to the first generation. In this model, the gauge symmetries, $SU(3)_{C}\times U(1)_{em}$, are $S_{3}$-invariant and preserve the semi-spinors. As a result of the generalised embedding of the $S_{3}$ automorphisms of $\mathbb{S}$ into $\mathbb{C}\ell(8)$, the three generations are linearly independent, in contrast to earlier work.

1 Introduction

In the search for a minimal mathematical framework yielding a derivation of the Standard Model (SM), the endeavour to link the four division algebras (and their left multiplication algebras) to particle physics has gained substantial traction in recent years, with numerous lecture series on the subject [FureyLectures1, FureyLectures2, OSMU2023, BoyleLectures] and a proliferation of research papers [Wilson2022, Furey2023One, Furey2023Two, Furey2023Three, Todorov2022, todorov2020superselection, gresnigt2020topological, gresnigt2021topological, gresnigt2018braids, Furey2022a, Furey2022, Gording2019, Krasnov2021, Krasnov2022, Lu2024, Masi2021, Stoica2017, Perelman2019, Vaibhav2021, furey2016standard, Furey2018, Furey2018a, singh2021quantum, finster2024causal]. In such approaches, the gauge groups, leptons, and quarks are contained within the multiplication algebra obtained from a composition of the four division algebras acting on itself. Although a composition of division algebras need not be associative (nor alternative), its multiplication algebra, obtained from the algebra acting on itself via its endomorphisms, is associative.

There are only four normed division algebras; $\mathbb{R}$, $\mathbb{C}$, $\mathbb{H}$ (quaternions) and $\mathbb{O}$ (octonions), of dimensions one, two, four and eight, respectively. Shortly after the discovery of quarks, Güynadin and Gürsey [Gunaydin1973] constructed a model of quark colour symmetry based on the algebra of the split octonions. This early association of $\mathbb{O}$ with quarks was expanded upon by Dixon [Dixon1990, Dixon2004, Dixon1994] who revealed that the mathematical characteristics of the SM, encompassing its gauge symmetries and corresponding multiplets to which a single generation of fermions is subject, are inherent in $\mathbb{T}^{2}$, where $\mathbb{T}=\mathbb{R}\otimes\mathbb{C}\otimes\mathbb{H}\otimes\mathbb{O}$, commonly referred to as the Dixon algebra [Dixon1990, Dixon1994, Dixon2010]. Here, $\mathbb{T}^{2}$ corresponds to a complexified (hyper) spinor in 1+9D spacetime.

A closely related approach by Furey encompasses both bosons and fermions within Clifford algebras, arising as the multiplication algebras of compositions of division algebras. Two minimal left ideals of $\mathbb{C}\ell(6)$, the left (or right) multiplication algebra of $\mathbb{C}\otimes\mathbb{O}$, transform as a single generation under $SU(3)_{C}\times U(1)_{em}$, whereas two $\mathbb{C}\ell(4)$ minimal ideals, the left and right multiplication algebra of $\mathbb{C}\otimes\mathbb{H}$, transform as a single generation of chiral fermions under weak $SU(2)$ [Furey2018b]. These findings integrate into a $\mathbb{C}\ell(10)$ model [Gresnigt2020, Furey2023Three, todorov2020superselection], whose $Spin(10)$ group generated from the bivectors of $\mathbb{C}\ell(10)$ can be systematically broken by requiring invariance under a series of division algebraic reflections, leading to a cascade of grand unified theories (GUTs) [Furey2023One, Furey2023Two, Furey2023Three].

The representation of spinors as minimal left ideals dates back to the 1930s [Juvet1930, Sauter1930] and 1940s [riesz2013clifford]. In Furey’s model, fermions are identified with the basis states of minimal left ideals in the multiplication algebra. This association is established through a standard procedure utilising a Witt decomposition, as proposed by Abłamowicz [Ablamowicz1995]. The gauge symmetries within this framework are characterised as those unitary symmetries that preserve these minimal left ideals (or equivalently, the Witt basis from which these minimal ideals are constructed).

Most division-algebraic based constructions to date are, however, limited to describing a single generation of fermions. Despite various attempts [Furey2018a, Furey2014, dixon2014, Gillard2019], an algebraic foundation for the existence of three generations within the division algebraic framework remains elusive. Likewise, the most popular GUTs, such as those based on $SU(5)$, $SO(10)$, and the Pati-Salam model are also inherently single-generation models, lacking a theoretical basis for three generations.

Furey endeavours to depict three generations directly from the algebra $\mathbb{C}\ell(6)$ [Furey2018a, Furey2014]. After defining two copies of $SU(3)$, the remaining 48 degrees of freedom are found to transform as three generations of colour states. The states are, however, no longer described in terms of minimal left ideals, as is the case for a single generation [furey2016standard]. Additionally, the $U(1)_{em}$ generator (corresponding to the number operator), which assigns the correct electric charge in the context of a one generation model, fails to work in this three generation model, although a modified construction allows for $U(1)_{em}$ to be included [Furey2018a]. A similar construction based on $\mathbb{C}\ell(6)$, which includes an $SU(2)$ gauge symmetry, is given by Gording and Schmidt-May [Gording2019]. Dixon, on the other hand, characterises three generations using the algebra $\mathbb{T}^{6}=\mathbb{C}\otimes\mathbb{H}^{2}\otimes\mathbb{O}^{3}$ [Dixon2004]. The choice of $\mathbb{T}^{6}$, over other $\mathbb{T}^{2n}$ options, may seem arbitrary but can be motivated from the Leech lattice. Other authors have sought to encode three generations within the exceptional Jordan algebra $J_{3}(\mathbb{O})$, comprising three-by-three matrices over $\mathbb{O}$ [dubois2016exceptional1, dubois2019exceptional2, todorov2018octonions, todorov2018deducing, boyle2020standard2, boyle2020standard]. Each octonion is associated with one generation through the three canonical $J_{2}(\mathbb{O})$ subalgebras of $J_{3}(\mathbb{O})$.

The three-generation model proposed here differs from the aforementioned ones. Central to our construction is the algebra $\mathbb{C}\ell(8)$, corresponding to the left multiplication algebra of the complex sedenions, $\mathbb{C}\otimes\mathbb{S}$. The sedenion algebra $\mathbb{S}$ is generated from $\mathbb{O}$ via the Cayley-Dickson process. The motivation to consider $\mathbb{S}$ as a basis for a three-generation model stems from the fact that its automorphism group is $Aut(\mathbb{S})=Aut(\mathbb{O})\times{S_{3}}$, where $Aut(\mathbb{O})=G_{2}$, indicating a threefold multiplicity of the symmetries associated with $\mathbb{O}$.

The initial model in [Gillard2019] used three $\mathbb{O}$ subalgebras of $\mathbb{S}$ to represent three generations, generalising the earlier three-generation lepton model using three $\mathbb{H}$ subalgebras of $\mathbb{O}$ [Manogue1998]. However, this model had two drawbacks: each generation required its own copy of $SU(3)_{C}$ (suggesting three generations of gluons), and the $S_{3}$ automorphisms of $\mathbb{S}$ lacked a clear physical interpretation.

The more recent model [Gresnigt2023] addressed these issues by utilising the order-three $S_{3}$ automorphism of $\mathbb{S}$ to generate three generations. Instead of associating three $\mathbb{O}$ subalgebras with three generations, these subalgebras were used to construct a single generation of colour states, with the $S_{3}$ automorphism of order-three generating two additional copies. This provides a clear physical interpretation for the order-three $S_{3}$ automorphism, and keeps $SU(3)_{C}$ invariant under $S_{3}$, thereby avoiding three generations of gluons.

However, the updated model did not include a $U(1)$ symmetry that assigns the correct electric charge to states Also, while the $S_{3}$ symmetry of order-three was given a clear physical interpretation, the order-two symmetry was not. Furthermore, in both models, the three generations were not linearly independent.

Here, we build on this previous model by including an $S_{3}$-invariant $U(1)$ symmetry that assigns the correct electric charge. This is achieved by generalising the embedding of the $S_{3}$ automorphisms of $\mathbb{S}$ into $\mathbb{C}\ell(8)$. We initially represent one generation of electrocolour states in terms of a single $\mathbb{C}\ell(8)$ minimal left ideal. To generalise our construction to three generations, the generalised $S_{3}$ symmetry of order-three is applied to the minimal left ideal. Being invariant under $S_{3}$, the $SU(3)$ symmetry transforms all states correctly. The $U(1)$ generator, however, fails to assign the correct electric charge because it is not invariant under $S_{3}$.

To include a $U(1)$ symmetry, we instead split each spinor into its even and odd-grade semi-spinors via a projector. Subsequently, we apply the order-two $S_{3}$ symmetry to the even semi-spinor, which results in a second even semi-spinor belonging to a different minimal left ideal. Applying the generalised order-three $S_{3}$ symmetry generates two additional pairs of semi-spinors. We then identify the unbroken gauge symmetries with those unitary symmetries that both preserve the semi-spinors, and are invariant under this $S_{3}$ action. The required $S_{3}$-invariant $U(1)$ symmetry then arises as the sum of three individual $U(1)$ symmetries, and the three pairs of semi-spinors transform as three generations of fermions under the SM unbroken gauge symmetry $SU(3)_{C}\times U(1)_{em}$. As an unexpected byproduct, all three generations are now linearly independent.

2 The algebra $\mathbb{C}\ell(8)$ as the multiplication algebra of $\mathbb{C}\otimes\mathbb{S}$

2.1 Normed division algebras

A division algebra $\mathbb{A}$ is an algebra over a field where division is always well-defined, except by zero. A normed division algebra is a division algebra equipped with a norm, defined in terms of a conjugate. Hurwitz [HurwitzUeberDC] showed that there are only four normed divisional algebras; $\mathbb{R}$, $\mathbb{C}$, $\mathbb{H}$ (quaternions) and $\mathbb{O}$ (octonions), of dimensions one, two, four and eight, respectively. Starting from the real numbers $\mathbb{R}$, the Cayley-Dickson (CD) construction produces a sequence of algebras of dimension $2^{n}$, the first three being the remaining three division algebras.

The octonions $\mathbb{O}$ are the largest normed division algebra with seven mutually anti-commuting imaginary units $u_{i}$ ($i={1,...,7}$), together with the identity $u_{0}$. A general octonion $x$ can be written as;

$$x=x_{0}u_{0}+x_{1}u_{1}+...+x_{7}u_{7},\hskip 10.00002ptx_{0},...,x_{7}\in\mathbb{R}.$$ (1)

The multiplication of two octonion elements is calculated as $u_{i}u_{j}=-\delta_{ij}+f_{ijk}u_{k}$δでるた start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT + italic_f start_POSTSUBSCRIPT italic_i italic_j italic_k end_POSTSUBSCRIPT italic_u start_POSTSUBSCRIPT italic_k end_POSTSUBSCRIPT for $i,j,k\in\{1,2,...,7\}$, where $f_{ijk}$ is an anti-symmetric tensor, $f_{ijk}=1$ when $ijk\in\{123,145,176,\allowbreak 246,257,347,365\}$, and zero otherwise. The standard involution of an octonion element $x$ is given by $\overline{x}=x_{o}u_{0}-x_{1}u_{1}-...-x_{7}u_{7}$. The norm $|x|$ is defined by $|x|^{2}=x\overline{x}=\overline{x}x$, and the inverse of $x$ is $x^{-1}={\overline{x}}/{|x|^{2}}$.

Elements $q_{i}$, $q_{j}$ and $q_{k}\in\mathbb{O}$ satisfying $q_{i}q_{j}=q_{k}$ correspond to a quaternion subalgebra of $\mathbb{O}$. There are seven such subalgebras, with elements as in $f_{ijk}$. Multiplication of octonion elements not belonging to the same quaternion subalgebra is non-associative. The automorphism group of $\mathbb{O}$ is the exceptional Lie group $G_{2}$. This exceptional group contains $SU(3)$ as one of its maximal subgroups, corresponding to the stabilizer subgroup of one of the octonion imaginary units. $SU(3)$ is thus the group that preserves a complex structure.

For a more detailed overview, see [Gresnigt2023, baez2002octonions].

2.2 Sedenions, and the left multiplication algebra of $\mathbb{C}\otimes\mathbb{S}$

Applying the CD construction to $\mathbb{O}$ generates the 16-dimensional algebra of sedenions $\mathbb{S}$. This algebra is non-commutative, non-associative, non-alternative, and contains zero divisors. An orthonormal basis consists of 15 imaginary units $s_{i}$ ($i={1,...,15}$), as well as the identity $s_{0}$. The imaginary units $s_{1},...,s_{7}$ correspond to the octonion imaginary units $u_{1},...,u_{7}$. A general sedenion $w$ may then be written as;

$$w=w_{0}s_{0}+w_{1}s_{1}+...+w_{15}s_{15},\hskip 10.00002ptw_{0},...,w_{15}\in\mathbb{R}.$$ (2)

The product of two sedenions can be determined using the multiplication table in Appendix A. The involution of a sedenion element $w$ is given by $\overline{w}=w_{0}s_{0}-w_{1}s_{1}-...-w_{15}s_{15}$. The norm $|w|$ is defined by $|w|^{2}=w\overline{w}=\overline{w}w$ and the inverse of $w$ (if it exists) is $w^{-1}={\overline{w}}/{|w|^{2}}$.

It is shown in [Brown1967] that the automorphism group of $\mathbb{S}$ is $Aut(\mathbb{S})=Aut(\mathbb{O})\times{S_{3}}$. These automorphisms can be explicitly stated as follows;

$$\phi:A+Bs_{8} \rightarrow\phi(A)+\phi(B)s_{8},$$ (3)

$$\epsilon:A+Bs_{8} \rightarrow A-Bs_{8},$$ (4)

$$\psi:A+Bs_{8} \rightarrow\frac{1}{4}[A+3A^{*}+\sqrt{3}(B-B^{*})]$$ψぷさい : italic_A + italic_B italic_s start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT

$$+\frac{1}{4}[B+3B^{*}-\sqrt{3}(A-A^{*})]s_{8},$$ (5)

where $A,B\in\mathbb{O}$. $A^{*}$ is an octonion involution of $A$ such that $(A^{*})^{*}=A$ and $(AB)^{*}=B^{*}A^{*}$, and $\phi$ is an element of $G_{2}$, the automorphism group of $\mathbb{O}$. These result in the following identities; $\epsilon^{2}=Id$, $\psi^{3}=Id$ψぷさい start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT = italic_I italic_d, $\psi\phi=\phi\psi$ψぷさい italic_ϕ = italic_ϕ italic_ψぷさい, $\epsilon\phi=\phi\epsilon$ and $\epsilon\psi=\psi^{2}\epsilon$ψぷさい = italic_ψぷさい start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT italic_ϵ. It follows that $\epsilon$ and $\psi$ψぷさい generate $S_{3}$.

Although $\mathbb{S}$ is non-associative and non-alternative, we can consider compositions of left or right actions of $\mathbb{S}$ on itself as linear operators, thereby generating an associative algebra. This process is almost identical to the derivation of $\mathbb{C}\ell(6)$ as the left (or right) multiplication algebra of the complex octonions, $\mathbb{C}\otimes\mathbb{O}$ [furey2016standard, Dixon1994, lohmus1994]. This paper will exclusively focus on the left associative multiplication algebra. Let $L_{a}$ denote the linear operator of left multiplication by $a\in\mathbb{C}\otimes\mathbb{S}$;

$$L_{a}[w]:=aw,\hskip 10.00002pt\forall a,w\in\mathbb{C}\otimes\mathbb{S}.$$ (6)

Since $L_{a}$ corresponds to linear operators, they can be represented as $16\times 16$ complex matrices (acting on the vector space $\mathbb{C}\otimes\mathbb{S}$ written as a $16\times 1$ column vector). Due to the non-associativity of $\mathbb{S}$, the left multiplication algebra of $\mathbb{C}\otimes\mathbb{S}$ contains new maps which are not captured by $\mathbb{C}\otimes\mathbb{S}$, because in general;

$$L_{a}L_{b}[w] =a(bw) \neq L_{ab}[w] =(ab)w,\quad a,b,w\in\mathbb{C}\otimes\mathbb{S}.$$ (7)

There are a total of 256 distinct left-acting complex-linear maps from $\mathbb{C}\otimes\mathbb{S}$ to itself, and these provide a faithful representation of $\mathbb{C}\ell(8)$. It can be shown that $L_{s_{i}}L_{s_{j}}[w]=-L_{s_{j}}L_{s_{i}}[w]$ and $L_{s_{i}}L_{s_{i}}[w]=-w$, $i,j=1,...,8$, $i\neq j$. As a result, the left multiplication actions $L_{s_{i}}$, $i=0,1,...,8$, are a generating basis for $\mathbb{C}\ell(8)$. From now on, the left actions will be denoted simply by $e_{i}:=L_{s_{i}}$ and assumed to be acting on arbitrary $w\in\mathbb{C}\otimes{\mathbb{S}}$. The maps $e_{k}$, $k=9,...,15$, can then be expressed in terms of these $e_{i}$, $i=1,...,8$. For example;

$${e_{9}}=-\frac{1}{2}{e_{1}e_{2}e_{3}e_{4}e_{5}e_{8}}+\frac{1}{2}{e_{1}e_{2}e_{3}e_{6}e_{7}e_{8}}+\frac{1}{2}{e_{1}e_{4}e_{5}e_{6}e_{7}e_{8}}-\frac{1}{2}{e_{1}e_{8}}.$$ (8)

2.3 The minimal left ideals of $\mathbb{C}\ell(8)$

A general construction for creating spinor spaces as minimal left ideals of Clifford algebras is given by Abłamowicz [Ablamowicz1995]. The process for $\mathbb{C}\ell(2n)$ is particularly simple. A Witt basis can be formed from the algebra’s canonical orthonormal basis vectors, which are then used to create primitive idempotents on which the Clifford algebra is left multiplied. For $\mathbb{C}\ell(8)$, we define the Witt basis;

$$a_{1} :=\frac{1}{2}(-e_{1}+ie_{5}),\hskip 20.00003pta_{1}^{\dagger} :=\frac{1}{2}(e_{1}+ie_{5}),$$ (9)

$$a_{2} :=\frac{1}{2}(-e_{2}+ie_{6}),\hskip 20.00003pta_{2}^{\dagger} :=\frac{1}{2}(e_{2}+ie_{6}),$$

$$a_{3} :=\frac{1}{2}(-e_{3}+ie_{7}),\hskip 20.00003pta_{3}^{\dagger} :=\frac{1}{2}(e_{3}+ie_{7}),$$

$$a_{4} :=\frac{1}{2}(-e_{4}+ie_{8}),\hskip 20.00003pta_{4}^{\dagger} :=\frac{1}{2}(e_{4}+ie_{8}).$$

These satisfy the fermionic anticommutation relations;

$$\{a_{i},a_{j}\}=\{a_{i}^{\dagger},a_{j}^{\dagger}\}=0,\hskip 10.00002pt\{a_{i},a_{j}^{\dagger}\}=\delta_{ij}.$$δでるた start_POSTSUBSCRIPT italic_i italic_j end_POSTSUBSCRIPT . (10)

The nilpotents ${\mathrm{<ruby><rb>\Omega </rb><rt>おめが</rt></ruby>}}_1=a_1{a}_2{a}_3{a}_4$Ωおめが1subscript𝑎1subscript𝑎2subscript𝑎3subscript𝑎4\Omega_{1}=a_{1}a_{2}a_{3}a_{4}roman_Ωおめが start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT and ${\mathrm{<ruby><rb>\Omega </rb><rt>おめが</rt></ruby>}}_1^{†}=a_4^{†}{a}_3^{†}{a}_2^{†}{a}_1^{†}$Ωおめが1†superscriptsubscript𝑎4†superscriptsubscript𝑎3†superscriptsubscript𝑎2†superscriptsubscript𝑎1†\Omega_{1}^{\dagger}=a_{4}^{\dagger}a_{3}^{\dagger}a_{2}^{\dagger}a_{1}^{\dagger}roman_Ωおめが start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT = italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT can be combined into a primitive idempotent; $v_1:={\mathrm{<ruby><rb>\Omega </rb><rt>おめが</rt></ruby>}}_1{\mathrm{<ruby><rb>\Omega </rb><rt>おめが</rt></ruby>}}_1^{†}$Ωおめが1superscriptsubscriptΩおめが1†v_{1}:=\Omega_{1}\Omega_{1}^{\dagger}italic_v start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT := roman_Ωおめが start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT roman_Ωおめが start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT, physically representing the vacuum state. A minimal left ideal is then;

$$\mathbb{C}\ell(8)v_{1}.$$ (11)

Explicitly, we can write the 16 complex-dimensional ideal as follows;

$$T_{1}=(r_{0}+r_{j}a_{j}^{\dagger}+r_{jk}a_{j}^{\dagger}{a_{k}^{\dagger}}+r_{jkl}a_{j}^{\dagger}{a_{k}^{\dagger}}a_{l}^{\dagger}+r_{1234}a_{1}^{\dagger}{a_{2}^{\dagger}}a_{3}^{\dagger}{a_{4}^{\dagger}})v_{1},$$ (12)

where $j,k,l\in\{1,2,3,4\}$, $j\neq k\neq l$ and $r_{0},r_{j},r_{jk},r_{jkl},r_{1234}$ are complex coefficients. Notably, $T_{1}$ can be expressed as $(\mathbb{C}\ell(6)\hskip 1.00006pt\oplus\hskip 1.00006pt\mathbb{C}\ell(6)a_{4}^{\dagger})v_{1}$, where $a_{k}$ and $a_{k}^{\dagger}$, $k=1,2,3$, constitutes a Witt basis for $\mathbb{C}\ell(6)$. This splitting can be obtained by applying the projectors $\eta_{\pm}=\frac{1}{2}(1\pm{ie_{4}e_{8}})$ηいーた start_POSTSUBSCRIPT ± end_POSTSUBSCRIPT = divide start_ARG 1 end_ARG start_ARG 2 end_ARG ( 1 ± italic_i italic_e start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT ) to $T_{1}$, with $\eta_{-}T_{1}=\mathbb{C}\ell(6)v_{1}$ηいーた start_POSTSUBSCRIPT - end_POSTSUBSCRIPT italic_T start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = blackboard_C roman_ℓ ( 6 ) italic_v start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and $\eta_{+}T_{1}=\mathbb{C}\ell(6)a_{4}^{\dagger}{v_{1}}$ηいーた start_POSTSUBSCRIPT + end_POSTSUBSCRIPT italic_T start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = blackboard_C roman_ℓ ( 6 ) italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_v start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT.

3 One fermion generation with unbroken gauge symmetries

The bi-vectors of $\mathbb{C}\ell(8)$ generate the Lie algebra for $Spin(8)$. However, not all of $Spin(8)$ preserves $T_{1}$, only an $SU(4)$ subgroup does. The generators for $SU(4)$ are listed below, written in terms of the Witt basis;

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_1$Λらむだ1\displaystyle\Lambda_{1}roman_Λらむだ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT $=-a_2^{†}{a}_1-a_1^{†}{a}_2,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_2$Λらむだ2\displaystyle=-a_{2}^{\dagger}{a_{1}}-a_{1}^{\dagger}{a_{2}},\hskip 10.00002pt% \Lambda_{2}= - italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT $=i{a}_2^{†}{a}_1-i{a}_1^{†}{a}_2,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_3=a_2^{†}{a}_2-a_1^{†}{a}_1,$Λらむだ3superscriptsubscript𝑎2†subscript𝑎2superscriptsubscript𝑎1†subscript𝑎1\displaystyle=ia_{2}^{\dagger}{a_{1}}-ia_{1}^{\dagger}{a_{2}},\hskip 10.00002% pt\Lambda_{3}=a_{2}^{\dagger}{a_{2}}-a_{1}^{\dagger}{a_{1}},= italic_i italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_i italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT = italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , (13)

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_4$Λらむだ4\displaystyle\Lambda_{4}roman_Λらむだ start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT $=-a_1^{†}{a}_3-a_3^{†}{a}_1,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_5$Λらむだ5\displaystyle=-a_{1}^{\dagger}{a_{3}}-a_{3}^{\dagger}{a_{1}},\hskip 10.00002pt% \Lambda_{5}= - italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 5 end_POSTSUBSCRIPT $=-i{a}_1^{†}{a}_3+i{a}_3^{†}{a}_1,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_6=-a_3^{†}{a}_2-a_2^{†}{a}_3,$Λらむだ6superscriptsubscript𝑎3†subscript𝑎2superscriptsubscript𝑎2†subscript𝑎3\displaystyle=-ia_{1}^{\dagger}{a_{3}}+ia_{3}^{\dagger}{a_{1}},\hskip 10.00002% pt\Lambda_{6}=-a_{3}^{\dagger}{a_{2}}-a_{2}^{\dagger}{a_{3}},= - italic_i italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT + italic_i italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 6 end_POSTSUBSCRIPT = - italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT ,

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_7$Λらむだ7\displaystyle\Lambda_{7}roman_Λらむだ start_POSTSUBSCRIPT 7 end_POSTSUBSCRIPT $=i{a}_3^{†}{a}_2-i{a}_2^{†}{a}_3,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_8$Λらむだ8\displaystyle=ia_{3}^{\dagger}{a_{2}}-ia_{2}^{\dagger}{a_{3}},\hskip 10.00002% pt\Lambda_{8}= italic_i italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_i italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT $\displaystyle=-\frac{1}{\sqrt{3}}(a_{1}^{\dagger}{a_{1}}+a_{2}^{\dagger}{a_{2}}-2a_{3}^{\dagger}{a_{3}}),$

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_9$Λらむだ9\displaystyle\Lambda_{9}roman_Λらむだ start_POSTSUBSCRIPT 9 end_POSTSUBSCRIPT $=-a_4^{†}{a}_1-a_1^{†}{a}_4,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{10}$Λらむだ10\displaystyle=-a_{4}^{\dagger}{a_{1}}-a_{1}^{\dagger}{a_{4}},\hskip 10.00002pt% \Lambda_{10}= - italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 10 end_POSTSUBSCRIPT $=i{a}_4^{†}{a}_1-i{a}_1^{†}{a}_4,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{11}=-a_4^{†}{a}_2-a_2^{†}{a}_4,$Λらむだ11superscriptsubscript𝑎4†subscript𝑎2superscriptsubscript𝑎2†subscript𝑎4\displaystyle=ia_{4}^{\dagger}{a_{1}}-ia_{1}^{\dagger}{a_{4}},\hskip 10.00002% pt\Lambda_{11}=-a_{4}^{\dagger}{a_{2}}-a_{2}^{\dagger}{a_{4}},= italic_i italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - italic_i italic_a start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 11 end_POSTSUBSCRIPT = - italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ,

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{12}$Λらむだ12\displaystyle\Lambda_{12}roman_Λらむだ start_POSTSUBSCRIPT 12 end_POSTSUBSCRIPT $=i{a}_4^{†}{a}_2-i{a}_2^{†}{a}_4,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{13}$Λらむだ13\displaystyle=ia_{4}^{\dagger}{a_{2}}-ia_{2}^{\dagger}{a_{4}},\hskip 10.00002% pt\Lambda_{13}= italic_i italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT - italic_i italic_a start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 13 end_POSTSUBSCRIPT $=-a_4^{†}{a}_3-a_3^{†}{a}_4,{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{14}=i{a}_4^{†}{a}_3-i{a}_3^{†}{a}_4,$Λらむだ14𝑖superscriptsubscript𝑎4†subscript𝑎3𝑖superscriptsubscript𝑎3†subscript𝑎4\displaystyle=-a_{4}^{\dagger}{a_{3}}-a_{3}^{\dagger}{a_{4}},\hskip 10.00002pt% \Lambda_{14}=ia_{4}^{\dagger}{a_{3}}-ia_{3}^{\dagger}{a_{4}},= - italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT - italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT , roman_Λらむだ start_POSTSUBSCRIPT 14 end_POSTSUBSCRIPT = italic_i italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT - italic_i italic_a start_POSTSUBSCRIPT 3 end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT italic_a start_POSTSUBSCRIPT 4 end_POSTSUBSCRIPT ,

${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{15}$Λらむだ15\displaystyle\Lambda_{15}roman_Λらむだ start_POSTSUBSCRIPT 15 end_POSTSUBSCRIPT $\displaystyle=-\frac{1}{\sqrt{6}}(a_{1}^{\dagger}{a_{1}}+a_{2}^{\dagger}{a_{2}}+$ $\displaystyle a_{3}^{\dagger}{a_{3}}-3a_{4}^{\dagger}{a_{4}}).$

The $SU(4)$ generators, $\{{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_1,\mathrm{\dots },{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{15}\}$Λらむだ1…subscriptΛらむだ15\{\Lambda_{1},...,\Lambda_{15}\}{ roman_Λらむだ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , … , roman_Λらむだ start_POSTSUBSCRIPT 15 end_POSTSUBSCRIPT }, transform the basis of $T_{1}$ as $1\oplus 4\oplus 6\oplus 4\oplus 1$. $SU(4)$ can be broken to $SU(3)\times{U(1)}$, where the $SU(3)$ generators are $\{{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_1,\mathrm{\dots },{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_8\}$Λらむだ1…subscriptΛらむだ8\{\Lambda_{1},...,\Lambda_{8}\}{ roman_Λらむだ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , … , roman_Λらむだ start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT }. These are the generators that preserve both $\mathbb{C}\ell(6)v_{1}$ and $\mathbb{C}\ell(6)a_{4}^{\dagger}{v_{1}}$ individually. Under the action of $SU(3)$, the basis elements of both $\mathbb{C}\ell(6)v_{1}$ and $\mathbb{C}\ell(6)a_{4}^{\dagger}{v_{1}}$ transform as $1\oplus\overline{3}\oplus 3\oplus{1}$, matching that of a single generation of fermions under $SU(3)_{C}$. The $U(1)$ generator ${\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_{15}$Λらむだ15\Lambda_{15}roman_Λらむだ start_POSTSUBSCRIPT 15 end_POSTSUBSCRIPT commutes with $\{{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_1,\mathrm{\dots },{\mathrm{<ruby><rb>\Lambda </rb><rt>らむだ</rt></ruby>}}_8\}$Λらむだ1…subscriptΛらむだ8\{\Lambda_{1},...,\Lambda_{8}\}{ roman_Λらむだ start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT , … , roman_Λらむだ start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT } and is proportional to the electric charge generator defined as;

$$Q_{1}:=\frac{1}{3}(a_{1}^{\dagger}{a_{1}}+a_{2}^{\dagger}{a_{2}}+a_{3}^{\dagger}{a_{3}}-3a_{4}^{\dagger}{a_{4}})\hskip 10.00002pt\in\mathfrak{su}(4),$$ (14)

assigning the correct electric charge to each state. Thus, one generation of fermions with unbroken $SU(3)_{C}\times U(1)_{em}$ gauge symmetry can be represented in terms of the minimal left ideal $T_{1}$. The $\mathbb{C}\ell(6)v_{1}$ terms represent the isospin-up states whereas the $\mathbb{C}\ell(6)a_{4}^{\dagger}{v_{1}}$ terms represent the isospin-down states. This construction is different to the construction in [furey2016standard], which uses two $\mathbb{C}\ell(6)$ ideals, generated from $\mathbb{C}\otimes\mathbb{O}$, with $U(1)_{em}$ arising from the $\mathbb{C}\ell(6)$ number operator ($N:=\frac{1}{3}(a_{1}^{\dagger}{a_{1}}+a_{2}^{\dagger}{a_{2}}+a_{3}^{\dagger}{a_{3}})$).

The $SU(3)$ generators above correspond to a subgroup of $Spin(6)$ generated from $\mathbb{C}\ell(6)$ bi-vectors. This same $SU(3)$ corresponds to a subgroup of the octonion automorphism group $G_{2}$, and in fact $SU(3)=G_{2}\cap Spin(6)$ [Dixon1994]. On the other hand, $Q_{1}$ defined above is not a subgroup of this $Spin(6)$ or $G_{2}$, and can only be defined within $Spin(8)$.

4 $S_{3}$ as a discrete generation symmetry

We now wish to extend the representation of a single generation of fermions, with unbroken gauge symmetry in terms of a single minimal left ideal of $\mathbb{C}\ell(8)$, to three generations. Recall that $\mathbb{C}\ell(8)$ arises as the left multiplication algebra of $\mathbb{C}\otimes\mathbb{S}$, and that $Aut(\mathbb{S})=G_{2}\times S_{3}$. Our approach is to extend the $S_{3}$ automorphisms of $\mathbb{C}\otimes\mathbb{S}$ to a symmetry of $\mathbb{C}\ell(8)$, and subsequently interpret this discrete group as a generation symmetry.

4.1 Extending the $S_{3}$ discrete symmetry to $\mathbb{C}\ell(8)$

It is clear to see that $\phi$ defined in eqn. (3) has a natural extension to $\mathbb{C}\ell(8)$: the $\phi$ automorphisms act on the left actions $e_{i}$ instead of sedenion elements $s_{i}$. The $\epsilon$ automorphism likewise has an obvious embedding into $\mathbb{C}\ell(8)$: $e_{i}\xrightarrow{\epsilon}{e_{i}}$ for $i\in\{0,1,...,7\}$ and $e_{8}\xrightarrow{\epsilon}{-e_{8}}$ within $\mathbb{C}\ell(8)$.

In our previous paper [Gresnigt2023], we defined the extension of $\psi$ψぷさい into $\mathbb{C}\ell(8)$ via the map;

$$e_{i}\xrightarrow{\psi_{old}}\begin{cases}-\frac{1}{2}e_{i}-\frac{\sqrt{3}}{2}e_{i+8}&i=\{1,...,7\},\\ -\frac{1}{2}e_{i}+\frac{\sqrt{3}}{2}e_{i-8}&i=\{9,...,15\},\\ e_{i}&i=\{0,8\}.\end{cases}$$ψぷさい start_POSTSUBSCRIPT italic_o italic_l italic_d end_POSTSUBSCRIPT end_OVERACCENT → end_ARROW { start_ROW start_CELL - divide start_ARG 1 end_ARG start_ARG 2 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT - divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 2 end_ARG italic_e start_POSTSUBSCRIPT italic_i + 8 end_POSTSUBSCRIPT end_CELL start_CELL italic_i = { 1 , … , 7 } , end_CELL end_ROW start_ROW start_CELL - divide start_ARG 1 end_ARG start_ARG 2 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT + divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 2 end_ARG italic_e start_POSTSUBSCRIPT italic_i - 8 end_POSTSUBSCRIPT end_CELL start_CELL italic_i = { 9 , … , 15 } , end_CELL end_ROW start_ROW start_CELL italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT end_CELL start_CELL italic_i = { 0 , 8 } . end_CELL end_ROW (15)

This definition does not allow for the construction of a $U(1)$ generator that assigns the correct electric charge to three generations of states. Furthermore, $e_{i}$, $\psi_{old}(e_{i})$ψぷさい start_POSTSUBSCRIPT italic_o italic_l italic_d end_POSTSUBSCRIPT ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) and $\psi_{old}^{2}(e_{i})$ψぷさい start_POSTSUBSCRIPT italic_o italic_l italic_d end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) are linearly dependent, satisfying;

$$e_{i}+\psi_{old}(e_{i})+\psi_{old}^{2}(e_{i})=0,$$ψぷさい start_POSTSUBSCRIPT italic_o italic_l italic_d end_POSTSUBSCRIPT ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) + italic_ψぷさい start_POSTSUBSCRIPT italic_o italic_l italic_d end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) = 0 , (16)

leading to an inevitable linear dependence between some states.

Instead, here we propose a generalised extension of $\psi\in Aut(\mathbb{S})$ψぷさい ∈ italic_A italic_u italic_t ( blackboard_S ) into $\mathbb{C}\ell(8)$ by defining;

$$e_{i}\xrightarrow{\psi}\begin{cases}\frac{1}{4}e_{i}-\frac{\sqrt{3}}{4}e_{i}e_{8}+\frac{\sqrt{3}}{4}e_{i+8}-\frac{3}{4}e_{i+8}e_{8}&i=\{0,...,7\},\\ \frac{1}{4}e_{i}-\frac{\sqrt{3}}{4}e_{i}e_{8}-\frac{\sqrt{3}}{4}e_{i-8}+\frac{3}{4}e_{i-8}e_{8}&i=\{8,...,15\}.\end{cases}$$ψぷさい → end_ARROW { start_ROW start_CELL divide start_ARG 1 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT - divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT + divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i + 8 end_POSTSUBSCRIPT - divide start_ARG 3 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i + 8 end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT end_CELL start_CELL italic_i = { 0 , … , 7 } , end_CELL end_ROW start_ROW start_CELL divide start_ARG 1 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT - divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT - divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i - 8 end_POSTSUBSCRIPT + divide start_ARG 3 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT italic_i - 8 end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT end_CELL start_CELL italic_i = { 8 , … , 15 } . end_CELL end_ROW (17)

For example;

$$e_{1} \xrightarrow{\psi}\frac{1}{4}e_{1}-\frac{\sqrt{3}}{4}e_{1}e_{8}+\frac{\sqrt{3}}{4}e_{9}-\frac{3}{4}e_{9}e_{8}.$$ ψぷさい → end_ARROW divide start_ARG 1 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT - divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT + divide start_ARG square-root start_ARG 3 end_ARG end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT 9 end_POSTSUBSCRIPT - divide start_ARG 3 end_ARG start_ARG 4 end_ARG italic_e start_POSTSUBSCRIPT 9 end_POSTSUBSCRIPT italic_e start_POSTSUBSCRIPT 8 end_POSTSUBSCRIPT . (18)

While the $e_{i+8}$ terms can be rewritten as $\mathbb{C}\ell(8)$ elements as per eqn. (8), it is more convenient to leave them as $e_{i+8}$. It is easily checked that $\psi^{3}=\text{Id}$ψぷさい start_POSTSUPERSCRIPT 3 end_POSTSUPERSCRIPT = Id, and that both $e_{0}$ and $e_{8}$ are invariant under $\psi$ψぷさい. The $\mathbb{C}\ell(8)$ maps $\epsilon$ and $\psi$ψぷさい can then be seen to generate $S_{3}$.

Because;

$$\psi(e_{i})\psi(e_{i})=\psi(e_{i}^{2}) =-1,$$ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) italic_ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) = italic_ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ) (19)

$$\psi(e_{i})\psi(e_{j})+\psi(e_{j})\psi(e_{i}) =0,$$ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) italic_ψぷさい ( italic_e start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ) + italic_ψぷさい ( italic_e start_POSTSUBSCRIPT italic_j end_POSTSUBSCRIPT ) italic_ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) (20)

for $i,j\in\{0,1,...,8\},\;i\neq j$, and likewise for $\epsilon$, these maps extend to $\mathbb{C}\ell(8)$ homomorphisms [harvey1990spinors]. It is trivial to show that both maps are one-to-one and thus extend to algebra automorphisms. Unlike in our previous paper, $e_{i}$, $\psi(e_{i})$ψぷさい ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) and $\psi^{2}(e_{i})$ψぷさい start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_e start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) are now linearly independent in $\mathbb{C}\ell(8)$.

4.2 Including two additional generations using the order-three symmetry $\psi$ψぷさい

Applying $\psi$ψぷさい to $a_{i}$ and $a_{i}^{\dagger}$ generates two additional Witt bases;

$$b_{i}=\psi(a_{i}),\hskip 10.00002ptb_{i}^{\dagger}=\psi(a_{i}^{\dagger}),\hskip 10.00002ptc_{i}=\psi^{2}(a_{i}),\hskip 10.00002ptc_{i}^{\dagger}=\psi^{2}(a_{i}^{\dagger}),\quad i=\{1,2,3,4\},$$ψぷさい ( italic_a start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) , italic_b start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT = italic_ψぷさい ( italic_a start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT ) , italic_c start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT = italic_ψぷさい start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_a start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT ) , italic_c start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT = italic_ψぷさい start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_a start_POSTSUBSCRIPT italic_i end_POSTSUBSCRIPT start_POSTSUPERSCRIPT † end_POSTSUPERSCRIPT ) , italic_i = { 1 , 2 , 3 , 4 } , (21)

satisfying the same anticommutation relations as our original Witt basis. From these we construct two additional minimal left ideals;

$$T_{2}=(r_{0}^{{}^{\prime}}+r_{j}^{{}^{\prime}}b_{j}^{\dagger}+r_{jk}^{{}^{\prime}}b_{j}^{\dagger}{b_{k}^{\dagger}}+r_{jkl}^{{}^{\prime}}b_{j}^{\dagger}{b_{k}^{\dagger}}b_{l}^{\dagger}+r_{1234}^{{}^{\prime}}b_{1}^{\dagger}{b_{2}^{\dagger}}b_{3}^{\dagger}{b_{4}^{\dagger}})v_{2},$$ (22)

$$T_{3}=(r_{0}^{{}^{\prime\prime}}+r_{j}^{{}^{\prime\prime}}C_{j}^{\dagger}+r_{jk}^{{}^{\prime\prime}}c_{j}^{\dagger}{c_{k}^{\dagger}}+r_{jkl}^{{}^{\prime\prime}}c_{j}^{\dagger}{c_{k}^{\dagger}}c_{l}^{\dagger}+r_{1234}^{{}^{\prime\prime}}c_{1}^{\dagger}{c_{2}^{\dagger}}c_{3}^{\dagger}{c_{4}^{\dagger}})v_{3},$$ (23)

where $v_{2}=\psi(v_{1})$ψぷさい ( italic_v start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ), $v_{3}=\psi^{2}(v_{1})$ψぷさい start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ( italic_v start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT ), $j,k,l\in\{1,2,3,4\}$, $j\neq{k}\neq{l}$ and $r_0^{{}^{\prime }},r_j^{{}^{\prime }},r_{jk}^{{}^{\prime }},r_{jkl}^{{}^{\prime }},r_{1234}^{{}^{\prime }}$