Matrix Representations Conventions
Introduction
The algebra is read in matrices in three ways, and the three answer different questions. The axis-free representation reads multiplication in the basis the algebra already carries, and writes the two operators $L_{\mathrm{axf}}(\tilde Q)$ and $R_{\mathrm{axf}}(\tilde Q)$; it needs no frame, its entries are the structure constants, and it is the display a computation uses when no direction is to be singled out. The Pauli representation reads the same two operators in the Peirce frame of the standard pair; it costs the choice of a frame and buys the $2\times2$ block, the five-operator Clifford frame and the gamma matrices. The $2\times2$ isomorphism $\mathsf{M}_2:\mathbb{B}\to M_2(\mathbb{C})$ reads the element itself as a complex matrix of the algebra's own size; it is not a regular representation at all, and it is the block from which the Pauli display is built.
The order of the article is that order. The axis-free display comes first because it follows from the multiplication table alone, with nothing chosen. The Pauli display comes second because it needs the standard pair, and once it is written the switch between the two frames is one matrix and its inverse, §The Switch Between the Two Frames. The $2\times2$ isomorphism comes last because it is the algebra and not an operator on it, and it is read best after both $4\times4$ displays have shown what a matrix of an operator is. The rule that keeps the two legacy formulations out of every body section is stated before all three, in §The Rule on Pauli and Gamma Matrices.
Three displays, two objects. The two $4\times4$ displays are one operator in two bases: the same four-dimensional subalgebra $\mathbb{B}_L$ of the ambient $M_4(\mathbb{C})=\mathrm{End}_{\mathbb{C}}(\mathbb{B})$, written twice. The $2\times2$ display is a different object, the algebra itself. A change of frame never changes the operator, and it never touches the $2\times2$ image, because that image carries no frame.
The order renames nothing. The corpus calls the Peirce frame the canonical frame, and this article keeps the name: the word names the frame, not its position here. The axis-free frame is the coefficient basis read as a frame, and its label is $\mathrm{axf}$.
The algebra is $M_2(\mathbb{C})$, and it is never $M_4(\mathbb{C})$. The ambient $M_4(\mathbb{C})=\mathrm{End}_{\mathbb{C}}(\mathbb{B})$ is the operator algebra, it carries no biquaternion structure and takes no biquaternion name; the left and right copies are named $\mathbb{B}_L$ and $\mathbb{B}_R$, and a matrix of the left copy is a matrix of $\mathbb{B}_L$.
The article assumes the base of Algebra and Element Conventions and the Peirce convention of Special Element and Peirce Conventions, whose standard pair fixes the Pauli frame.
The Rule on Pauli and Gamma Matrices
The rule. No Pauli matrix and no gamma matrix is written in the body of an article. The two formulations are legacy, and they appear in two places only: in a legacy section of an article, and in this article, which states them once so that a legacy section has something to refer to.
What the rule forbids. The arrays $\sigma_k$ and $\gamma_\mu$. A symbol built on them and used as an object of the algebra, such as $\gamma_5$ or $\sigma_3$. A sentence that reads a biquaternion off such an array by name.
What the rule allows. The names of the frames. The Pauli frame and the gamma frame are bases and not arrays, and the names stay. A reference to a legacy section stays, and so does a historical remark inside one.
The exception. The two articles whose subject is the Pauli display print the three arrays once, in the section that fixes the model: $M_2(\mathbb{C})_{\mathbb{H}}$ Pauli Representation in the quaternion category and $M_2(\mathbb{C})$ Pauli Representation in the biquaternion category. The reason is that in those two the display is the subject and not a notation, and a reader who is told that the images are $-i\sigma_k$ must be able to see what $\sigma_k$ is. The exception is confined to those two articles and to the section that fixes the model; every other body section follows the rule, and a sigma in one is still a violation.
The replacement. The basis operators. $L_aR_b$ writes an operator, $L_{\mathrm{axf}}$ and $R_{\mathrm{axf}}$ write a display in the axis-free frame, and $\mathsf{V}_{\mathrm{vol}}=-iL_{e_t}R_{\vec e_z}$ writes the chirality operator of a frame. Every array the rule forbids has a basis operator that says the same thing and names no axis.
The test. Open the article at a body section and read the symbols. A sigma or a gamma in a body section is a violation, the two sections named in §The Rule on Pauli and Gamma Matrices aside. The display is replaced by its basis operators before the article is published, and the array moves to the legacy section if it is worth keeping.
Why this is a rule and not a taste. An array carries a frame, a dimension and an inertia inside a symbol, and the sentence that prints it does not say so. A basis operator carries none of the three. That is the defect stated in full in Why AXF Matrices instead of Pauli and Gamma Matrices, and it is the reason the two formulations are kept as a record and not used as a language.
The Matrix Models
Two kinds of matrix model read the algebra. The isomorphism $\mathsf{M}_2:\mathbb{B}\to M_2(\mathbb{C})$ reads the element itself as a $2\times2$ complex matrix, of the algebra's own size. The regular representation reads multiplication as a linear map on the algebra taken as a complex vector space, and its target is $\mathrm{End}_{\mathbb{C}}(\mathbb{B})=M_4(\mathbb{C})$; its image is a four-dimensional subspace of that ambient algebra, not the ambient algebra. The two are different representations, not two sizes of one.
The matrices are the matrices of operators. Every array of this chapter is the matrix of a complex-linear operator of $\mathbb{B}$ read in a basis; the operator is the object and the array is one display of it. The two elementary operators are the multiplications $L_{\tilde Q}$ and $R_{\tilde Q_2}$, and their product is the two-sided operator of Operators Acting on Two Sides, $$ L_{\tilde Q}R_{\tilde Q_2}:\mathbb{B}\longrightarrow\mathbb{B}, \qquad L_{\tilde Q}R_{\tilde Q_2}(X)=\tilde Q\,X\,\tilde Q_2, $$ so the matrix of a two-sided operator is the product of the two regular matrices, taken in the same frame on both factors, $$ \mathsf{M}_{4L}(\tilde Q)\,\mathsf{M}_{4R}(\tilde Q_2). $$ This is the whole reason the article writes both families and their products: the object displayed is an operator of the two-sided family, and the matrix is what a computation can multiply.
The sixteen basis operators are the sixteen two-sided products. The pair $(\mu,\nu)$ names the basis two-sided operator $e_\mu[\,]e_\nu$, $X\mapsto e_\mu Xe_\nu$, and its matrix is the basis operator $L_{e_\mu}R_{e_\nu}$. The sixteen basis operators are therefore the matrices of a basis: by the generation theorem of that article, §The Sixteen Products as a Basis, they are linearly independent and they span the ambient $\operatorname{End}_{\mathbb{C}}(\mathbb{B})=M_4(\mathbb{C})$, of complex dimension sixteen. Every operator of $\mathbb{B}$ is a complex combination of the sixteen basis operators, and the sixteen coefficients are its coordinates in that basis.
The coefficients of a two-sided operator are an outer product. Expanding in the operator basis, $$ L_{\tilde Q}R_{\tilde Q_2}=\sum_{\mu,\nu}Q_\mu Q_{2\nu}\,L_{e_\mu}R_{e_\nu}, \qquad c\bigl(L_{\tilde Q}R_{\tilde Q_2}\bigr)=QQ_2^{\mathsf T}, $$ so the coefficient array $c(T)$ of The Enveloping Operator Algebra of the Biquaternion Algebra and the Bi-module Structure has rank one, and the length of the operator, $\ell(T)=\operatorname{rank}c(T)$, is one. A general operator has a coefficient matrix of any rank up to four, and a sum $L_{\tilde Q}R_{\tilde Q_2}+L_{\tilde Q_3}R_{\tilde Q_4}$ is one two-sided operator exactly when the sum of the two outer products still has rank one. The rank-one set of coefficient arrays is the Segre variety of the two-sided family, and the linear space of all sixteen coefficients is the enveloping algebra $\mathbb{B}\otimes\mathbb{B}_{\mathrm{opp}}$ of The Enveloping Operator Algebra of the Biquaternion Algebra and the Bi-module Structure, the basis operator $L_{e_\mu}R_{e_\nu}$ being the image of $e_\mu\otimes e_{\mathrm{opp},\nu}$. One table of sixteen basis operators therefore serves three readings: the matrices of the basis two-sided operators, the coordinates of the enveloping algebra, and, in §The Clifford Structure, the Clifford algebra itself.
| object | writing | meaning |
|---|---|---|
| the algebra | $\mathbb{B}$ | $\mathbb{C}\otimes_{\mathbb{R}}\mathbb{H}\cong M_2(\mathbb{C})$ |
| the ambient | $M_4(\mathbb{C})=\mathrm{End}_{\mathbb{C}}(\mathbb{B})$ | operators on the algebra as a vector space |
| the left copy | $\mathbb{B}_L=M_4(\mathbb{C})_{\mathbb{B}_L}=\rho_L(\mathbb{B})$ | four-dimensional subalgebra |
| the right copy | $\mathbb{B}_R=M_4(\mathbb{C})_{\mathbb{B}_R}=\rho_R(\mathbb{B})$ | the commutant of $\mathbb{B}_L$ |
| the isomorphism | $\mathsf{M}_2:\mathbb{B}\to M_2(\mathbb{C})$ | the element read as a $2\times2$ matrix |
| the axis-free matrix | $\mathsf{M}_{\mathrm{axf},4L}(\tilde Q)$, $\mathsf{M}_{\mathrm{axf},4R}(\tilde Q)$ | the regular matrices in the coefficient basis, written $L_{\mathrm{axf}}(\tilde Q)$, $R_{\mathrm{axf}}(\tilde Q)$ |
| the Pauli matrix | $\mathsf{M}_{\mathrm{pau},4L}(\tilde Q)$, $\mathsf{M}_{\mathrm{pau},4R}(\tilde Q)$ | the regular matrices in the Peirce frame, written $L(\tilde Q)$, $R(\tilde Q)$ |
The subscript names the copy: $M_4(\mathbb{C})_{\mathbb{B}_L}$ is the left copy, $M_4(\mathbb{C})_{\mathbb{B}_R}$ the right, and $M_4(\mathbb{C})_{\mathbb{B}_L,\mathbb{B}_R}$ both. A bare $M_4(\mathbb{C})$ is the ambient. The two cones are written $M_4(\mathbb{C})_{\mathcal{T}}$ for the two-sided operators and $M_4(\mathbb{C})_{\Theta}$ for the two-sided operators, whose names are in Operator Family Conventions.
The subscript names the action, not a tensor slot. $L$ is left multiplication and $R$ is right multiplication, read as on $L_{\tilde Q}$ and $R_{\tilde Q}$ in Operator and Calculus Conventions, §The Operators. In the Pauli frame the left copy happens to be the second tensor factor and the right copy the first, and another frame exchanges the two slots without exchanging the two actions. A subscript naming a slot would therefore change meaning with the frame, and the action does not.
General Matrix Conventions
A matrix carries no tilde. The rule is that of Algebra and Element Conventions, §The Elements and their Coefficients, read here on the arrays of this article: the tilde marks the element and never the array that holds it. The image of an element is $\mathsf{M}_2(\tilde Q)$ or $\mathsf{M}_{4L}(\tilde Q)$, with the tilde inside the argument, and the array of a regular matrix is $L_{\mathrm{axf}}(\tilde Q)$, $R_{\mathrm{axf}}(\tilde Q)$, $L(\tilde Q)$ or $R(\tilde Q)$, without one. The identity is $I_n$, and the two regular families $\mathsf{M}_{4L}$ and $\mathsf{M}_{4R}$ take no tilde either, their subscript giving the size first and the side of the multiplication second. An array whose entries are elements is written without a tilde as well, the tildes sitting on the entries, $\mathsf{Q}=(\tilde Q_{\lambda\rho})\in\mathbb{B}^{n\times m}$.
The ban is on the array, not on its argument. A tilde on the array, as in $\tilde I_n$, $\tilde{\mathsf{M}}_2$ or $\tilde{\mathcal Q}$, is forbidden and is a defect to be rewritten, not a variant to be read: no object stands behind it, and the element it would name is already named by the argument.
The letter $M$ belongs to matrices. It names a matrix and nothing else: the images $\mathsf{M}_2(\tilde Q)$ and $\mathsf{M}_{4L}(\tilde Q)$, the matrix algebras $M_2(\mathbb{C})$ and $M_4(\mathbb{C})$, and an entrywise matrix in a proof. No element of the algebra is written $M$, with or without a tilde.
The copy is named, the ambient is not. The left copy is $\mathbb{B}_L=M_4(\mathbb{C})_{\mathbb{B}_L}=\rho_L(\mathbb{B})$ and the right copy $\mathbb{B}_R=M_4(\mathbb{C})_{\mathbb{B}_R}=\rho_R(\mathbb{B})$, each a subalgebra of $M_4(\mathbb{C})$ isomorphic to $M_2(\mathbb{C})$ and the commutant of the other. The ambient $M_4(\mathbb{C})=\mathrm{End}_{\mathbb{C}}(\mathbb{B})$ is the operator algebra and carries no biquaternion name. The subscript names the copy, so $M_4(\mathbb{C})_{\mathbb{B}_L}$ is $\mathbb{B}_L$ and $M_4(\mathbb{C})_{\mathbb{B}_R}$ is $\mathbb{B}_R$; a subscript $L$ or $R$ alone names no algebra.
The frame label names the display. $\mathsf{M}_{4L}$ names the operator and carries no frame. The array of that operator carries the frame, $L_{\mathrm{axf}}(\tilde Q)$ in the coefficient basis and $L(\tilde Q)$ in the Peirce frame, or the frame is named in the sentence; an array of no named frame is not written. A bare $\mathsf{M}_4$ is written for neither the operator's family nor the frame.
The $M_4(\mathbb{C})$ Axis-Free Representation
The basis is the one the algebra carries. The algebra is $\mathbb{B}=\mathbb{C}\otimes_{\mathbb{R}}\mathbb{H}$, and its basis is the tensor basis of Introduction to the Algebra of Biquaternions, §The Basis and the Complexification, $$ e_t=1\otimes1,\qquad \vec e_x=1\otimes\vec e_x,\qquad \vec e_y=1\otimes\vec e_y,\qquad \vec e_z=1\otimes\vec e_z, $$ the vector units on the right being those of $\mathbb{H}$, and the scalar imaginary $i=i\otimes1$ being central and occurring only as a coefficient. The four are a $\mathbb{C}$-basis, and the element is $$ \tilde Q=Q_te_t+Q_x\vec e_x+Q_y\vec e_y+Q_z\vec e_z,\qquad Q_\mu=q_\mu+iq'_\mu, $$ the developed form of The Element, the Coordinates and the Conjugations in the Biquaternion Algebra, §The Coordinate Systems. This basis is called the coefficient basis, because its four coordinates are the coefficients of the developed form, and it is called axis-free, because no one of the three vector directions enters by a rule of its own.
The multiplication table. Every matrix below is read from this table and from nothing else:
| $e_\mu e_\nu$ | $e_t$ | $\vec e_x$ | $\vec e_y$ | $\vec e_z$ |
|---|---|---|---|---|
| $e_t$ | $e_t$ | $\vec e_x$ | $\vec e_y$ | $\vec e_z$ |
| $\vec e_x$ | $\vec e_x$ | $-e_t$ | $\vec e_z$ | $-\vec e_y$ |
| $\vec e_y$ | $\vec e_y$ | $-\vec e_z$ | $-e_t$ | $\vec e_x$ |
| $\vec e_z$ | $\vec e_z$ | $\vec e_y$ | $-\vec e_x$ | $-e_t$ |
Each of the sixteen products is $\pm$ one of the four basis elements, and the signs are those of the quaternion algebra, $\vec e_k^2=-e_t$ and $\vec e_x\vec e_y=\vec e_z$, $\vec e_y\vec e_z=\vec e_x$, $\vec e_z\vec e_x=\vec e_y$. No complex number appears in the table, because $i$ is central and is carried by the coefficients. In the compact form, $$ e_\mu e_\nu=\sum_\rho\varepsilon_{\mu\nu}^{\ \ \rho}\,e_\rho,\qquad \varepsilon_{\mu\nu}^{\ \ \rho}\in\{0,+1,-1\}, $$ and the array $\varepsilon$ is the algebra: the table is that array written out.
Why this frame is first. The construction uses the multiplication of the algebra and nothing else: no idempotent is chosen, no direction is singled out, no root is fixed. The display is therefore unique once the basis is, and the basis is the one the algebra comes with. That is why it is the first display of the article, and why a computation that must not favour an axis starts here.
The label. Write $L_{\mathrm{axf}}(\tilde Q)$ and $L_{\mathrm{axf},e_\mu}$, and $R_{\mathrm{axf}}(\tilde Q)$ and $R_{\mathrm{axf},e_\mu}$. The label is a name in lowercase roman letters, like $\mathrm{opp}$ and $\mathrm{fop}$, not an exponent: $L_{\mathrm{axf},\vec e_x}$ is the left matrix of $\vec e_x$, and its square is written $(L_{\mathrm{axf},\vec e_x})^2$. The label is carried by the family and not by the entry, because the frame is a property of the carrier and the side is a property of the action.
The Left Family
The definition. The $\nu$-th column of the left matrix is the coordinate column of the product $\tilde Qe_\nu$, $$ L_{\mathrm{axf}}(\tilde Q)_{\mu\nu}=\bigl[\tilde Qe_\nu\bigr]_{\mu},\qquad L_{\mathrm{axf}}(\tilde Q):=\bigl(\bigl[\tilde Qe_t\bigr]\;\bigl[\tilde Q\vec e_x\bigr]\;\bigl[\tilde Q\vec e_y\bigr]\;\bigl[\tilde Q\vec e_z\bigr]\bigr), $$ which is the definition of Families of Matrix Representations, §The Two Sides, read in the basis $(e_t,\vec e_x,\vec e_y,\vec e_z)$. The four matrices of the units are $L_{\mathrm{axf},e_\mu}:=L_{\mathrm{axf}}(e_\mu)$.
The display. For $\tilde Q=Q_te_t+Q_x\vec e_x+Q_y\vec e_y+Q_z\vec e_z$, $$ L_{\mathrm{axf}}(\tilde Q)= \begin{pmatrix} Q_t&-Q_x&-Q_y&-Q_z\\ Q_x&Q_t&-Q_z&Q_y\\ Q_y&Q_z&Q_t&-Q_x\\ Q_z&-Q_y&Q_x&Q_t \end{pmatrix}. $$ The first column is $(Q_t,Q_x,Q_y,Q_z)^{\mathsf T}$, the coordinate column of $\tilde Q$ itself, because $\tilde Qe_t=\tilde Q$. The other three columns are the products of $\tilde Q$ with the three vector units, one rule applied to the other three columns of the table.
The four left matrices. On the units, $$ L_{\mathrm{axf},e_t}=\mathrm I_4,\qquad L_{\mathrm{axf},\vec e_x}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&1&0\end{pmatrix}, $$ $$ L_{\mathrm{axf},\vec e_y}=\begin{pmatrix}0&0&-1&0\\0&0&0&1\\1&0&0&0\\0&-1&0&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_z}=\begin{pmatrix}0&0&0&-1\\0&0&-1&0\\0&1&0&0\\1&0&0&0\end{pmatrix}. $$ Each is a column of the multiplication table written as an array: the $z$ matrix has $-1$ where the table sends $e_t\mapsto\vec e_z$, $\vec e_y\mapsto-\vec e_x$, $\vec e_z\mapsto-e_t$ and $\vec e_x\mapsto\vec e_y$.
Signed permutations. Each of the four is a signed permutation matrix: exactly one nonzero entry in each row and in each column, of modulus one, hence orthogonal with entries in $\{0,+1,-1\}$. Each has determinant $1$; the identity has trace $4$ and the three vector matrices have trace $0$. The four are linearly independent, since the assignment is injective and the four units are a basis, so they are a basis of $\mathbb{B}_L$.
The group. The four matrices together with their negatives are eight distinct signed permutation matrices forming a group isomorphic to the quaternion group $Q_8$: the vector matrices square to $-\mathrm I_4$, distinct ones anticommute up to sign, and the identity is the unit. The left family of this frame is a copy of $Q_8$ inside $O(4)\subset M_4(\mathbb{R})$.
The law. For all $\tilde Q_1,\tilde Q_2$, $$ L_{\mathrm{axf}}(\tilde Q_1\tilde Q_2)=L_{\mathrm{axf}}(\tilde Q_1)L_{\mathrm{axf}}(\tilde Q_2), $$ which is associativity and nothing else: the $\nu$-th column of the left side is the coordinate column of $(\tilde Q_1\tilde Q_2)e_\nu=\tilde Q_1(\tilde Q_2e_\nu)$, and the $\nu$-th column of the right side is $\tilde Q_1$ applied to the coordinate column of $\tilde Q_2e_\nu$. The first column recovers $\tilde Q$, so the assignment is injective, and its image is the four-dimensional complex subalgebra $\mathbb{B}_L$ of $M_4(\mathbb{C})$ whose unit is $\mathrm I_4$.
The transpose is the conjugation. In this frame, $$ L_{\mathrm{axf}}(\tilde Q)^{\mathsf T}=L_{\mathrm{axf}}(\tilde Q^{\natural}),\qquad \tilde Q^{\natural}=Q_te_t-Q_x\vec e_x-Q_y\vec e_y-Q_z\vec e_z, $$ because transposing swaps the $(\mu,\nu)$ and the $(\nu,\mu)$ entries, and those two carry opposite signs exactly when $s_\mu s_\nu=-1$, with $s_t=1$ and $s_x=s_y=s_z=-1$; multiplying the three vector coefficients by $-1$ produces that sign pattern. The dagger gives the star conjugate, $L_{\mathrm{axf}}(\tilde Q)^{\dagger}=L_{\mathrm{axf}}(\tilde Q^{*})$, because the entrywise conjugate of the display is the display of the element with $\overline{Q_\mu}$ in place of $Q_\mu$. Both identities are properties of this frame and not of the algebra.
The Right Family
The definition. The $\nu$-th column of the right matrix is the coordinate column of $\vec e_\nu\tilde Q$, $$ R_{\mathrm{axf}}(\tilde Q)_{\mu\nu}=\bigl[\vec e_\nu\tilde Q\bigr]_{\mu},\qquad R_{\mathrm{axf}}(\tilde Q):=\bigl(\bigl[e_t\tilde Q\bigr]\;\bigl[\vec e_x\tilde Q\bigr]\;\bigl[\vec e_y\tilde Q\bigr]\;\bigl[\vec e_z\tilde Q\bigr]\bigr), $$ the same construction with the order of the two factors exchanged.
The display. $$ R_{\mathrm{axf}}(\tilde Q)= \begin{pmatrix} Q_t&-Q_x&-Q_y&-Q_z\\ Q_x&Q_t&Q_z&-Q_y\\ Q_y&-Q_z&Q_t&Q_x\\ Q_z&Q_y&-Q_x&Q_t \end{pmatrix}. $$ The two displays agree on the diagonal, in the first row and in the first column, and differ in the sign of the six entries whose two indices are both vector.
The four right matrices. On the units, $$ R_{\mathrm{axf},e_t}=\mathrm I_4,\qquad R_{\mathrm{axf},\vec e_x}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&1\\0&0&-1&0\end{pmatrix}, $$ $$ R_{\mathrm{axf},\vec e_y}=\begin{pmatrix}0&0&-1&0\\0&0&0&-1\\1&0&0&0\\0&1&0&0\end{pmatrix},\qquad R_{\mathrm{axf},\vec e_z}=\begin{pmatrix}0&0&0&-1\\0&0&1&0\\0&-1&0&0\\1&0&0&0\end{pmatrix}. $$ Each is a signed permutation matrix of determinant $1$ and trace $4\delta_{\mu t}$, and the four are a basis of $\mathbb{B}_R$.
The law is reversed. For all $\tilde Q_1,\tilde Q_2$, $$ R_{\mathrm{axf}}(\tilde Q_1\tilde Q_2)=R_{\mathrm{axf}}(\tilde Q_2)R_{\mathrm{axf}}(\tilde Q_1), $$ because associativity turns $\vec e_\nu(\tilde Q_1\tilde Q_2)$ into $(\vec e_\nu\tilde Q_1)\tilde Q_2$, so the columns compose in the reverse order. Powers and inverses are not reversed, since they involve one element only: $R_{\mathrm{axf}}(\tilde Q^n)=R_{\mathrm{axf}}(\tilde Q)^n$ and $R_{\mathrm{axf}}(\tilde Q^{-1})=R_{\mathrm{axf}}(\tilde Q)^{-1}$. The products of the unit matrices follow the reverse order too, $R_{\mathrm{axf},e_\mu}R_{\mathrm{axf},e_\nu}=\varepsilon'R_{\mathrm{axf},e_{\nu\mu}}$ with $e_\nu e_\mu=\varepsilon'e_{\nu\mu}$, so $R_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_y}=-R_{\mathrm{axf},\vec e_z}$, the opposite sign to the left family.
The two families commute. For all $\tilde Q_1,\tilde Q_2$, $$ L_{\mathrm{axf}}(\tilde Q_1)R_{\mathrm{axf}}(\tilde Q_2)=R_{\mathrm{axf}}(\tilde Q_2)L_{\mathrm{axf}}(\tilde Q_1), $$ because the two factors sit on opposite sides of the argument and never meet. The two families meet in the scalars: $L_{\mathrm{axf}}(\tilde Q)=R_{\mathrm{axf}}(\tilde Q)$ says that $\tilde Q$ commutes with every element, that is $\tilde Q\in\mathbb{C}e_t$, and then the common matrix is $\lambda\mathrm I_4$.
The Stitch Between the Two Families
The general rule. In any frame the right matrix is the transpose of the left one between the switch matrix and its inverse, $$ R(\tilde Q)=\mathsf W^{-1}L(\tilde Q)^{\mathsf T}\mathsf W, $$ where $\mathsf W$ is the Gram matrix of the symmetric form $\beta(\tilde X,\tilde Y)=\operatorname{Sc}(\tilde X\tilde Y)$ on the frame (Families of Matrix Representations).
The switch of this frame. On the four units, $\beta(e_\mu,e_\nu)=\operatorname{Sc}(e_\mu e_\nu)$ is $+1$ on $e_t$, $-1$ on the three vector units and $0$ off the diagonal, so $$ \mathsf W=D=\operatorname{diag}(1,-1,-1,-1), \qquad R_{\mathrm{axf}}(\tilde Q)=D\,L_{\mathrm{axf}}(\tilde Q)^{\mathsf T}D, \qquad D=D^{\mathsf T}=D^{-1}. $$ The same array $D$ is the sign vector $\varepsilon$ of the products: the transpose dictionary and the plain weight are one object here. This is the cheapest stitch in the corpus, and it is the reason the transpose is the conjugation above.
The switch matrix, written out. It is its own transpose and its own inverse, $$ D=\begin{pmatrix}1&0&0&0\\0&-1&0&0\\0&0&-1&0\\0&0&0&-1\end{pmatrix}, \qquad D^{\mathsf T}=\begin{pmatrix}1&0&0&0\\0&-1&0&0\\0&0&-1&0\\0&0&0&-1\end{pmatrix}=D, \qquad D^{-1}=D . $$ Entrywise the stitch reads $$ \bigl(R_{\mathrm{axf}}(\tilde Q)\bigr)_{ij}=d_i\,d_j\,\bigl(L_{\mathrm{axf}}(\tilde Q)\bigr)_{ji}, \qquad d_1=1,\quad d_2=d_3=d_4=-1, $$ so transposing already writes $R$ except in the bottom-right $3\times3$ block, and the two factors of $D$ negate that block and only that block. On the general element the two displays of §The Left Family and §The Right Family differ in exactly those six entries, and on the unit $\vec e_x$ the computation is $$ D\,\bigl(L_{\mathrm{axf},\vec e_x}\bigr)^{\mathsf T}D =\begin{pmatrix}1&0&0&0\\0&-1&0&0\\0&0&-1&0\\0&0&0&-1\end{pmatrix} \begin{pmatrix}0&1&0&0\\-1&0&0&0\\0&0&0&1\\0&0&-1&0\end{pmatrix} \begin{pmatrix}1&0&0&0\\0&-1&0&0\\0&0&-1&0\\0&0&0&-1\end{pmatrix} =\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&1\\0&0&-1&0\end{pmatrix} =R_{\mathrm{axf},\vec e_x} . $$ The four matrix displays of the two families are the same arrays read in the two ways, and nothing in the algebra is changed by the choice.
The invariants. The trace and the determinant are $$ \operatorname{tr}L_{\mathrm{axf}}(\tilde Q)=4Q_t,\qquad \det L_{\mathrm{axf}}(\tilde Q)=\langle\tilde Q\rangle_{\natural}^{2}, $$ and the same values hold for $R$, since the stitch changes neither a trace nor a determinant. They are the values of the matrix of the operator, not of the frame.
The Sixteen Products in This Frame
They are real. The sixteen products $L_{\mathrm{axf},e_\mu}R_{\mathrm{axf},e_\nu}$ are real signed permutation matrices: four entries $\pm1$, one in each row and each column, of determinant $1$, of Hilbert–Schmidt norm two, and mutually orthogonal. Their entries are the structure constants themselves, so the product table of the algebra and the multiplication of the sixteen arrays are one table.
The sixteen, one by one. Each is a signed permutation matrix, with four entries $\pm1$, and each is written here with both factors named:
$$ L_{\mathrm{axf},e_t}R_{\mathrm{axf},e_t}=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix},\qquad L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_x}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&1\\0&0&-1&0\end{pmatrix} $$
$$ L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_y}=\begin{pmatrix}0&0&-1&0\\0&0&0&-1\\1&0&0&0\\0&1&0&0\end{pmatrix},\qquad L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_z}=\begin{pmatrix}0&0&0&-1\\0&0&1&0\\0&-1&0&0\\1&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},e_t}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&1&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_x}=\begin{pmatrix}-1&0&0&0\\0&-1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_y}=\begin{pmatrix}0&0&0&1\\0&0&-1&0\\0&-1&0&0\\1&0&0&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_z}=\begin{pmatrix}0&0&-1&0\\0&0&0&-1\\-1&0&0&0\\0&-1&0&0\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},e_t}=\begin{pmatrix}0&0&-1&0\\0&0&0&1\\1&0&0&0\\0&-1&0&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_x}=\begin{pmatrix}0&0&0&-1\\0&0&-1&0\\0&-1&0&0\\-1&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_y}=\begin{pmatrix}-1&0&0&0\\0&1&0&0\\0&0&-1&0\\0&0&0&1\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_z}=\begin{pmatrix}0&1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&-1&0\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},e_t}=\begin{pmatrix}0&0&0&-1\\0&0&-1&0\\0&1&0&0\\1&0&0&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_x}=\begin{pmatrix}0&0&1&0\\0&0&0&-1\\1&0&0&0\\0&-1&0&0\end{pmatrix} $$
$$ L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_y}=\begin{pmatrix}0&-1&0&0\\-1&0&0&0\\0&0&0&-1\\0&0&-1&0\end{pmatrix},\qquad L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_z}=\begin{pmatrix}-1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&-1\end{pmatrix} $$
The product closes on the basis operator. They multiply by multiplying the left factors together and the right factors together, $$ L_{\mathrm{axf},e_\mu}R_{\mathrm{axf},e_\nu}\,L_{\mathrm{axf},e_\rho}R_{\mathrm{axf},e_\sigma} =L_{\mathrm{axf},e_\mu e_\rho}R_{\mathrm{axf},e_\sigma e_\nu}, $$ which is again a basis operator, so the sixteen span a sixteen-dimensional algebra: the whole ambient $M_4(\mathbb{C})$, as §The Clifford Structure reads it in the other frame.
The sixteen basis operators, named on the units. Each basis operator acts on the basis as a signed permutation, so the four images of $(e_t,\vec e_x,\vec e_y,\vec e_z)$ carry the whole array, and the sixteen pairs are read off the arrays above:
| basis operator | images of $(e_t,\vec e_x,\vec e_y,\vec e_z)$ | basis operator | images of $(e_t,\vec e_x,\vec e_y,\vec e_z)$ |
|---|---|---|---|
| $L_{\mathrm{axf},e_t}R_{\mathrm{axf},e_t}$ | $(e_t,\vec e_x,\vec e_y,\vec e_z)$ | $L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},e_t}$ | $(\vec e_y,-\vec e_z,-e_t,\vec e_x)$ |
| $L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_x}$ | $(\vec e_x,-e_t,-\vec e_z,\vec e_y)$ | $L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_x}$ | $(-\vec e_z,-\vec e_y,-\vec e_x,-e_t)$ |
| $L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_y}$ | $(\vec e_y,\vec e_z,-e_t,-\vec e_x)$ | $L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_y}$ | $(-e_t,\vec e_x,-\vec e_y,\vec e_z)$ |
| $L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_z}$ | $(\vec e_z,-\vec e_y,\vec e_x,-e_t)$ | $L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_z}$ | $(\vec e_x,e_t,-\vec e_z,-\vec e_y)$ |
| $L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},e_t}$ | $(\vec e_x,-e_t,\vec e_z,-\vec e_y)$ | $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},e_t}$ | $(\vec e_z,\vec e_y,-\vec e_x,-e_t)$ |
| $L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_x}$ | $(-e_t,-\vec e_x,\vec e_y,\vec e_z)$ | $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_x}$ | $(\vec e_y,-\vec e_z,e_t,-\vec e_x)$ |
| $L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_y}$ | $(\vec e_z,-\vec e_y,-\vec e_x,e_t)$ | $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_y}$ | $(-\vec e_x,-e_t,-\vec e_z,-\vec e_y)$ |
| $L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_z}$ | $(-\vec e_y,-\vec e_z,-e_t,-\vec e_x)$ | $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_z}$ | $(-e_t,\vec e_x,\vec e_y,-\vec e_z)$ |
The same sixteen basis operators in the Pauli frame. A basis operator does not change with the frame: the pair $(\mu,\nu)$ names one basis two-sided operator $e_\mu[\,]e_\nu$ in both displays, so the sixteen basis operators correspond one to one and index for index, $$ L_{\mathrm{axf},e_\mu}R_{\mathrm{axf},e_\nu}\;\longleftrightarrow\;L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}, \qquad \mu,\nu\in\{t,x,y,z\}, $$ and the coefficients $Q_\mu Q_{2\nu}$ of a two-sided operator are the same sixteen numbers in both frames. Only the arrays differ: here real signed permutations, in the Peirce frame complex arrays with entries in $\{0,\pm1,\pm i\}$. §The Switch Between the Two Frames writes the sixteen expressions one by one, each an equality of the Pauli basis operator with its axis-free partner conjugated.
The cost. No $2\times2$ block appears, no Peirce idempotent is visible, and the sixteen arrays are all of one kind, so no five of them are singled out. The frame that gives the transpose dictionary takes away the block and the gamma frame; the next sections buy them back.
The $M_4(\mathbb{C})$ Pauli Representation
A second frame, and what it buys. The same two operators are displayed in the Peirce frame of the standard pair. The frame is a choice, and it is the choice the rest of the corpus makes: it is the frame in which the left matrix is a $2\times2$ block matrix, in which every basis operator is a Kronecker product of two $2\times2$ matrices, and in which five of them are the gamma matrices. Its label is $\mathrm{pau}$.
The Peirce Frame
The two idempotents. The standard pair is $$ \tilde\Pi_\pm=\tfrac12(e_t\pm i\vec e_z),\qquad \tilde\Pi_+^2=\tilde\Pi_+,\quad \tilde\Pi_-^2=\tilde\Pi_-,\quad \tilde\Pi_+\tilde\Pi_-=0,\quad \tilde\Pi_++\tilde\Pi_-=e_t, $$ and its two mixed units are $$ \vec\Upsilon_\pm=\tfrac12(i\vec e_x\pm\vec e_y),\qquad \vec\Upsilon_\pm^2=0,\quad \vec\Upsilon_+\vec\Upsilon_-=\tilde\Pi_-,\quad \vec\Upsilon_-\vec\Upsilon_+=\tilde\Pi_+ . $$ So $(\tilde\Pi_+,\vec\Upsilon_+,\vec\Upsilon_-,\tilde\Pi_-)$ is a Peirce basis, and the frame is that basis in that order: the two elements of $\mathbb{B}\tilde\Pi_+$ first and those of $\mathbb{B}\tilde\Pi_-$ second. The Peirce convention fixes no family (Special Element and Peirce Conventions, §The Biquaternion Peirce Convention); the choice here is not a preferred pair, it is the frame this display is written in, and the corpus calls it the canonical frame.
The frame vectors over the basis. Written in the coefficient basis, the four frame vectors are $$ \tilde\Pi_+=\tfrac12\bigl(e_t+i\vec e_z\bigr),\qquad \vec\Upsilon_+=\tfrac12\bigl(i\vec e_x+\vec e_y\bigr),\qquad \vec\Upsilon_-=\tfrac12\bigl(i\vec e_x-\vec e_y\bigr),\qquad \tilde\Pi_-=\tfrac12\bigl(e_t-i\vec e_z\bigr), $$ so each is a combination of two units and the switch is two $2\times2$ blocks, as §The Switch Between the Two Frames writes it out.
The root, and the broken symmetry. The frame singles out the direction $i\vec e_z$ through $\tilde\Pi_\pm$, and the two mixed units pair $\vec e_x$ with $\vec e_y$ and leave $\vec e_z$ in the idempotents. That is the price of the block: the axis-free frame has no root and no block, and this frame has both.
The Two Tensor Factors
The Pauli matrices. The $4\times4$ arrays of this frame are built from three $2\times2$ matrices, and they are fixed here once for the whole chapter: $$ \sigma_1=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad \sigma_2=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\qquad \sigma_3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}. $$ They are Hermitian, traceless, of determinant $-1$ and of square $I_2$, and they multiply by $$ \sigma_1\sigma_2=i\sigma_3,\qquad \sigma_2\sigma_3=i\sigma_1,\qquad \sigma_3\sigma_1=i\sigma_2, \qquad \sigma_1\sigma_2\sigma_3=iI_2, $$ so $\sigma_j\sigma_k=-\sigma_k\sigma_j$ for $j\ne k$ and $\{\sigma_j,\sigma_k\}=2\delta_{jk}I_2$. The image of the algebra is their complex linear combination, $$ \mathsf{M}_2(\tilde Q)=Q_tI_2-i\bigl(Q_x\sigma_1+Q_y\sigma_2+Q_z\sigma_3\bigr) =\begin{pmatrix}Q_t-iQ_z&-iQ_x-Q_y\\-iQ_x+Q_y&Q_t+iQ_z\end{pmatrix}, $$ and every $4\times4$ array of this frame is a Kronecker product of these three with $I_2$ or with each other. That is why each $4\times4$ display carries the same $2\times2$ arrays twice, and why the four left matrices below are read as $-i\sigma_k$ repeated in the two diagonal blocks, the four right ones as $-i\sigma_k^{\mathsf T}$ multiplying the four blocks of $I_2$.
The left matrix. In the Peirce frame the left matrix of $\tilde Q$ is the $2\times2$ matrix taken twice, $$ \rho_L(\tilde Q)=I_2\otimes\mathsf{M}_2(\tilde Q),\qquad \mathsf{M}_{\mathrm{pau},4L}(\tilde Q)=L(\tilde Q)=I_2\otimes\mathsf{M}_2(\tilde Q), $$ so its four blocks are $\mathsf{M}_2(\tilde Q)$, $0$, $0$, $\mathsf{M}_2(\tilde Q)$.
The four left matrices. Each is the image of an invertible element, $$ L_{\mathrm{pau},e_t}=\mathsf{M}_{4L}(e_t)=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_x}=\mathsf{M}_{4L}(\vec e_x)=\begin{pmatrix}0&-i&0&0\\-i&0&0&0\\0&0&0&-i\\0&0&-i&0\end{pmatrix}, $$ $$ L_{\mathrm{pau},\vec e_y}=\mathsf{M}_{4L}(\vec e_y)=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&1&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_z}=\mathsf{M}_{4L}(\vec e_z)=\begin{pmatrix}-i&0&0&0\\0&i&0&0\\0&0&-i&0\\0&0&0&i\end{pmatrix}, $$ each equal to $I_2\otimes\mathsf{M}_2(\vec e_k)$, with $\mathsf{M}_2(\vec e_k)=-i\sigma_k$: $$ L_{\mathrm{pau},e_t}=\operatorname{diag}\bigl(I_2,I_2\bigr),\qquad L_{\mathrm{pau},\vec e_x}=\operatorname{diag}\bigl(-i\sigma_1,-i\sigma_1\bigr),\qquad L_{\mathrm{pau},\vec e_y}=\operatorname{diag}\bigl(-i\sigma_2,-i\sigma_2\bigr),\qquad L_{\mathrm{pau},\vec e_z}=\operatorname{diag}\bigl(-i\sigma_3,-i\sigma_3\bigr). $$ The visible structure of the four arrays is the structure of the three Pauli matrices: $L_{\mathrm{pau},\vec e_z}$ is diagonal because $\sigma_3$ is, $L_{\mathrm{pau},\vec e_x}$ and $L_{\mathrm{pau},\vec e_y}$ are off the diagonal inside each block because $\sigma_1$ and $\sigma_2$ are, and the factor $i$ is what makes each square $-I_4$. A general element is their linear combination, $$ \mathsf{M}_{4L}(\tilde Q)=Q_tL_{\mathrm{pau},e_t}+Q_xL_{\mathrm{pau},\vec e_x}+Q_yL_{\mathrm{pau},\vec e_y}+Q_zL_{\mathrm{pau},\vec e_z}=I_2\otimes\mathsf{M}_2(\tilde Q). $$
The four right matrices. Right multiplication is the transpose of left multiplication in the exchanged frame. Let $\mathsf P$ be the linear involution of the carrier that fixes the two idempotents and exchanges the two mixed Peirce units, $$ \mathsf P\tilde\Pi_\pm=\tilde\Pi_\pm,\qquad \mathsf P\vec\Upsilon_+=\vec\Upsilon_-,\qquad \mathsf P\vec\Upsilon_-=\vec\Upsilon_+, $$ the permutation of the second and third canonical coordinates. Its matrix there is $$ \mathsf P=\begin{pmatrix}1&0&0&0\\0&0&1&0\\0&1&0&0\\0&0&0&1\end{pmatrix}, $$ symmetric and involutive, $\mathsf P^{\mathsf T}=\mathsf P=\mathsf P^{-1}$, of trace $2$ and determinant $-1$; in the coefficient basis the same operator is the reflection $\operatorname{diag}(1,1,-1,1)$, which fixes $e_t,\vec e_x,\vec e_z$ and negates $\vec e_y$. It is a linear involution of the carrier, not an algebra map, not one of the sixteen basis operators and not a one-sided multiplication. Then $$ \mathsf M_{4R}(\tilde Q)=\mathsf P\,\mathsf M_{4L}(\tilde Q)^{\mathsf T}\mathsf P^{-1},\qquad R_{\mathrm{pau},e_\nu}=\mathsf P\,{L_{\mathrm{pau},e_\nu}}^{\mathsf T}\,\mathsf P^{-1}, \qquad \mathsf{M}_{\mathrm{pau},4R}(\tilde Q)=R(\tilde Q)=\mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2 . $$ Explicitly, $$ R_{\mathrm{pau},e_t}=\mathsf{M}_{4R}(e_t)=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix},\qquad R_{\mathrm{pau},\vec e_x}=\mathsf{M}_{4R}(\vec e_x)=\begin{pmatrix}0&0&-i&0\\0&0&0&-i\\-i&0&0&0\\0&-i&0&0\end{pmatrix}, $$ $$ R_{\mathrm{pau},\vec e_y}=\mathsf{M}_{4R}(\vec e_y)=\begin{pmatrix}0&0&1&0\\0&0&0&1\\-1&0&0&0\\0&-1&0&0\end{pmatrix},\qquad R_{\mathrm{pau},\vec e_z}=\mathsf{M}_{4R}(\vec e_z)=\begin{pmatrix}-i&0&0&0\\0&-i&0&0\\0&0&i&0\\0&0&0&i\end{pmatrix}. $$ The same Pauli matrices, in the other slot. The right matrices are $$ R_{\mathrm{pau},\vec e_k}=\mathsf{M}_2(\vec e_k)^{\mathsf T}\otimes I_2=\bigl(-i\sigma_k^{\mathsf T}\bigr)\otimes I_2,\qquad R_{\mathrm{pau},e_t}=I_2\otimes I_2=\operatorname{diag}(I_2,I_2), $$ so the entry $\bigl(-i\sigma_k^{\mathsf T}\bigr)_{rc}I_2$ sits in the $(r,c)$ block. The exchange of the two factors is not a transposition of the array: it moves the Pauli matrix from the diagonal blocks to the block-scaling role, and its blocks become multiples of $I_2$ instead of being $I_2$-multiples of the same $2\times2$ array.
What the exchange does, case by case. $\sigma_1$ and $\sigma_2$ have zero diagonal, so $R_{\mathrm{pau},\vec e_x}$ and $R_{\mathrm{pau},\vec e_y}$ have both diagonal blocks zero and their two off-diagonal blocks equal to $-iI_2$. $\sigma_3$ is diagonal, so the two matrices differ only in the order of four diagonal entries, $$ L_{\mathrm{pau},\vec e_z}=\operatorname{diag}(-i,\,i,\,-i,\,i),\qquad R_{\mathrm{pau},\vec e_z}=\operatorname{diag}(-i,\,-i,\,i,\,i), $$ which is the exchange written on the diagonal: the left array repeats $-i,i$ twice, the right array repeats $-i$ twice and then $i$ twice. The transposition itself leaves $R_{\mathrm{pau},\vec e_x}$ and $R_{\mathrm{pau},\vec e_z}$ as the exchange alone would give, because $\sigma_1$ and $\sigma_3$ are symmetric, and it negates the blocks of $R_{\mathrm{pau},\vec e_y}$ because $\sigma_2$ is antisymmetric: $\sigma_2^{\mathsf T}=-\sigma_2$.
The laws and the invariants. Both families obey the laws of the axis-free frame, with the same side conventions, $$ L(\tilde Q_1\tilde Q_2)=L(\tilde Q_1)L(\tilde Q_2),\qquad R(\tilde Q_1\tilde Q_2)=R(\tilde Q_2)R(\tilde Q_1),\qquad L(\tilde Q_1)R(\tilde Q_2)=R(\tilde Q_2)L(\tilde Q_1), $$ and the trace and determinant are the same numbers, $\operatorname{tr}L(\tilde Q)=4Q_t$ and $\det L(\tilde Q)=\langle\tilde Q\rangle_{\natural}^2$, because they are properties of the matrix of the operator. The dagger is the star conjugate, $L(\tilde Q)^{\dagger}=L(\tilde Q^{*})$. Transposition is not the conjugation in this frame: with the exchange it gives the other family, $R(\tilde Q)=\mathsf P L(\tilde Q)^{\mathsf T}\mathsf P^{-1}$, and the conjugation of this frame is no transpose at all, $L(\tilde Q^{\natural})=2Q_t\,\mathrm I_4-L(\tilde Q)$. The transpose is the conjugation in the axis-free frame alone, §The Stitch Between the Two Families.
The block form of the element. Since $\mathsf{M}_2(\tilde Q)=\begin{pmatrix}a&b\\ c&d\end{pmatrix}$ with $$ a=Q_t-iQ_z,\qquad b=-iQ_x-Q_y,\qquad c=-iQ_x+Q_y,\qquad d=Q_t+iQ_z, $$ the left matrix is $$ L(\tilde Q)=\begin{pmatrix}a&b&0&0\\ c&d&0&0\\0&0&a&b\\0&0&c&d\end{pmatrix}, $$ so the algebra acts on each Peirce component by the same $2\times2$ matrix, and the two components never mix. This is the block the $2\times2$ display names, §The $2\times2$ Pauli Representation.
The zero block is the point. Every entry linking an index in $\{1,2\}$ to one in $\{3,4\}$ vanishes for every element. The Peirce frame does not block an arbitrary operator; it blocks the left copy $\mathbb{B}_L$, and the vanishing is the statement that left multiplication preserves the two Peirce components.
The Sixteen Basis Operators in the Peirce Frame
The sixteen basis operators, entry by entry. The basis operators multiply left multipliers with left multipliers and right with right, $$ L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}\,L_{\mathrm{pau},e_\rho}R_{\mathrm{pau},e_\sigma} =L_{\mathrm{pau},e_\mu e_\rho}R_{\mathrm{pau},e_\sigma e_\nu}, $$ so each basis operator is fixed by its left unit $e_\mu$ and its right unit $e_\nu$, and the four left matrices and the four right matrices generate all sixteen. In blocks of $2\times2$, $$ L_{\mathrm{pau},e_t}=\begin{pmatrix}1&0\\0&1\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&-i\\-i&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&-1\\1&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_z}=\begin{pmatrix}-i&0\\0&i\end{pmatrix}, $$ $$ R_{\mathrm{pau},e_t}=\begin{pmatrix}1&0\\0&1\end{pmatrix},\qquad R_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&-i\\-i&0\end{pmatrix},\qquad R_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&1\\-1&0\end{pmatrix},\qquad R_{\mathrm{pau},\vec e_z}=\begin{pmatrix}-i&0\\0&i\end{pmatrix}, $$ each a $2\times2$ block repeated twice, $I_2\otimes\mathsf{M}_2(e_\mu)=\operatorname{diag}\bigl(\mathsf{M}_2(e_\mu),\mathsf{M}_2(e_\mu)\bigr)$ for the left matrices and $\mathsf{M}_2(e_\nu)^{\mathsf T}\otimes I_2=\operatorname{diag}\bigl(\mathsf{M}_2(e_\nu)^{\mathsf T},\mathsf{M}_2(e_\nu)^{\mathsf T}\bigr)$ for the right ones.
The sixteen basis operators are Kronecker products. Combining the two tensor forms, $$ L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}=\bigl(I_2\otimes\mathsf{M}_2(e_\mu)\bigr)\bigl(\mathsf{M}_2(e_\nu)^{\mathsf T}\otimes I_2\bigr) =\mathsf{M}_2(e_\nu)^{\mathsf T}\otimes\mathsf{M}_2(e_\mu), $$ so each basis operator is a Kronecker product: the right unit's matrix, transposed, in the first slot, and the left unit's matrix in the second. The left factor fixes the second slot and the right factor the first, and the basis operator is the pair $(\mu,\nu)$ read in the two slots:
| basis operator | $\mathsf{M}_2(e_\nu)^{\mathsf T}\otimes \mathsf{M}_2(e_\mu)$ | basis operator | $\mathsf{M}_2(e_\nu)^{\mathsf T}\otimes \mathsf{M}_2(e_\mu)$ |
|---|---|---|---|
| $L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}$ | $I_2\otimes I_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}$ | $I_2\otimes \mathsf{M}_2(\vec e_y)$ |
| $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}$ | $\mathsf{M}_2(\vec e_x)^{\mathsf T}\otimes I_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}$ | $\mathsf{M}_2(\vec e_x)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_y)$ |
| $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}$ | $\mathsf{M}_2(\vec e_y)^{\mathsf T}\otimes I_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}$ | $\mathsf{M}_2(\vec e_y)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_y)$ |
| $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ | $\mathsf{M}_2(\vec e_z)^{\mathsf T}\otimes I_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}$ | $\mathsf{M}_2(\vec e_z)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_y)$ |
| $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}$ | $I_2\otimes \mathsf{M}_2(\vec e_x)$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}$ | $I_2\otimes \mathsf{M}_2(\vec e_z)$ |
| $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}$ | $\mathsf{M}_2(\vec e_x)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_x)$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}$ | $\mathsf{M}_2(\vec e_x)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_z)$ |
| $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}$ | $\mathsf{M}_2(\vec e_y)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_x)$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}$ | $\mathsf{M}_2(\vec e_y)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_z)$ |
| $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}$ | $\mathsf{M}_2(\vec e_z)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_x)$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}$ | $\mathsf{M}_2(\vec e_z)^{\mathsf T}\otimes \mathsf{M}_2(\vec e_z)$ |
All sixteen are invertible, of rank four and of Hilbert–Schmidt norm two, because $\langle X\otimes Y\rangle_*^{1/2}=\langle X\rangle_*^{1/2}\langle Y\rangle_*^{1/2}$ and every $\mathsf{M}_2(e_\mu)$ has norm $\sqrt2$; the only one of trace $4$ is $L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}=I_4$, and the other fifteen have trace $0$.
The sixteen, one by one. Written out in full:
$$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix},\qquad L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&0&-i&0\\0&0&0&-i\\-i&0&0&0\\0&-i&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&0&1&0\\0&0&0&1\\-1&0&0&0\\0&-1&0&0\end{pmatrix},\qquad L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}=\begin{pmatrix}-i&0&0&0\\0&-i&0&0\\0&0&i&0\\0&0&0&i\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}=\begin{pmatrix}0&-i&0&0\\-i&0&0&0\\0&0&0&-i\\0&0&-i&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&0&0&-1\\0&0&-1&0\\0&-1&0&0\\-1&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&0&0&-i\\0&0&-i&0\\0&i&0&0\\i&0&0&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}=\begin{pmatrix}0&-1&0&0\\-1&0&0&0\\0&0&0&1\\0&0&1&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&1&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&0&0&i\\0&0&-i&0\\0&i&0&0\\-i&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&0&0&-1\\0&0&1&0\\0&1&0&0\\-1&0&0&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}=\begin{pmatrix}0&i&0&0\\-i&0&0&0\\0&0&0&-i\\0&0&i&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}=\begin{pmatrix}-i&0&0&0\\0&i&0&0\\0&0&-i&0\\0&0&0&i\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}=\begin{pmatrix}0&0&-1&0\\0&0&0&1\\-1&0&0&0\\0&1&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}=\begin{pmatrix}0&0&-i&0\\0&0&0&i\\i&0&0&0\\0&-i&0&0\end{pmatrix},\qquad L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}=\begin{pmatrix}-1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&-1\end{pmatrix} $$
The Pauli Form
The Pauli form. The matrices $I_2\otimes\mathsf{M}_2(\tilde Q)$ and $\mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2$ are the two tensor forms of the same element. The first reads the element on the left, the second on the right, and the exchange $\mathsf P$ is the difference between them: $$ \mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2=\mathsf P\,\bigl(I_2\otimes\mathsf{M}_2(\tilde Q)^{\mathsf T}\bigr)\mathsf P^{-1} =\mathsf P\,\bigl(I_2\otimes\mathsf{M}_2(\tilde Q)\bigr)^{\mathsf T}\mathsf P^{-1}. $$ The four left matrices, the four right matrices and the sixteen basis operators are written above in this frame; the gamma framing of the same five basis operators, and the comparison with the Clifford algebra, are in §The Clifford Structure and §The Legacy Pauli Gamma Formulation.
The Switch Between the Two Frames
One matrix, and its inverse. The two frames are two bases of the same four-dimensional vector space, so the two displays of one operator are related by one change of matrix. Write $C_{\Pi}$ for the matrix whose columns are the coordinates of the frame vectors in the coefficient basis, $$ C_{\Pi}=\tfrac12\begin{pmatrix}1&0&0&1\\0&i&i&0\\0&1&-1&0\\i&0&0&-i\end{pmatrix}, $$ the columns being $\tilde\Pi_+$, $\vec\Upsilon_+$, $\vec\Upsilon_-$, $\tilde\Pi_-$ in the order $(e_t,\vec e_x,\vec e_y,\vec e_z)$. Then for every element and every side, $$ \boxed{\;L_{\mathrm{axf}}(\tilde Q)=C_{\Pi}\,L(\tilde Q)\,C_{\Pi}^{-1},\qquad R_{\mathrm{axf}}(\tilde Q)=C_{\Pi}\,R(\tilde Q)\,C_{\Pi}^{-1}\;} $$ and inversely $L(\tilde Q)=C_{\Pi}^{-1}L_{\mathrm{axf}}(\tilde Q)C_{\Pi}$, $R(\tilde Q)=C_{\Pi}^{-1}R_{\mathrm{axf}}(\tilde Q)C_{\Pi}$. The reason is one line: the $\nu$-th column of a regular matrix is the coordinate column of a product, and a change of basis rewrites every coordinate column by the same matrix. Neither side changes, and both displays change by the same conjugation.
The inverse. The inverse is $$ C_{\Pi}^{-1}=\begin{pmatrix}1&0&0&-i\\0&-i&1&0\\0&-i&-1&0\\1&0&0&i\end{pmatrix}=2C_{\Pi}^{\dagger}, $$ so $C_{\Pi}^{\dagger}C_{\Pi}=\tfrac12\mathrm I_4$ and the change is a scaled unitary, not a unitary. Its determinant is $-\tfrac14$: the frame vectors are not normalised to modulus one, and no normalisation is imposed, because the frame is the Peirce basis and not an orthonormal basis.
Two $2\times2$ blocks. In $2\times2$ blocks over the splitting $(1,2)\mid(3,4)$, $$ C_{\Pi}=\tfrac12\begin{pmatrix} \begin{pmatrix}1&0\\0&i\end{pmatrix}&\begin{pmatrix}0&1\\i&0\end{pmatrix}\\[2pt] \begin{pmatrix}0&1\\i&0\end{pmatrix}&\begin{pmatrix}-1&0\\0&-i\end{pmatrix} \end{pmatrix}, \qquad C_{\Pi}^{-1}=\begin{pmatrix} \begin{pmatrix}1&0\\0&-i\end{pmatrix}&\begin{pmatrix}0&1\\-i&0\end{pmatrix}\\[2pt] \begin{pmatrix}0&1\\-i&0\end{pmatrix}&\begin{pmatrix}-1&0\\0&i\end{pmatrix} \end{pmatrix}, $$ and the two off-diagonal blocks are equal in each matrix: the switch does not mix the two Peirce components independently, it pairs them symmetrically. The block form is what makes the switch two $2\times2$ computations and not one $4\times4$ one.
The eight unit matrices. Apply the switch to each unit matrix, using $C_{\Pi}L_{\mathrm{axf},e_\mu}C_{\Pi}^{-1}=L_{\mathrm{pau},e_\mu}$ and $C_{\Pi}R_{\mathrm{axf},e_\mu}C_{\Pi}^{-1}=R_{\mathrm{pau},e_\mu}$. In the axis-free frame the four left matrices are real signed permutations and the four right ones are real signed permutations, and after the switch, $$ L_{\mathrm{axf},e_t}\mapsto L_{\mathrm{pau},e_t}=\mathrm I_4,\qquad L_{\mathrm{axf},\vec e_x}\mapsto L_{\mathrm{pau},\vec e_x},\qquad L_{\mathrm{axf},\vec e_y}\mapsto L_{\mathrm{pau},\vec e_y},\qquad L_{\mathrm{axf},\vec e_z}\mapsto L_{\mathrm{pau},\vec e_z}, $$ $$ R_{\mathrm{axf},e_t}\mapsto R_{\mathrm{pau},e_t}=\mathrm I_4,\qquad R_{\mathrm{axf},\vec e_x}\mapsto R_{\mathrm{pau},\vec e_x},\qquad R_{\mathrm{axf},\vec e_y}\mapsto R_{\mathrm{pau},\vec e_y},\qquad R_{\mathrm{axf},\vec e_z}\mapsto R_{\mathrm{pau},\vec e_z}, $$ with the eight displays of §The Two Tensor Factors. The identity is the identity in both frames, and the seven others change. The two families agree on the identity and nowhere else, as they must: $L_{\mathrm{axf}}(\tilde Q)=R_{\mathrm{axf}}(\tilde Q)$ forces $\tilde Q\in\mathbb{C}e_t$.
The sixteen basis operators, one by one. The switch carries the basis operators one by one, and each basis operator of the Pauli frame is the conjugate of the basis operator of the same index in the axis-free frame. The sixteen expressions follow, and in each the Pauli basis operator stands on the left, the conjugation of the axis-free basis operator stands in the middle, and the computed Pauli array stands on the right: $$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}=C_{\Pi}^{-1}\,L_{\mathrm{axf},e_t}R_{\mathrm{axf},e_t}\,C_{\Pi}=\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix} $$
$$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}=C_{\Pi}^{-1}\,L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_x}\,C_{\Pi}=\begin{pmatrix}0&0&-i&0\\0&0&0&-i\\-i&0&0&0\\0&-i&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}=C_{\Pi}^{-1}\,L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_y}\,C_{\Pi}=\begin{pmatrix}0&0&1&0\\0&0&0&1\\-1&0&0&0\\0&-1&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}=C_{\Pi}^{-1}\,L_{\mathrm{axf},e_t}R_{\mathrm{axf},\vec e_z}\,C_{\Pi}=\begin{pmatrix}-i&0&0&0\\0&-i&0&0\\0&0&i&0\\0&0&0&i\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},e_t}\,C_{\Pi}=\begin{pmatrix}0&-i&0&0\\-i&0&0&0\\0&0&0&-i\\0&0&-i&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_x}\,C_{\Pi}=\begin{pmatrix}0&0&0&-1\\0&0&-1&0\\0&-1&0&0\\-1&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_y}\,C_{\Pi}=\begin{pmatrix}0&0&0&-i\\0&0&-i&0\\0&i&0&0\\i&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_z}\,C_{\Pi}=\begin{pmatrix}0&-1&0&0\\-1&0&0&0\\0&0&0&1\\0&0&1&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},e_t}\,C_{\Pi}=\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&-1\\0&0&1&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_x}\,C_{\Pi}=\begin{pmatrix}0&0&0&i\\0&0&-i&0\\0&i&0&0\\-i&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_y}\,C_{\Pi}=\begin{pmatrix}0&0&0&-1\\0&0&1&0\\0&1&0&0\\-1&0&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_z}\,C_{\Pi}=\begin{pmatrix}0&i&0&0\\-i&0&0&0\\0&0&0&-i\\0&0&i&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},e_t}\,C_{\Pi}=\begin{pmatrix}-i&0&0&0\\0&i&0&0\\0&0&-i&0\\0&0&0&i\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_x}\,C_{\Pi}=\begin{pmatrix}0&0&-1&0\\0&0&0&1\\-1&0&0&0\\0&1&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_y}\,C_{\Pi}=\begin{pmatrix}0&0&-i&0\\0&0&0&i\\i&0&0&0\\0&-i&0&0\end{pmatrix} $$
$$ L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}=C_{\Pi}^{-1}\,L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_z}\,C_{\Pi}=\begin{pmatrix}-1&0&0&0\\0&1&0&0\\0&0&1&0\\0&0&0&-1\end{pmatrix} $$
The sixteen equalities were checked by multiplying the three matrices of each line; the middle expression is the axis-free display of §The Sixteen Products in This Frame conjugated by $C_{\Pi}^{-1}$ on the left and $C_{\Pi}$ on the right, and the right-hand array is the Pauli display of §The Sixteen Basis Operators in the Peirce Frame. The pair $(\mu,\nu)$ is the same operator on both sides of each line, $e_\mu[\,]e_\nu$, so the list is a dictionary of arrays and not a table of operators. Two of the sixteen have the same array in both frames, $L_{\mathrm{axf},e_t}R_{\mathrm{axf},e_t}=L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}=\mathrm I_4$ and $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_z}=L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}=\operatorname{diag}(-1,1,1,-1)$, two more differ by the single sign $L_{\mathrm{axf}}=-L_{\mathrm{pau}}$ at $(t,y)$ and $(z,x)$, and the other twelve differ in their entries. The explanation is commutation: $L_{e_t}R_{e_t}=\mathrm I_4$ commutes with everything, and $L_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_z}=\operatorname{diag}(-1,1,1,-1)$ commutes with $C_{\Pi}$, since both fix the two $2\times2$ blocks that $C_{\Pi}$ pairs. The two basis operators at $(t,y)$ and $(z,x)$ are the ones that anticommute with $C_{\Pi}$, which is why their two displays differ by $-1$ and not in their entries.
A worked element. Take $\tilde Q=e_t+\vec e_x$, so that $Q_t=Q_x=1$ and $Q_y=Q_z=0$. In the axis-free frame, $$ L_{\mathrm{axf}}(e_t+\vec e_x)=L_{\mathrm{axf},e_t}+L_{\mathrm{axf},\vec e_x} =\begin{pmatrix}1&-1&0&0\\1&1&0&0\\0&0&1&-1\\0&0&1&1\end{pmatrix}, $$ and in the Peirce frame the same element reads $$ L(e_t+\vec e_x)=\mathsf{M}_{4L}(e_t)+\mathsf{M}_{4L}(\vec e_x)=\begin{pmatrix}1&-i&0&0\\-i&1&0&0\\0&0&1&-i\\0&0&-i&1\end{pmatrix}, $$ which is the block $\mathsf{M}_2(e_t+\vec e_x)=\begin{pmatrix}1&-i\\-i&1\end{pmatrix}$ written twice. The switch is checked by multiplying, $L= C_{\Pi}^{-1}L_{\mathrm{axf}}C_{\Pi}$, and it holds entry by entry. The right displays of the same element are obtained the same way, $R(e_t+\vec e_x)=\mathsf{P}\,L(e_t+\vec e_x)^{\mathsf T}\mathsf{P}^{-1}$.
What the switch preserves, and what it does not. It preserves the trace, the determinant, the spectrum, the ranks of all powers, and the algebra of the sixteen basis operators: it is a conjugation, so it is an isomorphism of the ambient $M_4(\mathbb{C})$ carrying $\mathbb{B}_L$ onto $\mathbb{B}_L$ and $\mathbb{B}_R$ onto $\mathbb{B}_R$. It does not preserve the transpose dictionary, the reality of the sixteen arrays, the vanishing of the cross blocks, or the property that a matrix is a signed permutation. Those four are properties of the axis-free frame; the vanishing of the cross blocks and the tensor form are properties of the Peirce frame.
Where the $2\times2$ display sits. The switch moves the $4\times4$ displays and leaves the $2\times2$ image alone: $\mathsf{M}_2(\tilde Q)$ appears as a block before the switch and after it, and it never meets $C_{\Pi}$. §The $2\times2$ Pauli Representation, §The Bridge to the $4\times4$ Displays states the consequence.
The Clifford Structure
The Clifford algebra is the basis operator algebra. The sixteen basis operators $L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}$ are linearly independent, and their complex span is all of $M_4(\mathbb{C})$. They multiply by multiplying the left multipliers together and the right multipliers together, $$ L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}\,L_{\mathrm{pau},e_\rho}R_{\mathrm{pau},e_\sigma} =L_{\mathrm{pau},e_\mu e_\rho}R_{\mathrm{pau},e_\sigma e_\nu}; $$ the result is again a left multiplication and a right multiplication, so the assignment $e_\mu\otimes e_{\mathrm{opp},\nu}\mapsto L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}$ is an algebra isomorphism $$ \mathbb{B}\otimes_{\mathbb{C}}\mathbb{B}_{\mathrm{opp}}\longrightarrow M_4(\mathbb{C})\cong\mathrm{Cl}_{1,3}(\mathbb{C}). $$ The sixteen products of the five frame operators span over $\mathbb{R}$ the algebra $\mathrm{Cl}_{1,3}(\mathbb{R})\cong M_2(\mathbb{H})$, of real dimension $16$, and the sixteen basis operators span $\mathrm{Cl}_{3,1}(\mathbb{R})\cong M_4(\mathbb{R})$, the opposite real form. The Dirac, Weyl and Majorana matrices are three presentations of the same complex algebra, related by a change of frame; a change of frame is a change of basis, not a change of object.
The five basis operators of the frame. The five frame operators are $$ \boxed{\;iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x},\qquad iL_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y},\qquad iL_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y},\qquad iL_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y},\qquad -iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}\;} $$ Each is an invertible multiple of a basis operator; the left factor is invertible and the right factor a spatial unit, written $\vec e_{\nu(\mu)}$ with $\nu(t)=x$ and $\nu(k)=y$. For the first and the last the left factor is $e_t$, so the basis operator collapses to a single right multiplication, $iR_{\mathrm{pau},\vec e_x}$ and $-iR_{\mathrm{pau},\vec e_z}$, and the two spellings are the same operator. In the Peirce frame the five are monomial matrices, written out in §The Legacy Pauli Gamma Formulation; in the axis-free frame they are $i$ times real signed permutation matrices, displayed there too.
The anticommutation relations. $$ \{iL_{\mathrm{pau},e_\mu}R_{\mathrm{pau},\vec e_{\nu(\mu)}},iL_{\mathrm{pau},e_\nu}R_{\mathrm{pau},\vec e_{\nu(\nu)}}\}=2\varepsilon_\mu\delta_{\mu\nu}I_4,\qquad (iL_{\mathrm{pau},\vec e_k}R_{\mathrm{pau},\vec e_y})^2=-I_4, $$ $$ -iR_{\mathrm{pau},\vec e_z}=i\,(iR_{\mathrm{pau},\vec e_x})(iL_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y})(iL_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y})(iL_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}), $$ with $\varepsilon=(1,-1,-1,-1)$ the plain weight. No index is raised or lowered, and a contravariant index is not allowed: the metric is the unit metric $\delta_{\mu\nu}=(1,1,1,1)$, which is the identity, and the sign is carried by the weight inside the product.
The sixteen products are the sixteen basis operators. The five frame operators are five basis operators up to a factor $i$, and the sixteen products of the five are the sixteen basis operators:
| grade | basis operators |
|---|---|
| $0$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}$ |
| $1$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}$, $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}$, $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}$, $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}$ |
| $2$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}$, $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}$, $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}$, $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}$, $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}$, $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}$ |
| $3$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}$, $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}$, $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}$, $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}$ |
| $4$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ |
Each row lists the basis operators of one grade, and the counts are $1+4+6+4+1=16$. The frame is one frame among many: another orthogonal pair $(\beta,\delta)$ gives another five basis operators, the relations survive, and the basis operator labels do not. The proof, and the freedom in the choice of frame, are in Operators Acting on Two Sides, §Composition and the Two Closures.
The opposite algebra is written with three letters. The opposite of an algebra, with the product reversed, carries the superscript $\mathrm{opp}$: the reversed product is $a\cdot_{\mathrm{opp}}b=ba$, and the opposite of $\mathbb{B}$ is $\mathbb{B}_{\mathrm{opp}}$. The two-letter form $\mathrm{op}$ is not used at all, neither for the opposite algebra nor for the operator norm, which is written $\lVert A\rVert$ (Index and Notation Conventions, §The Labels). The competing norms keep their own labels, $\langle A\rangle_*^{1/2}$, $\lVert A\rVert_{\mathrm{nuc}}$, $\langle A\rangle_*^{1/2}$, $\lVert A\rVert_*$.
The comparison map. The $\mathbb{R}$-linear map $\iota_{\mathrm{Cl}}$ of the Dirac articles is read in basis operators as $\iota_{\mathrm{Cl}}(e_\mu)=L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_t}$ and $\iota_{\mathrm{Cl}}(i)=L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$. Its image is the real span of the eight basis operators $L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_t}$ and $L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},\vec e_z}$, of real dimension $8$, and it satisfies $\iota_{\mathrm{Cl}}(i\tilde Q)=L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}\,\iota_{\mathrm{Cl}}(\tilde Q)$. It is $\mathbb{R}$-linear and not $\mathbb{C}$-linear: it is the comparison with the even subalgebra of $\mathrm{Cl}_{1,3}$, not a second representation of $\mathbb{B}$. In the axis-free frame the same map reads $\iota_{\mathrm{Cl}}(e_\mu)=L_{\mathrm{axf},e_\mu}R_{\mathrm{axf},e_t}$, because the conjugation carries the whole basis operator algebra with it.
The slash is not used. The Dirac slash is not part of the notation of the matrix models. In the Pauli frame the momentum contraction is a basis operator sum, $$ \sum_\mu iL_{\mathrm{pau},e_\mu}R_{\mathrm{pau},\vec e_{\nu(\mu)}}p_\mu =i\bigl(L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}p_t+L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}p_x+L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}p_y+L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}p_z\bigr), $$ so the frame is visible in the formula.
The Legacy Pauli Gamma Formulation
The Pauli and gamma formulation is legacy: the reasons are explained in the article Why AXF Matrices instead of Pauli and Gamma Matrices. The generators carry no frame label: $\gamma_\mu$ is one set of matrices, $\gamma_0^2=I_4$ and $\gamma_k^2=-I_4$. The frame belongs to the coordinate, $Q_\mu$ or $x_\mu$, not to the array. The fifth is written $\gamma_5$ with no bracket and no frame, because it is the product of the four and not the image of a direction.
The gammas are the five-operator frame as matrices. The body writes the basis operators; this section writes the arrays, in both frames.
The five gammas. The five Dirac matrices are two-sided operators of the frame. The fifth is the fifth frame operator, written $\mathsf{V}_{\mathrm{vol}}$ where the operator and not its array is meant, with the frame named in the sentence; the articles on the sides use that glyph and cite the gamma display for the arrays. $$ \boxed{\;\gamma_0=iR_{\mathrm{pau},\vec e_x}=iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x},\qquad \gamma_1=iL_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y},\qquad \gamma_2=iL_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y},\qquad \gamma_3=iL_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y},\qquad \gamma_5=-iR_{\mathrm{pau},\vec e_z}=-iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}\;} $$
In the Peirce frame. $$ \gamma_0=\begin{pmatrix}0&0&1&0\\0&0&0&1\\1&0&0&0\\0&1&0&0\end{pmatrix},\qquad \gamma_1=\begin{pmatrix}0&0&0&1\\0&0&1&0\\0&-1&0&0\\-1&0&0&0\end{pmatrix}, $$ $$ \gamma_2=\begin{pmatrix}0&0&0&-i\\0&0&i&0\\0&i&0&0\\-i&0&0&0\end{pmatrix},\qquad \gamma_3=\begin{pmatrix}0&0&1&0\\0&0&0&-1\\-1&0&0&0\\0&1&0&0\end{pmatrix},\qquad \gamma_5=\begin{pmatrix}-1&0&0&0\\0&-1&0&0\\0&0&1&0\\0&0&0&1\end{pmatrix}. $$ They satisfy $\{\gamma_\mu,\gamma_\nu\}=2\varepsilon_\mu\delta_{\mu\nu}I_4$, $(\gamma_k)^2=-I_4$ and $\gamma_5=i\gamma_0\gamma_1\gamma_2\gamma_3$. In the Pauli form the same five are $$ \gamma_0=\sigma_1\otimes I_2,\qquad \gamma_k=i\,\sigma_2\otimes\sigma_k,\qquad \gamma_5=-\sigma_3\otimes I_2 . $$
In the axis-free frame. The conjugation §The Switch Between the Two Frames carries them, and there they are $i$ times real signed permutation matrices, $$ \gamma_0=i\begin{pmatrix}0&-1&0&0\\1&0&0&0\\0&0&0&1\\0&0&-1&0\end{pmatrix},\qquad \gamma_1=i\begin{pmatrix}0&0&0&1\\0&0&-1&0\\0&-1&0&0\\1&0&0&0\end{pmatrix}, $$ $$ \gamma_2=i\begin{pmatrix}-1&0&0&0\\0&1&0&0\\0&0&-1&0\\0&0&0&1\end{pmatrix},\qquad \gamma_3=i\begin{pmatrix}0&-1&0&0\\-1&0&0&0\\0&0&0&-1\\0&0&-1&0\end{pmatrix},\qquad \gamma_5=i\begin{pmatrix}0&0&0&1\\0&0&-1&0\\0&1&0&0\\-1&0&0&0\end{pmatrix}. $$ The entries are $\pm i$ or $0$ in both frames, and the axis-free arrays are exactly the basis operators $iR_{\mathrm{axf},\vec e_x}$, $iL_{\mathrm{axf},\vec e_x}R_{\mathrm{axf},\vec e_y}$, $iL_{\mathrm{axf},\vec e_y}R_{\mathrm{axf},\vec e_y}$, $iL_{\mathrm{axf},\vec e_z}R_{\mathrm{axf},\vec e_y}$ and $-iR_{\mathrm{axf},\vec e_z}$: the same five basis operators, drawn in the other frame.
The sixteen products. The five frame operators are five of the sixteen products up to a factor $i$, and their products in increasing grade are the sixteen basis operators:
| grade | $\gamma$-product | basis operator | equality |
|---|---|---|---|
| $0$ | $1$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},e_t}=I_4$ |
| $1$ | $\gamma_0$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}=-i\,\gamma_0$ |
| $1$ | $\gamma_1$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}=-i\,\gamma_1$ |
| $1$ | $\gamma_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}=-i\,\gamma_2$ |
| $1$ | $\gamma_3$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}=-i\,\gamma_3$ |
| $2$ | $\gamma_0\gamma_1$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_z}=\gamma_0\gamma_1$ |
| $2$ | $\gamma_0\gamma_2$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_z}=\gamma_0\gamma_2$ |
| $2$ | $\gamma_0\gamma_3$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_z}=\gamma_0\gamma_3$ |
| $2$ | $\gamma_1\gamma_2$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},e_t}=\gamma_1\gamma_2$ |
| $2$ | $\gamma_1\gamma_3$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},e_t}=-\gamma_1\gamma_3$ |
| $2$ | $\gamma_2\gamma_3$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},e_t}=\gamma_2\gamma_3$ |
| $3$ | $\gamma_0\gamma_1\gamma_2$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}$ | $L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_x}=-i\,\gamma_0\gamma_1\gamma_2$ |
| $3$ | $\gamma_0\gamma_1\gamma_3$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}$ | $L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_x}=i\,\gamma_0\gamma_1\gamma_3$ |
| $3$ | $\gamma_0\gamma_2\gamma_3$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}$ | $L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_x}=-i\,\gamma_0\gamma_2\gamma_3$ |
| $3$ | $\gamma_1\gamma_2\gamma_3$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_y}=i\,\gamma_1\gamma_2\gamma_3$ |
| $4$ | $\gamma_0\gamma_1\gamma_2\gamma_3$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ | $L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}=i\,\gamma_5=-\gamma_0\gamma_1\gamma_2\gamma_3$ |
Each row reads the basis operator as a unit coefficient times the $\gamma$-product; the inverse reading carries the conjugate coefficient, since $c\in\{1,-1,i,-i\}$.
The comparison map. In gammas, $\iota_{\mathrm{Cl}}(\vec e_x)=\gamma_2\gamma_3$, $\iota_{\mathrm{Cl}}(\vec e_y)=\gamma_3\gamma_1$, $\iota_{\mathrm{Cl}}(\vec e_z)=\gamma_1\gamma_2$ and $\iota_{\mathrm{Cl}}(i)=-\gamma_0\gamma_1\gamma_2\gamma_3$, the same map as $\iota_{\mathrm{Cl}}(e_\mu)=L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_t}$ and $\iota_{\mathrm{Cl}}(i)=L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ of §The Clifford Structure.
The contractions. In gamma notation a momentum contraction is $\sum_\mu \gamma_\mu p_\mu=i(L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}p_t+L_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}p_x+L_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}p_y+L_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}p_z)$, a derivative contraction is $i\sum_\mu \gamma_\mu\partial/\partial x_\mu$ with $x_0=q'_t$, and the same contraction in the $\mathbb{B}$ coordinate is $\sum_\mu \gamma_\mu\partial/\partial Q_\mu$. The two differ by the coordinate, $Q_t=ix_0$, and not by the matrices: $\partial/\partial x_\mu$ and $\partial/\partial Q_\mu$ already name the frame, so the array needs no label to say which is meant.
The identities the bodies cite. In the $\gamma$ reading, with $\varepsilon=(1,-1,-1,-1)$ the sign vector of the basis operators, $$ \{\gamma_\mu,\gamma_\nu\}=2\varepsilon_\mu\delta_{\mu\nu}I_4,\qquad (\gamma_k)^2=-I_4,\qquad \gamma_5^2=I_4,\qquad \{\gamma_\mu,\gamma_5\}=0, $$ $$ [\gamma_\mu,\gamma_\nu]=2\gamma_\mu\gamma_\nu\quad(\mu\ne\nu),\qquad [\gamma_a,S_{cd}]=\varepsilon_a\delta_{ac}\gamma_d-\varepsilon_a\delta_{ad}\gamma_c,\qquad S_{cd}=\tfrac14[\gamma_c,\gamma_d], $$ $$ \mathrm{tr}(\gamma_\mu\gamma_\nu)=4\varepsilon_\mu\delta_{\mu\nu},\qquad \mathrm{tr}(\gamma_5\gamma_\mu\gamma_\nu\gamma_\rho\gamma_\sigma)=-4i\,\epsilon_{\mu\nu\rho\sigma}, $$ $$ \mathrm{tr}(\gamma_\mu\gamma_\nu\gamma_\rho\gamma_\sigma) =4\bigl(\varepsilon_\mu\delta_{\mu\nu}\varepsilon_\rho\delta_{\rho\sigma} -\varepsilon_\mu\delta_{\mu\rho}\varepsilon_\nu\delta_{\nu\sigma} +\varepsilon_\mu\delta_{\mu\sigma}\varepsilon_\nu\delta_{\nu\rho}\bigr). $$ The contraction sums follow from the squares. No index is raised: the summed index carries the weight $\varepsilon_\mu$, so that $\sum_\mu\varepsilon_\mu\gamma_\mu\gamma_\mu=4I_4$, $\sum_\mu\varepsilon_\mu\gamma_\mu\gamma_\alpha\gamma_\mu=-2\gamma_\alpha$, $\sum_\mu\varepsilon_\mu\gamma_\mu\gamma_\alpha\gamma_\beta\gamma_\mu=4\varepsilon_\alpha\delta_{\alpha\beta}I_4$ and $\sum_\mu\varepsilon_\mu\gamma_\mu\gamma_\gamma\gamma_\beta\gamma_\alpha\gamma_\mu=2\gamma_\gamma\gamma_\beta\gamma_\alpha$.
The $2\times2$ Pauli Representation
Read last, and not because it is small. The $2\times2$ display is the algebra itself, not an operator on it. It is placed last because it is best read after both $4\times4$ displays have shown what a matrix of an operator is, and because its matrices are the block of the Pauli display: the object it displays is the one the other two displays act on.
The Isomorphism
The map. The isomorphism is $\mathbb{C}$-linear, multiplicative and bijective, $$ \mathsf{M}_2:\mathbb{B}\longrightarrow M_2(\mathbb{C}),\qquad \mathsf{M}_2(\tilde Q_1+\tilde Q_2)=\mathsf{M}_2(\tilde Q_1)+\mathsf{M}_2(\tilde Q_2),\qquad \mathsf{M}_2(\tilde Q_1\tilde Q_2)=\mathsf{M}_2(\tilde Q_1)\mathsf{M}_2(\tilde Q_2), $$ and it is the reading every numerical check in the corpus is performed with.
The four unit images. The images of the invertibles are fixed once, because every article depends on them agreeing: $$ \mathsf{M}_2(e_t)=I_2,\qquad \mathsf{M}_2(\vec e_x)=\begin{pmatrix}0&-i\\-i&0\end{pmatrix}=-i\sigma_1,\qquad \mathsf{M}_2(\vec e_y)=\begin{pmatrix}0&-1\\1&0\end{pmatrix}=-i\sigma_2,\qquad \mathsf{M}_2(\vec e_z)=\begin{pmatrix}-i&0\\0&i\end{pmatrix}=-i\sigma_3, $$ with $\sigma_1,\sigma_2,\sigma_3$ the Pauli matrices of §The Two Tensor Factors. The central scalar is $\mathsf{M}_2(i)=iI_2$, and $\mathbb{C}$-linearity gives the Hermitian units, $\mathsf{M}_2(i\vec e_k)=i\,\mathsf{M}_2(\vec e_k)$. A general element is $$ \mathsf{M}_2(\tilde Q)= \begin{pmatrix} Q_t-iQ_z&-iQ_x-Q_y\\ -iQ_x+Q_y&Q_t+iQ_z \end{pmatrix} =\begin{pmatrix}a&b\\ c&d\end{pmatrix}, $$ which is the $a,b,c,d$ of §The Two Tensor Factors, and the reason the Pauli display has that block.
The assignment is forced. $\vec e_k^2=-e_t$ requires the factor $i$, and $\vec e_x\vec e_y=\vec e_z$ fixes the sign. The three images cannot be made real, nor all three negated; Introduction to the $2\times2$ Matrix Element Representation $M_2(\mathbb{C})_{\mathbb{H}}$ of Quaternions derives both.
No frame is needed. The isomorphism is a map of algebras and does not depend on a frame of $\mathbb{B}$. The four matrices above are the images of the units, not the entries of a matrix read in a basis, and no Peirce family enters. The frame enters only in the two $4\times4$ displays, which read operators on the algebra.
The invariants. The determinant of the image is the biquaternion norm and its trace is twice the scalar part, $$ \det\mathsf{M}_2(\tilde Q)=\langle\tilde Q\rangle_{\natural},\qquad \operatorname{tr}\mathsf{M}_2(\tilde Q)=2Q_t . $$ The trace in that display is the trace of the image, and the corpus uses the word for the image alone: an element is read in $\operatorname{Sc}$. The determinant is multiplicative, which is what makes the matrix model compute every norm in the corpus; the invertibility criterion, the group of invertibles and the unit-norm group are read from it.
The Bridge to the $4\times4$ Displays
The block. In the Pauli frame the left regular matrix is $I_2\otimes\mathsf{M}_2(\tilde Q)$, so the $2\times2$ image is the block of the $4\times4$ display, and the right regular matrix is $\mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2$, the same block transposed in the other slot. This is the whole reason the Peirce frame earns its root: it turns a $4\times4$ operator into a $2\times2$ matrix written twice.
The frame moves the block, not the image. In the axis-free frame the same operator reads $$ L_{\mathrm{axf}}(\tilde Q)=C_{\Pi}\,\bigl(I_2\otimes\mathsf{M}_2(\tilde Q)\bigr)\,C_{\Pi}^{-1}, $$ so the axis-free display is the conjugated block. The $2\times2$ image $\mathsf{M}_2(\tilde Q)$ is untouched by $C_{\Pi}$: a change of frame conjugates the $4\times4$ displays and leaves the $2\times2$ display alone. That asymmetry is what separates the two kinds of model in §The Matrix Models.
The determinant squares, the trace doubles. From the block form, $\det L(\tilde Q)=\det\mathsf{M}_2(\tilde Q)^2=\langle\tilde Q\rangle_{\natural}^2$ and $\operatorname{tr}L(\tilde Q)=2\operatorname{tr}\mathsf{M}_2(\tilde Q)=4Q_t$, in either frame. The two invariants of the operator are the invariants of the element, counted twice.
Summary
The algebra is displayed in three ways, in the order in which the displays are built.
The axis-free representation comes first. Its basis is the tensor basis $(e_t,\vec e_x,\vec e_y,\vec e_z)$, its product is the multiplication table, and its two matrices are read off that table by columns: $$ L_{\mathrm{axf}}(\tilde Q)= \begin{pmatrix} Q_t&-Q_x&-Q_y&-Q_z\\ Q_x&Q_t&-Q_z&Q_y\\ Q_y&Q_z&Q_t&-Q_x\\ Q_z&-Q_y&Q_x&Q_t \end{pmatrix}, \qquad R_{\mathrm{axf}}(\tilde Q)= \begin{pmatrix} Q_t&-Q_x&-Q_y&-Q_z\\ Q_x&Q_t&Q_z&-Q_y\\ Q_y&-Q_z&Q_t&Q_x\\ Q_z&Q_y&-Q_x&Q_t \end{pmatrix}. $$ Its eight unit matrices are real signed permutations, its stitch is $R_{\mathrm{axf}}=D(L_{\mathrm{axf}})^{\mathsf T}D$ with $D=\operatorname{diag}(1,-1,-1,-1)$, its transpose is the conjugation, and its sixteen products are real. Nothing is chosen but the basis the algebra already carries.
The Pauli representation comes second. Its frame is the Peirce basis $(\tilde\Pi_+,\vec\Upsilon_+,\vec\Upsilon_-,\tilde\Pi_-)$ of the standard pair, and in it the left matrix is $I_2\otimes\mathsf{M}_2(\tilde Q)$ and the right one $\mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2=\mathsf P L(\tilde Q)^{\mathsf T}\mathsf P^{-1}$. The four left matrices $L_{\mathrm{pau},e_\mu}$, the four right matrices $R_{\mathrm{pau},e_\nu}$ and the sixteen basis operators $L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}$ are written out, and the basis operators are the Clifford algebra: $e_\mu\otimes e_{\mathrm{opp},\nu}\mapsto L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}$ is an isomorphism onto $M_4(\mathbb{C})\cong\mathrm{Cl}_{1,3}(\mathbb{C})$, the sixteen products of the five frame operators span $\mathrm{Cl}_{1,3}(\mathbb{R})$ and the sixteen basis operators span $\mathrm{Cl}_{3,1}(\mathbb{R})$. Five basis operators are the gamma matrices, written in this frame and in the axis-free one. The basis operators are the matrices of the sixteen basis two-sided operators $e_\mu[\,]e_\nu$, so one table of them serves the two-sided family, the enveloping algebra and the Clifford algebra at once.
The switch comes between them, and belongs to both. It is $C_{\Pi}$ and its inverse, of determinant $-\tfrac14$ and normalisation $C_{\Pi}^{\dagger}C_{\Pi}=\tfrac12\mathrm I_4$, with $L_{\mathrm{axf}}=C_{\Pi}LC_{\Pi}^{-1}$ and $R_{\mathrm{axf}}=C_{\Pi}RC_{\Pi}^{-1}$. It is two $2\times2$ blocks, it carries the eight unit matrices and the sixteen basis operators, and it leaves the trace $4Q_t$ and the determinant $\langle\tilde Q\rangle_{\natural}^2$ alone.
The $2\times2$ isomorphism comes last. It is $\mathsf{M}_2(e_t)=I_2$, $\mathsf{M}_2(\vec e_k)=-i\sigma_k$ and $\mathsf{M}_2(i)=iI_2$, forced by $\vec e_k^2=-e_t$ and $\vec e_x\vec e_y=\vec e_z$, with $\det\mathsf{M}_2(\tilde Q)=\langle\tilde Q\rangle_{\natural}$ and $\operatorname{tr}\mathsf{M}_2(\tilde Q)=2Q_t$. It needs no frame, and it is the block of the Pauli display: the change of frame moves the block and never the image.
The subscript names the copy, and a bare $\mathsf{M}_4$ is not written. A regular matrix carries the family letter $L$ or $R$ in its subscript and the frame label $\mathrm{axf}$ or $\mathrm{pau}$ in its caret.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\rho_L,\rho_R$ | Left and right multiplication as linear maps |
| $\mathbb{B}_L=M_4(\mathbb{C})_{\mathbb{B}_L}$ | The left copy, $\rho_L(\mathbb{B})$; $\mathbb{B}_R$ is its commutant |
| $L_{\mathrm{axf}}(\tilde Q)$, $L_{\mathrm{axf},e_\mu}$ | The left regular matrix in the axis-free frame, and the matrix of the unit $e_\mu$ |
| $R_{\mathrm{axf}}(\tilde Q)$, $R_{\mathrm{axf},e_\mu}$ | The right regular matrix in the axis-free frame |
| $D=\operatorname{diag}(1,-1,-1,-1)$ | The stitch of the axis-free frame, $R_{\mathrm{axf}}=D(L_{\mathrm{axf}})^{\mathsf T}D$; the Gram matrix of $\operatorname{Sc}(XY)$ there |
| $L_{\mathrm{pau}}(\tilde Q)=I_2\otimes\mathsf{M}_2(\tilde Q)$ | The left regular matrix in the Peirce frame, written $L(\tilde Q)$ |
| $R(\tilde Q)=\mathsf{M}_2(\tilde Q)^{\mathsf T}\otimes I_2=\mathsf P L(\tilde Q)^{\mathsf T}\mathsf P^{-1}$ | The right regular matrix in the Peirce frame |
| $\mathsf P$ | The exchange of $\vec\Upsilon_+$ and $\vec\Upsilon_-$ fixing $\tilde\Pi_\pm$; the swap of the second and third canonical coordinates, $\operatorname{diag}(1,1,-1,1)$ in the coefficient basis |
| $L_{\mathrm{pau},e_\mu}=\mathsf M_{4L}(e_\mu)$, $R_{\mathrm{pau},e_\mu}=\mathsf M_{4R}(e_\mu)$ | The four left and the four right matrices; $R_{\mathrm{pau},e_\nu}=\mathsf P {L_{\mathrm{pau},e_\nu}}^{\mathsf T}\mathsf P^{-1}$ |
| $L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_\nu}$ | The sixteen basis operators; their complex span is $M_4(\mathbb{C})\cong\mathrm{Cl}_{1,3}(\mathbb{C})$ |
| $L_{e_\mu}R_{e_\nu}=e_\mu[\,]e_\nu$ | The basis operator, which is the matrix of the basis two-sided operator $X\mapsto e_\mu Xe_\nu$ |
| $c(T)$, $\ell(T)=\operatorname{rank}c(T)$ | The coefficient array of an operator $T$ in the sixteen basis operators, and its length; $c(L_{\tilde Q}R_{\tilde Q_2})=QQ_2^{\mathsf T}$ |
| $C_{\Pi}$ | The change of frame, $L_{\mathrm{axf}}=C_{\Pi}LC_{\Pi}^{-1}$ and $R_{\mathrm{axf}}=C_{\Pi}RC_{\Pi}^{-1}$; $\det C_{\Pi}=-\tfrac14$, $C_{\Pi}^{-1}=2C_{\Pi}^{\dagger}$ |
| $\tilde\Pi_\pm$, $\vec\Upsilon_\pm$ | The standard pair and its two mixed units, which fix the Peirce frame |
| $iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_x}$, $iL_{\mathrm{pau},\vec e_x}R_{\mathrm{pau},\vec e_y}$, $iL_{\mathrm{pau},\vec e_y}R_{\mathrm{pau},\vec e_y}$, $iL_{\mathrm{pau},\vec e_z}R_{\mathrm{pau},\vec e_y}$, $-iL_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ | The five basis operators of the frame, timelike, three spatial, chirality; the gammas |
| $\iota_{\mathrm{Cl}}$ | The $\mathbb{R}$-linear comparison with the even subalgebra; $\iota_{\mathrm{Cl}}(e_\mu)=L_{\mathrm{pau},e_\mu}R_{\mathrm{pau},e_t}$, $\iota_{\mathrm{Cl}}(i)=L_{\mathrm{pau},e_t}R_{\mathrm{pau},\vec e_z}$ |
| $\mathsf{M}_2:\mathbb{B}\to M_2(\mathbb{C})$ | The isomorphism of the element; $\mathsf{M}_2(e_t)=I_2$, $\mathsf{M}_2(\vec e_k)=-i\sigma_k$ |
| $2\operatorname{Sc}\tilde Q=\operatorname{tr}\mathsf{M}_2(\tilde Q)$ | Twice the scalar part equals the matrix trace of the image |
| $\det\mathsf{M}_2(\tilde Q)=\langle\tilde Q\rangle_{\natural}$ | The determinant, the biquaternion norm |
Further Reading
- Why AXF Matrices instead of Pauli and Gamma Matrices, for the reasons the AXF matrices replace the Pauli and gamma matrices.
- S. J. Sangwine, T. A. Ell and N. Le Bihan, "Fundamental representations and algebraic properties of biquaternions or complexified quaternions", Advances in Applied Clifford Algebras 21 (2011) 607–636. The biquaternion conjugations and the conventions of the complexified algebra in the applied literature.
- Soo-Chang Pei, Ja-Han Chang and Jian-Jiun Ding, "Commutative reduced biquaternions and their Fourier transform for signal and image processing applications", IEEE Transactions on Signal Processing 52 (2004) 2012–2022. The reduced biquaternion algebra of the engineering literature, its idempotent decomposition, its ring-valued norm and its transform.
- $M_4(\mathbb{C})$ Axis-Free Representation and $M_4(\mathbb{C})$ Axis-Free Representation : Operators, for the matrices, the product, the trace, the determinant and the sixteen basis operators in the coefficient basis.
- $M_4(\mathbb{C})$ Pauli Representation and $M_4(\mathbb{C})$ Pauli Representation : Operators, for the Peirce frame, the switch and the same objects in it.
- $M_2(\mathbb{C})$ Pauli Representation, for the $2\times2$ isomorphism read in full.
- Operators Acting on Two Sides, for the sixteen basis operators, the frame and the proof of the basis operator table.
- Index and Notation Conventions, for the superscripts and the labels.