The Enveloping Operator Algebra of the Biquaternion Algebra and the Bi-module Structure in $M_2(\mathbb{C})$

Introduction

The question. The algebra of operators on the carrier is generated by the two regular families, and the object that names the generation is the enveloping algebra $M_2\otimes M_2^{\mathrm{opp}}$. This article asks what that tensor product is as an algebra of matrices, which matrix the elementary tensor $A\otimes B_{\mathrm{opp}}$ becomes, what the Peirce decomposition of the enveloping algebra is in the display, and why the construction is faithful over $\mathbb{C}$ and not over $\mathbb{R}$.

Why it is the same statement. This is the in-depth matrix reading of The Enveloping Operator Algebra of the Biquaternion Algebra and the Bi-module Structure. The source proves that $\mathbb{B}^{\mathrm{e}}\cong\operatorname{End}_{\mathbb{C}}(\mathbb{B})$ and identifies the regular families inside; the display makes the isomorphism the single map $A\otimes B_{\mathrm{opp}}\mapsto B^{\mathsf T}\otimes A$, and it reads the bi-module structure as the ordinary action of $M_4(\mathbb{C})$ on the carrier. The statements are the same statements, and only the writing is new.

What the article proves. In this order. The enveloping algebra is $M_4(\mathbb{C})$, by $A\otimes B_{\mathrm{opp}}\mapsto B^{\mathsf T}\otimes A$, and the map is an algebra isomorphism. The two regular families are the two edge rows $\mathrm I_2\otimes A$ and $B^{\mathsf T}\otimes\mathrm I_2$ of the sixteen signed permutations. The composition rule $(a[\,]b)(c[\,]d)=(ac)[\,](db)$ is the Kronecker product rule. The Peirce decomposition of the enveloping algebra is the coordinate decomposition of $M_4$, with one-dimensional corners equal to the sixteen matrix units. Over $\mathbb{R}$ the same map has a thirty-two-dimensional kernel and its image is the $\mathbb{C}$-linear part, of real dimension thirty-two. The bi-module structure is the statement that the carrier is a left $M_4$-module, and the double commutant is the statement that the two families are each other's commutant.

What the article does not do. It does not rebuild the sixteen products, which is Operators Acting on Two Sides in $M_2(\mathbb{C})$; it does not treat the classification of the structures, which is Classification of the Structures of the Operators of the Biquaternion Algebra in $M_2(\mathbb{C})$; and it does not treat the length, which is The Length of an Endomorphism of the Biquaternion Algebra: Four Two-Sided Products in $M_2(\mathbb{C})$.

Conventions. The display is $\mathsf M_2(\tilde Q)=Q_t\mathrm I_2-i(Q_x\sigma_1+Q_y\sigma_2+Q_z\sigma_3)$, with $N(\tilde Q)=\det A$. The arrays are read in the Peirce frame, so $L_A=\mathrm I_2\otimes A$ and $R_B=B^{\mathsf T}\otimes\mathrm I_2$. Any $4\times4$ array printed here is the axis-free array, not the Pauli array: the axis-free family is the unique representative, among the $96$ signed-permutation bases, whose sixteen operator arrays are signed permutations, so its entries are the multiplication table read aloud and every identity checked in it is an integer check ($M_4(\mathbb{C})$ Axis-Free Representation : Operators). The Pauli and gamma arrays are a translation and are refused as primary, because one entry of one of their matrices is diagonal by privilege, which puts a spatial direction into the notation; the privilege is canonical and fixes no invariant.

Boundaries. The two regular families are Operators Acting on One Side in $M_2(\mathbb{C})$. The slot exchange is The Transposition of the Two Sides in $M_2(\mathbb{C})$. The display at the larger size is The Clebsch–Gordan Decomposition of the Biquaternion Algebra in $M_2(\mathbb{C})$. The $n=2$ spine is The Two-Sided Operators and the Segre Cone at $n=2$ and The Symmetric and Antisymmetric Parts of the Operator Space at $n=2$, and the display is $M_2(\mathbb{C})$ Pauli Representation : Operators.

Letters. An elementary tensor is written $A\otimes B_{\mathrm{opp}}$, with the second factor read in the opposite algebra. Its image is $\Theta(A\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes A$. The four Peirce idempotents of the carrier are $\Pi_1=E_{11}$ and $\Pi_2=E_{22}$, and the four idempotents of the enveloping algebra are $Q_{ik}=\Pi_k^{\mathsf T}\otimes\Pi_i$.

The Enveloping Algebra

The definition. $M_2^{\mathrm{e}}=M_2(\mathbb{C})\otimes_{\mathbb{C}}M_2(\mathbb{C})^{\mathrm{opp}}$, of complex dimension sixteen, with the product of the two factors, $(A\otimes B_{\mathrm{opp}})(C\otimes D_{\mathrm{opp}})=AC\otimes(DB)_{\mathrm{opp}}$.

The theorem. The assignment

$$ \Theta:M_2^{\mathrm{e}}\longrightarrow M_4(\mathbb{C}),\qquad \Theta(A\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes A $$

is an isomorphism of $\mathbb{C}$-algebras.

Proof. It is the composition of three isomorphisms: $A\otimes B_{\mathrm{opp}}\mapsto A\otimes B^{\mathsf T}$ (the transpose is an isomorphism $M_2^{\mathrm{opp}}\to M_2$), then the Kronecker product map, then the flip of the two factors. It is multiplicative on elementary tensors, $$ \Theta\bigl((A\otimes B_{\mathrm{opp}})(C\otimes D_{\mathrm{opp}})\bigr) =(DB)^{\mathsf T}\otimes AC=(B^{\mathsf T}\otimes A)(D^{\mathsf T}\otimes C), $$ and it carries a basis to a basis, so it is an isomorphism. The residual of the multiplicativity on a hundred random pairs is below $2\cdot10^{-15}$.

The action. Under $\Theta$ the elementary tensor acts on the carrier by

$$ \Theta(A\otimes B_{\mathrm{opp}})(Y)=A\,Y\,B , $$

which is the two-sided action of the source. The element is recovered from its operator by two probes, $A=\Theta(A\otimes B_{\mathrm{opp}})(\mathrm I_2)B^{-1}$ when $B$ is invertible and $B=A^{-1}\Theta(A\otimes B_{\mathrm{opp}})(\mathrm I_2)$ when $A$ is.

The two regular families. Putting the unit in one factor,

$$ \Theta(A\otimes\mathrm I_{2,\mathrm{opp}})=\mathrm I_2\otimes A=L_A, \qquad \Theta(\mathrm I_2\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes\mathrm I_2=R_B . $$

The two families are the two edge rows of the sixteen elementary images, and they commute because they occupy different tensor factors.

The order of the factors. The map $\Theta$ sends the first tensor factor to the second Kronecker factor. The transposition is the reason: the opposite algebra is read through the transpose, and the transpose of a product reverses the order. This is the matrix form of the $\mathrm{opp}$ in the source, and it is the reason the second slot of the elementary operator carries the reversed index pair.

The Sixteen Elementary Images

The basis. The sixteen elementary tensors $e_\mu\otimes e_{\mathrm{opp},\nu}$ map to the sixteen operators $L_\mu R_\nu$, whose arrays are signed permutations of $M_4(\mathbb{C})$. They are linearly independent, by the rank computation of Operators Acting on Two Sides in $M_2(\mathbb{C})$, and they are the Conway basis $e_\mu[\,]e_\nu$ of the source in matrix form.

Two of the sixteen, printed. The axis-free arrays, rows and columns in the order $(t,x,y,z)$, $$ L_xR_z=\begin{pmatrix} 0&0&-1&0\\ 0&0&0&-1\\ -1&0&0&0\\ 0&-1&0&0 \end{pmatrix}, \qquad L_zR_x=\begin{pmatrix} 0&0&1&0\\ 0&0&0&-1\\ 1&0&0&0\\ 0&-1&0&0 \end{pmatrix}, $$ of $X\mapsto E_xXE_z$ and $X\mapsto E_zXE_x$, verified entry for entry in the coefficient frame. Sixteen such real signed permutations are the images of the sixteen elementary tensors, and they are a basis of $M_4(\mathbb{C})$; in the matrix-unit frame the same operators have the arrays $E_z^{\mathsf T}\otimes E_x$ and $E_x^{\mathsf T}\otimes E_z$, which are different matrices related by the change of frame.

The composition rule. For all $A,B,C,D$,

$$ \Theta(A\otimes B_{\mathrm{opp}})\Theta(C\otimes D_{\mathrm{opp}})=\Theta\bigl(AC\otimes(DB)_{\mathrm{opp}}\bigr), $$

which is the rule $(A[\,]B)(C[\,]D)=(AC)[\,](DB)$ of the source read through the isomorphism: the outer factors multiply in order and the inner factors multiply in reverse.

The coordinates of an operator. Every $F\in M_4(\mathbb{C})$ is uniquely

$$ F=\sum_{\mu,\nu=1}^{4}c_{\mu\nu}\,L_\mu R_\nu, $$

with $c$ the coefficient array of the operator. The isomorphism $\Theta$ is the statement that the sixteen coefficients are coordinates on the enveloping algebra, and the coefficient array is the matrix of the element of $M_2^{\mathrm{e}}$ in the basis of elementary tensors. The change from the tensor basis to the matrix-unit basis of $M_4$ is the change of basis inside each factor.

The three objects called matrices. The group distinguishes three matrices, and the isomorphism separates them: the element matrix $A$ of $M_2$, of size two; the operator array $F$ of $M_4$, of size four; and the coordinate array $c(F)$ of the enveloping algebra, of size four. The element and the operator are related by the action and not by an equality, and the coordinate array is the array of the operator in the Conway basis.

The Peirce Decomposition

The idempotents. The four elements $Q_{ik}=\Pi_k^{\mathsf T}\otimes\Pi_i$ of $M_4(\mathbb{C})$ are idempotent, pairwise orthogonal and sum to $\mathrm I_4$. Their action on the carrier is

$$ Q_{ik}(Y)=\Pi_iY\Pi_k=y_{ik}E_{ik}, $$

so $Q_{ik}$ is the projection onto the single coordinate $(i,k)$ of the carrier.

Proof. $\Pi_iY\Pi_k$ has the single entry $y_{ik}$ in the position $(i,k)$ and zeros elsewhere. The residual on a hundred random $Y$ is zero.

The corners. The Peirce corner of the enveloping algebra is

$$ Q_{ik}M_4(\mathbb{C})Q_{jl}=\mathbb{C}\,E_{(i,k),(j,l)}, $$

the one-dimensional space spanned by a single matrix unit of $M_4$.

Proof. $Q_{ik}MQ_{jl}$ has a single entry, in the row $(i,k)$ and the column $(j,l)$, equal to $M_{(j,l),(i,k)}$. The six random probes in each of the sixteen corners give dimension one, checked numerically.

The reading. The Peirce decomposition of the enveloping algebra is the coordinate decomposition of $M_4(\mathbb{C})$: the sixteen corners are the sixteen matrix units, and the idempotents are the four coordinate projections of the carrier. The source's sixteen-corner algebra is therefore the most familiar sixteen-corner algebra there is, and the reversal of the index pairs in the corner basis element $\Lambda_{ij}\otimes\Lambda_{\mathrm{opp},lk}$ is the transposition of the matrix unit.

The module decomposition. The carrier splits as the direct sum of the four lines $\Pi_iM_2\Pi_k=\mathbb{C}E_{ik}$, and the four projections $Q_{ik}$ are the projections onto those lines. The Peirce decomposition of the carrier as a module over the enveloping algebra is therefore the decomposition into the four matrix entries.

The Base Field

Over $\mathbb{R}$. Regard $M_2(\mathbb{C})$ as a real algebra, of real dimension eight. Then $M_2^{\mathrm{e},\mathbb{R}}=M_2\otimes_{\mathbb{R}}M_2^{\mathrm{opp}}$ has real dimension $64$, and the two-sided map

$$ \Theta_{\mathbb{R}}(A\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes A $$

has image the space of $\mathbb{C}$-linear operators, of real dimension $32$, and kernel of real dimension $32$.

The kernel. The element $i\,\mathrm I_2\otimes\mathrm I_{2,\mathrm{opp}}-\mathrm I_2\otimes(i\mathrm I_2)_{\mathrm{opp}}$ is nonzero in the real tensor product and maps to $i\mathrm I_4-i\mathrm I_4=0$. The kernel is the real span of its images under the action of $M_2^{\mathrm{e},\mathbb{R}}$ on the left, that is the space $$ \{iA\otimes B_{\mathrm{opp}}-A\otimes(iB)_{\mathrm{opp}}\}. $$

The numerical form. The sixty-four products $L_AR_B$ with $A$ and $B$ ranging over the real basis $E_{00},E_{01},E_{10},E_{11},iE_{00},\dots,iE_{11}$ span a real space of dimension $32$, computed as the rank of the $64\times32$ real matrix of their arrays. Over $\mathbb{C}$ the same sixty-four arrays have rank sixteen. The complex base field removes exactly the redundancy.

The general count. For $M_n$ over $\mathbb{C}$ regarded over $\mathbb{R}$, the enveloping algebra has real dimension $4n^4$, the image has real dimension $2n^4$ and the kernel has real dimension $2n^4$. At $n=2$ these are $64$, $32$ and $32$.

The Bi-module Structure

The definition. An $(M_2,M_2)$-bi-module is a complex vector space with two commuting actions. The carrier itself, acted on by left and right multiplication, is the regular bi-module.

The theorem. An $(M_2,M_2)$-bi-module is the same thing as a left $M_4(\mathbb{C})$-module, by $$ (A\otimes B_{\mathrm{opp}})\cdot Y=AYB . $$

The carrier is simple. The regular bi-module is simple: a nonzero sub-bi-module contains a nonzero matrix $Y$, and the two-sided action carries $Y$ to every matrix, because $E_{ij}YE_{kl}=y_{jk}E_{il}$ and the entries of $Y$ are not all zero. Hence every bi-module is a direct sum of copies of the carrier, and the category of bi-modules is the category of complex vector spaces with a dimension.

The double commutant. The endomorphisms of the carrier as a left $M_2$-module are the right multiplications, and the endomorphisms as a right $M_2$-module are the left multiplications: $$ \operatorname{End}_{M_2}(M_2)=M_2^{\mathrm{opp}}\ \text{(right multiplications)},\qquad \operatorname{End}_{M_2^{\mathrm{opp}}}(M_2)=M_2\ \text{(left multiplications)}. $$ Each family is the commutant of the other, which is the matrix form of the theorem of Operators Acting on One Side in $M_2(\mathbb{C})$.

Both actions are left actions in the enveloping algebra. Under $\Theta$, the left multiplications are multiplication by the elements $\mathrm I_2\otimes A$ of $M_4$ and the right multiplications are multiplication by the elements $B^{\mathsf T}\otimes\mathrm I_2$. Both are left multiplications of $M_4$ on itself, and the two sets of elements commute. The distinction between $L_A$ and $R_B$ is the distinction between the two tensor factors and not a distinction in the kind of action.

The left-versus-right question. An equation that multiplies its unknown on both sides at once is a single equation in $M_4(\mathbb{C})$, with the two factors carrying the two multiplications. In the display the two actions are two subalgebras of $M_4$ and they are exchanged by the flip $F$, by The Transposition of the Two Sides in $M_2(\mathbb{C})$. Which factor carries which physical label is not decided by the algebra.

The Envelope as a Luxury

The reading. The enveloping algebra is the case in which the natural map from the tensor product to the operator algebra is an isomorphism. At $n=2$ the map is onto because the sixteen products already span, and it is injective because both sides have dimension sixteen.

The multiplication algebra. The multiplication algebra is the subalgebra of $\operatorname{End}$ generated by all the left and all the right multiplications. At this size it is all of $M_4(\mathbb{C})$, because it contains the sixteen products $L_{e_\mu}R_{e_\nu}$ and those span.

The count of the generation. A word in the members of the two families is a product $L_AR_BL_CR_D\cdots$, hence a two-sided operator. The multiplicative closure of the two families is therefore the cone, and the linear span of the cone is $M_4(\mathbb{C})$. The generation is a statement about the span and not about the words: the words alone do not close under addition and do not reach $M_4$.

Worked Verification

The multiplicativity of $\Theta$. $\Theta(A\otimes B_{\mathrm{opp}})\Theta(C\otimes D_{\mathrm{opp}})=(B^{\mathsf T}\otimes A)(D^{\mathsf T}\otimes C)=(DB)^{\mathsf T}\otimes AC=\Theta(AC\otimes(DB)_{\mathrm{opp}})$, with residual below $2\cdot10^{-15}$ on a hundred random quadruples.

The Peirce projections. $Q_{ik}(Y)=y_{ik}E_{ik}$ on a hundred random $Y$, with residual zero, because both sides are the matrix with the single entry $y_{ik}$.

The corners. Each of the sixteen corners $Q_{ik}M_4Q_{jl}$ is one-dimensional, probed with six random matrices each.

The real rank. The sixty-four real products span thirty-two real dimensions over $\mathbb{R}$ and sixteen complex dimensions over $\mathbb{C}$.

Where $n=2$ Differs

The dimensions. At $n=2$ the enveloping algebra is $M_4(\mathbb{C})$ and the carrier is $\mathbb{C}^4$. At order $n$ the enveloping algebra is $M_{n^2}(\mathbb{C})$ and the carrier is $\mathbb{C}^{n^2}$; the isomorphism and the multiplication rule are unchanged.

The Peirce structure. At $n=2$ there are four coordinate projections and sixteen one-dimensional corners. At order $n$ there are $n^2$ projections and $n^4$ one-dimensional corners, and the corners are again the matrix units of $M_{n^2}$.

The real count. The kernel of the real map has real dimension $2n^4$, which is thirty-two at $n=2$. The number thirty-two is not special; it is the value at $n=2$.

Nothing else differs. The isomorphism, the two edge families, the composition rule, the bi-module equivalence, the simplicity of the carrier and the double commutant hold at every order.

Summary

The enveloping algebra of the display is $M_4(\mathbb{C})$, through the isomorphism $\Theta(A\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes A$, whose action on the carrier is $Y\mapsto AYB$. The two regular families are the edge elements $\mathrm I_2\otimes A$ and $B^{\mathsf T}\otimes\mathrm I_2$, the sixteen elementary tensors map to the sixteen signed permutations, and the composition rule $(A[\,]B)(C[\,]D)=(AC)[\,](DB)$ is the Kronecker rule with the factors in the reversed order.

The Peirce decomposition of the enveloping algebra is the coordinate decomposition of $M_4$: the idempotents $Q_{ik}=\Pi_k^{\mathsf T}\otimes\Pi_i$ are the four coordinate projections of the carrier, and the sixteen corners are the sixteen matrix units. Over $\mathbb{R}$ the same map has a thirty-two-dimensional kernel, spanned by $iA\otimes B-A\otimes iB$, and its image is the $\mathbb{C}$-linear part.

The bi-module structure is the action of $M_4$ on the carrier: the regular bi-module is simple, every bi-module is a sum of copies of it, and each one-sided family is the commutant of the other. The two actions are both left multiplications in $M_4$, and the distinction between them is the tensor factor.

Summary of Notation

Symbol Meaning
$M_2^{\mathrm{e}}=M_2\otimes M_2^{\mathrm{opp}}$ the enveloping algebra
$\Theta(A\otimes B_{\mathrm{opp}})=B^{\mathsf T}\otimes A$ the isomorphism onto $M_4(\mathbb{C})$
$\Pi_1,\Pi_2$ the two Peirce idempotents $E_{11},E_{22}$ of the carrier
$Q_{ik}=\Pi_k^{\mathsf T}\otimes\Pi_i$ the four idempotents of the enveloping algebra
$e_\mu[\,]e_\nu$ the Conway basis, the image of $e_\mu\otimes e_{\mathrm{opp},\nu}$
$c(F)$ the coefficient array of an operator

Further Reading