$\operatorname{End}(M_2(\mathbb{C}))$ as an Algebra: the Regular Representation
Introduction
The algebra of complex-linear endomorphisms of the four-dimensional space $M_2(\mathbb{C})$ is $M_4(\mathbb{C})$, and it is generated by the left and the right multiplications, which commute and together give an isomorphism $M_2(\mathbb{C})\otimes M_2(\mathbb{C})^{\mathrm{op}}\to M_4(\mathbb{C})$.
The article identifies the algebra of endomorphisms, defines the left and the right multiplications, proves that they commute, proves that the products $L_{X}R_{Y}$ span the whole endomorphism algebra, states the double centraliser theorem at order two, and computes the trace and the determinant of an operator on the matrix space. It assumes The Matrix Algebra $M_2(\mathbb{C})$, The Ideals, the Centre and the Simplicity of $M_2(\mathbb{C})$ and The Derivation $X\mapsto[A,X]$ and Its Anti-Self-Adjointness for the Trace Form.
It follows the biquaternion article The Endomorphisms of the Biquaternion Algebra for the same algebra acting on the eight real coordinates.
The Space of Endomorphisms
The matrix space has complex dimension four, so its endomorphism algebra has complex dimension sixteen.
Definition. Let $\operatorname{End}(M_2(\mathbb{C}))$ be the complex algebra of complex-linear maps from $M_2(\mathbb{C})$ to itself.
Proposition. The endomorphism algebra has complex dimension sixteen and real dimension thirty-two, and it is isomorphic to $M_4(\mathbb{C})$.
Proof. A linear map of a four-dimensional complex space is a $4\times4$ matrix in a chosen basis, so the algebra is $M_4(\mathbb{C})$, of complex dimension sixteen; the real dimension is twice that.
Corollary. The endomorphism algebra is simple, with centre the scalars, and its only two-sided ideals are zero and the whole algebra.
Proof. It is $M_4(\mathbb{C})$, and a full matrix algebra is simple with scalar centre.
Corollary. Every endomorphism has a trace and a determinant, computed in any basis of the matrix space.
Proof. The trace and the determinant are invariants of the conjugacy class of the representing matrix.
The Left and Right Multiplications
Definition. For a fixed $X$ let $L_{X}(Y)=XY$ and $R_{Y}(Z)=ZX$.
Proposition. Both $L_{X}$ and $R_{X}$ are linear endomorphisms of the matrix space, and the maps $X\mapsto L_{X}$ and $X\mapsto R_{X}$ are algebra homomorphisms, the second one into the opposite algebra.
Proof. The multiplication is bilinear, so both maps are linear in the argument and multiplicative in the parameter: $L_{X}L_{X'}=L_{XX'}$ and $R_{X}R_{X'}=R_{X'X}$.
Corollary. The left multiplications form a copy of $M_2(\mathbb{C})$ inside $M_4(\mathbb{C})$, and the right multiplications form a copy of the opposite algebra.
Proof. The two homomorphisms are injective, because $L_{X}=0$ forces $X=0$ by acting on the identity.
Corollary. The trace of $L_{X}$ is twice the trace of $X$, and its determinant is the square of the determinant of $X$.
Proof. In the basis of matrix units the left multiplication by $X$ is the operator $Z\mapsto XZ$, whose matrix is the tensor product of $X$ with the identity in the two factors of the matrix space; the trace of such a tensor product is twice the trace of $X$ and its determinant is the square of the determinant.
The Two Families Commute
Theorem. Every left multiplication commutes with every right multiplication, so $L_{X}R_{Y}=R_{Y}L_{X}$ for all $X,Y$.
Proof. Both sides send $Z$ to $XZY$, by associativity.
Corollary. The multiplication map $X\otimes Y\mapsto L_{X}R_{Y}$ is a homomorphism from $M_2(\mathbb{C})\otimes M_2(\mathbb{C})^{\mathrm{op}}$ into $M_4(\mathbb{C})$.
Proof. The two families commute, so the products are multiplicative in the pair.
Corollary. The image of the multiplication map is the whole endomorphism algebra, because the sixteen products of the matrix units span it.
Proof. The products $L_{E_{ij}}R_{E_{kl}}$ send $E_{kl}$ to $E_{ij}$ and every other matrix unit to zero, so they are the sixteen matrix units of the endomorphism algebra in the basis of matrix units.
The Double Centraliser
Theorem. The endomorphisms that commute with every left multiplication are exactly the right multiplications, and conversely.
Proof. A map $T$ with $TL_{X}=L_{X}T$ for all $X$ is determined by $T(I)$, because $T(X)=T(L_{X}I)=L_{X}T(I)=X\,T(I)=R_{T(I)}X$; so $T=R_{T(I)}$. The converse is the commutation theorem.
Corollary. The centraliser of the copy of $M_2(\mathbb{C})$ of the left multiplications is the four-dimensional copy of the right multiplications.
Proof. The theorem identifies it, and the right multiplications are parametrised by the four-dimensional matrix space.
Corollary. The double centraliser theorem holds at order two: the algebra equals its double centraliser inside the endomorphism algebra.
Proof. The centraliser of the right multiplications is again the left multiplications, and the two together generate the whole algebra.
The Dimension Count
Proposition. The algebra $M_2(\mathbb{C})\otimes M_2(\mathbb{C})^{\mathrm{op}}$ has complex dimension sixteen, and the multiplication map is an isomorphism onto $M_4(\mathbb{C})$.
Proof. The tensor product of two four-dimensional algebras has dimension sixteen, the map is injective because its image contains the sixteen independent matrix units, and an injective map between spaces of the same dimension is an isomorphism.
Corollary. Every endomorphism of the matrix space is a sum of at most sixteen products $L_{X}R_{Y}$, and in fact of exactly the products of the matrix units.
Proof. The sixteen products of matrix units form a basis.
Corollary. The endomorphism algebra is generated as an algebra by the eight-dimensional space of left multiplications together with the eight-dimensional space of right multiplications.
Proof. The products of the generators span the algebra, and the two spaces have real dimension eight each.
The Structure of the Algebra
Proposition. The two copies of $M_2(\mathbb{C})$, the left and the right, are mutual commutants, and neither is an ideal.
Proof. The commutant statement is the double centraliser theorem; neither is an ideal because the algebra is simple and each is a proper subalgebra.
Corollary. The endomorphism algebra is not the direct sum of the two copies, and the sum of the two copies is a proper subspace of dimension eight.
Proof. The sum has dimension at most eight and the algebra has dimension sixteen.
Corollary. The two copies intersect in the scalars, so the scalar operators are the only ones that are both left and right multiplications.
Proof. An operator that is both is in the intersection of the two mutual commutants, which is the centre, and the centre is the scalars.
The Trace and the Determinant of an Operator
Proposition. For an operator written as a sum of products $L_{X}R_{Y}$ the trace is computed in the basis of matrix units, and the trace of $L_{X}R_{Y}$ is $\operatorname{tr}(X)\operatorname{tr}(Y)$.
Proof. The operator $L_{X}R_{Y}$ sends $Z$ to $XZY$; in the basis of matrix units its diagonal entries are the coefficients of $E_{ij}$ in $XE_{ij}Y$, and the sum of those coefficients is the product of the two traces.
Corollary. The trace of an operator is a bilinear pairing between the two copies, and it is the tensor product of the two traces of the algebra.
Proof. The formula is the product of the two traces, which is the trace of the tensor product.
Corollary. The determinant of $L_{X}R_{Y}$ is $\det(X)^{2}\det(Y)^{2}$, so the two-sided operator is singular exactly when one of the factors is.
Proof. The determinant is multiplicative over the tensor product and the determinant of $L_{X}$ is $\det(X)^{2}$.
Summary
The endomorphism algebra of the four-dimensional matrix space is $M_4(\mathbb{C})$, of complex dimension sixteen. It is generated by the left multiplications and the right multiplications, which commute, and the multiplication map $X\otimes Y\mapsto L_{X}R_{Y}$ is an isomorphism from $M_2(\mathbb{C})\otimes M_2(\mathbb{C})^{\mathrm{op}}$ onto it.
The two families are mutual commutants and intersect in the scalars, so the double centraliser theorem holds at order two. The trace of $L_{X}R_{Y}$ is the product of the two traces, and the determinant is the product of the squared determinants.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\operatorname{End}(M_2(\mathbb{C}))$ | the algebra of linear endomorphisms, isomorphic to $M_4(\mathbb{C})$ |
| $L_{X}$ | the left multiplication $Y\mapsto XY$ |
| $R_{Y}$ | the right multiplication $Z\mapsto ZY$ |
| $M_2^{\mathrm{op}}$ | the opposite algebra |
| $E_{ij}$ | the matrix units of the matrix space |
Further Reading
- Frank W. Anderson and Kent R. Fuller, Rings and Categories of Modules (Springer, 1992), for the double centraliser theorem.
- Israel Nathan Herstein, Noncommutative Rings (Mathematical Association of America, 1968), for the structure of full matrix algebras.
- The Matrix Algebra $M_2(\mathbb{C})$, for the algebra whose endomorphisms are described here.
- The Derivation $X\mapsto[A,X]$ and Its Anti-Self-Adjointness for the Trace Form, for the commutator read as an operator on the matrix space.
- The Ideals, the Centre and the Simplicity of $M_2(\mathbb{C})$, for the simplicity used in the structure section.