The Frame, the Simple Nilpotents and the Gauge of $M_4(\mathbb{C})$
Introduction
The root fixes the flag, and a frame fixes one vector in each line of the flag; the freedom in that choice is a torus with one parameter per simple nilpotent. The root therefore gives the idempotents and the lines, and the gauge gives the vectors and the units. The two together are the frame.
The article constructs the frame from a basis adapted to the spectral flag, states the matrix-unit rule, isolates the three simple nilpotents $E_{12},E_{23},E_{34}$, computes the gauge action and its cocycle, and counts the parameters: $12$ complex from the root and $3$ complex from the gauge, $15$ in all. The idempotents are gauge invariant. The gauge is not a second primitive: at $n=2$ an involution anticommuting with $H$ is exactly a point of the gauge, and for $n=4$ there is no single such element, only the three parameters.
The idempotents are from Roots of Unity and the Lagrange Idempotents of $M_4(\mathbb{C})$; the corners are The Peirce Decomposition of $M_4(\mathbb{C})$; the principal $\mathfrak{sl}_2$ uses the nilpotents as ladder steps; the count of the root set is The Adjoint Action, the Orbits and the Flag Variety of $M_4(\mathbb{C})$. The general construction is The Frame, the Simple Nilpotents and the Gauge of $M_4(\mathbb{C})$ for $M_n(\mathbb{C})$.
The Adapted Frame
Definition. A frame adapted to the root $H$ is a basis $(v_1,v_2,v_3,v_4)$ of $\mathbb{C}^4$ with
$$ v_k\in V_k=\operatorname{im}P_k,\qquad v_k\ne0,\qquad k=1,2,3,4. $$
Let $(\ell_1,\ell_2,\ell_3,\ell_4)$ be the dual basis of the dual space, $\ell_k(v_m)=\delta_{km}$, and put
$$ E_{kl}=v_k\ell_l . $$
Proposition. The sixteen elements $E_{kl}$ are a basis of $M_4(\mathbb{C})$. They satisfy
$$ E_{kl}E_{mn}=\delta_{lm}E_{kn},\qquad E_{kk}=P_k,\qquad H=\sum_{k=1}^{4}\omega_kE_{kk},\qquad \operatorname{tr}E_{kl}=\delta_{kl}. $$
Proof. For $v\in\mathbb{C}^4$, write $v=\sum_m\ell_m(v)v_m$. Then
$$ E_{kl}E_{mn}v=v_k\ell_l\bigl(v_m\ell_n(v)\bigr)=\delta_{lm}v_k\ell_n(v)=\delta_{lm}E_{kn}v. $$
On $v_m$, $E_{kk}$ acts as $v_k\ell_k(v_m)=\delta_{km}v_k$, so $E_{kk}=P_k$. The sum $\sum_k\omega_kE_{kk}$ acts on $v_m$ as $\omega_mv_m=Hv_m$. The trace is $\ell_l(v_k)=\delta_{kl}$. The sixteen units are the matrix units of the basis, hence a basis of the algebra.
The frame is the choice of the vectors. The root fixes the four lines. The frame fixes a nonzero vector in each line. Different frames of the same flag give different matrix units and the same idempotents:
$$ E_{kk}=P_k $$
depends only on the line $V_k$, not on the vector $v_k$.
The Standard Frame and a Non-Standard One
In the standard frame $v_k=e_k$ and $\ell_k=e_k^{\mathsf T}$, the units $E_{kl}$ are the coordinate matrix units, $P_k=E_{kk}$, and $H=H_0=\operatorname{diag}(1,i,-1,-i)$.
In a general frame the units are the outer products $v_k\ell_l$. For the upper triangular frame of Roots of Unity and the Lagrange Idempotents of $M_4(\mathbb{C})$,
$$ v_1=\begin{pmatrix}1\\0\\0\\0\end{pmatrix},\quad v_2=\begin{pmatrix}1\\1\\0\\0\end{pmatrix},\quad v_3=\begin{pmatrix}0\\1\\1\\0\end{pmatrix},\quad v_4=\begin{pmatrix}0\\0\\1\\1\end{pmatrix}, $$
with dual functionals
$$ \ell_1=(1,-1,1,-1),\quad \ell_2=(0,1,-1,1),\quad \ell_3=(0,0,1,-1),\quad \ell_4=(0,0,0,1), $$
the first units are
$$ E_{12}=v_1\ell_2= \begin{pmatrix}0&1&-1&1\\0&0&0&0\\0&0&0&0\\0&0&0&0\end{pmatrix}, $$
$$ E_{23}=v_2\ell_3= \begin{pmatrix}0&0&1&-1\\0&0&1&-1\\0&0&0&0\\0&0&0&0\end{pmatrix}, $$
$$ E_{34}=v_3\ell_4= \begin{pmatrix}0&0&0&0\\0&0&0&1\\0&0&0&1\\0&0&0&0\end{pmatrix}. $$
They are not the coordinate matrix units, and their entries are not zeros and ones only. The frame, not the standard basis, is the data.
The Simple Nilpotents
Definition. The simple nilpotents of the frame are the three superdiagonal units
$$ N_1=E_{12},\qquad N_2=E_{23},\qquad N_3=E_{34}. $$
Proposition. Each simple nilpotent squares to zero:
$$ N_1^2=N_2^2=N_3^2=0. $$
The composites are the products of consecutive simple nilpotents:
$$ E_{13}=N_1N_2,\qquad E_{24}=N_2N_3,\qquad E_{14}=N_1N_2N_3. $$
The subdiagonal units are
$$ E_{21}=N_1^{*},\qquad E_{32}=N_2^{*},\qquad E_{43}=N_3^{*} $$
in an orthonormal frame, and in a general frame they are the dual units $E_{k+1,k}$. The diagonal units are recovered by the product rule:
$$ E_{11}=E_{12}E_{21},\quad E_{22}=E_{23}E_{32},\quad E_{33}=E_{34}E_{43},\quad E_{44}=E_{43}E_{34}. $$
Proof. $N_k^2=E_{k,k+1}E_{k,k+1}=0$ because the inner indices do not match. The composite formulas are the product rule $E_{k,k+1}E_{k+1,k+2}=E_{k,k+2}$ iterated. The subdiagonal identity is the definition of the conjugate transpose in an orthonormal frame; in a general frame the subdiagonal unit $E_{k+1,k}=v_{k+1}\ell_k$ is the dual unit, not the conjugate transpose of $N_k$ with respect to the standard inner product. The diagonal identities are the product rule with $E_{k,k+1}E_{k+1,k}=E_{kk}$ and $E_{k+1,k}E_{k,k+1}=E_{k+1,k+1}$.
The simple nilpotents generate the frame. The algebra generated by $N_1,N_2,N_3$ and the subdiagonal units $E_{21},E_{32},E_{43}$ is the whole of $M_4(\mathbb{C})$. The products of consecutive $N$'s give all $E_{kl}$ with $k
$$ N_1,\ N_2,\ N_3,\ N_1N_2,\ N_2N_3,\ N_1N_2N_3, $$
of dimension $6$; it is the algebra of the flag, and the lower triangular algebra is generated by the subdiagonal units.
The chain is the Peirce chain. The nilpotent $N_k$ maps the line $V_{k+1}$ to the line $V_k$ and kills the other lines. In the Peirce notation of The Peirce Decomposition of $M_4(\mathbb{C})$,
$$ N_k\in\mathcal{A}_{k,k+1},\qquad k=1,2,3. $$
The chain $N_1,N_2,N_3$ is the sequence of maps that steps down the flag. The composite $E_{kl}$ lies in $\mathcal{A}_{kl}$ and is the product of the simple nilpotents along the chain from $l$ to $k$.
The nilpotents are not polynomials in $H$. In the spectral frame a polynomial in $H$ is diagonal, because $H$ is diagonal; the $N_k$ are off-diagonal. They are therefore not part of the root data: they are frame data. This is the precise sense in which the root alone is not the frame.
The Gauge Action
Definition. A gauge transformation of the frame is a rescaling
$$ v_k\longmapsto t_kv_k,\qquad t_k\in\mathbb{C}^{\times},\qquad k=1,2,3,4. $$
The dual basis is rescaled inversely, $\ell_k\mapsto t_k^{-1}\ell_k$, so the matrix units transform by
$$ E_{kl}=v_k\ell_l\longmapsto t_kt_l^{-1}E_{kl}. $$
Proposition. The gauge fixes the idempotents and the root:
$$ P_k=E_{kk}\longmapsto E_{kk}=P_k,\qquad H=\sum_k\omega_kE_{kk}\longmapsto H. $$
The diagonal $t_1=t_2=t_3=t_4$ acts trivially on every $E_{kl}$, so the effective gauge group is
$$ T/\mathbb{C}^{\times}\cong(\mathbb{C}^{\times})^3. $$
Proof. The idempotent $E_{kk}=v_k\ell_k$ is invariant because $t_kt_k^{-1}=1$. The root is a linear combination of the idempotents with constant coefficients $\omega_k$, hence invariant. The common rescaling $t_k=t$ gives $t t^{-1}=1$ for every invertible element, so the kernel is the diagonal.
One parameter per simple nilpotent. Put
$$ \lambda_k=t_kt_{k+1}^{-1},\qquad k=1,2,3. $$
Then
$$ N_k=E_{k,k+1}\longmapsto\lambda_kN_k, $$
and every other unit is fixed by the $\lambda$'s through the cocycle
$$
E_{kl}\longmapsto \lambda_k\lambda_{k+1}\cdots\lambda_{l-1}\,E_{kl}\quad(k For $k>l$ the scale is the inverse of the product from $l$ to $k$: $$
E_{kl}\longmapsto \bigl(\lambda_l\lambda_{l+1}\cdots\lambda_{k-1}\bigr)^{-1}E_{kl}\quad(k>l).
$$ Proof. For $k The fourth ratio is determined. Put $\lambda_4=t_4/t_1$. Then $$
\lambda_1\lambda_2\lambda_3\lambda_4
=\frac{t_1}{t_2}\frac{t_2}{t_3}\frac{t_3}{t_4}\frac{t_4}{t_1}=1.
$$ There are three independent $\lambda$'s, one per simple nilpotent, and the fourth is their inverse product. There is no fourth independent scalar, and there is no independent scale for a composite corner. The cocycle forbids an extra parameter. Suppose that in addition to the $t_k$ one tries to rescale a single composite $E_{kl}$, $k $$
E_{kl}=E_{k,k+1}E_{k+1,k+2}\cdots E_{l-1,l},
$$ so the scale of $E_{kl}$ is the product of the scales of the factors, and $\mu$ is determined by them. Equivalently, the gauge action on the invertibles is the coboundary of the $0$-cochain $t$ in the torus, and a coboundary has no freedom beyond the cochain modulo constants. The gauge is not a second root. The parameters $\lambda_k$ act on the nilpotents and on the off-diagonal units. They never act on the diagonal, on the idempotents, or on the root. A statement that depends on $\lambda$ is a statement about a frame, not about the algebra or the idempotents. In the standard frame, the gauge is the diagonal matrix $$
D=\operatorname{diag}(t_1,t_2,t_3,t_4),
$$ acting by conjugation: $$
D E_{kl}D^{-1}=\frac{t_k}{t_l}E_{kl}.
$$ Take $$
(t_1,t_2,t_3,t_4)=\left(1,\frac12,\frac16,\frac1{30}\right).
$$ Then $$
\lambda_1=\frac{t_1}{t_2}=2,\qquad
\lambda_2=\frac{t_2}{t_3}=3,\qquad
\lambda_3=\frac{t_3}{t_4}=5,\qquad
\lambda_4=\frac{t_4}{t_1}=\frac1{30},
$$ and $$
\lambda_1\lambda_2\lambda_3\lambda_4=2\cdot3\cdot5\cdot\frac1{30}=1.
$$ The simple nilpotents and the composites scale as The idempotents $E_{11},E_{22},E_{33},E_{44}$ have scale $1$, and the root $H_0$ is unchanged. The fourth ratio $1/30$ is not an independent parameter; it is fixed by the other three and by the condition that the product around the cycle is $1$. A numerical check. With this gauge, $D H_0D^{-1}=H_0$ because $D$ is diagonal and $H_0$ is diagonal, and $D E_{kk}D^{-1}=E_{kk}$ for every $k$. The composite $E_{14}$ scales by $t_1/t_4=30$, which is the product $2\cdot3\cdot5$. Rescaling $E_{14}$ by any other number would violate $E_{14}=E_{12}E_{23}E_{34}$. A frame modulo overall scale has $15$ complex dimensions. The set of bases of $\mathbb{C}^4$ is $GL_4(\mathbb{C})$, of complex dimension $16$. Two bases that differ by an overall scalar give the same matrix units, since $E_{kl}=v_k\ell_l$ is unchanged when all $v_k$ are multiplied by $t$ and all $\ell_l$ by $t^{-1}$. The frame modulo scale is therefore a point of $PGL_4(\mathbb{C})$, of complex dimension $15$. The root and the gauge split the count. The map from frames to roots, $$
(v_1,v_2,v_3,v_4)\longmapsto H=\sum_{k=1}^{4}\omega_kE_{kk},
$$ has fibre the torus $T/\mathbb{C}^{\times}\cong(\mathbb{C}^{\times})^3$: the frames that give the same root are exactly the rescalings of the vectors inside the fixed lines. Hence $$
\underbrace{15}_{\text{frame}/\mathbb{C}^{\times}}
=
\underbrace{12}_{\text{root}}
+
\underbrace{3}_{\text{gauge}} .
$$ The root set is the base of the fibration, the gauge is the fibre, and the frame is the total space. The root count is The Adjoint Action, the Orbits and the Flag Variety of $M_4(\mathbb{C})$; the gauge is the stabiliser of the root in $PGL_4$. The case $n=2$. For the biquaternion algebra $M_2(\mathbb{C})$, $$
\underbrace{3}_{\text{frame}/\mathbb{C}^{\times}}
=
\underbrace{2}_{\text{root}}
+
\underbrace{1}_{\text{gauge}},
$$ all complex. In real dimensions the frame modulo scale has $6$, the root has $4$ and the gauge has $2$: the four real parameters of the biquaternion root and the two real parameters of the complex number $\lambda$. This is the six-dimensional basis of the biquaternion frame. At $n=2$ an involution of the frame that anticommutes with $H$ is exactly a point of the gauge. In the frame $(v_1,v_2)$ with $H=\operatorname{diag}(1,-1)$, let $R$ satisfy $$
R^2=I,\qquad RH=-HR.
$$ The anticommutation forces $R$ to be off-diagonal, $R=aE_{12}+bE_{21}$, and $R^2=abI=I$ forces $b=a^{-1}$. Hence $$
R=aE_{12}+a^{-1}E_{21},\qquad a\in\mathbb{C}^{\times}.
$$ Under the gauge $E_{12}\mapsto\lambda E_{12}$, $E_{21}\mapsto\lambda^{-1}E_{21}$, the element $R$ is carried to $$
a\lambda E_{12}+a^{-1}\lambda^{-1}E_{21},
$$ which is the same family with $a$ replaced by $a\lambda$. The gauge orbit of $R$ is the whole family, and the parameter $a$ is gauge, not root. What $R$ carries and what it does not. It names one nonzero element in each mixed Peirce line, $R$ having nonzero entries exactly in $\mathcal{A}_{12}$ and $\mathcal{A}_{21}$, and it normalises the two reciprocally, so it is a shorthand for a normalised pair. It supplies no idempotent: those are $P_\pm=(I\pm H)/2$, produced by $H$ alone. The biquaternion articles state the same data as the choice of one nonzero element of a mixed line, with the partner fixed by the normalisation; the primitive is the root, and the choice is the basis. For $n=4$ there is no single such element. Elements anticommuting with $H$ do exist: they occupy the two mixed blocks $\mathcal{A}_{13},\mathcal{A}_{31},\mathcal{A}_{24},\mathcal{A}_{42}$, and a general such element is a sum of two independent off-diagonal pairs. Under the gauge it sees only the two combinations $\lambda_1\lambda_2$ and $\lambda_2\lambda_3$, so its orbit has dimension two and it does not carry the third parameter. No single element therefore plays the role that $R$ plays at $n=2$; the gauge is a torus, and there is no second primitive. An adapted frame $(v_1,v_2,v_3,v_4)$ gives the matrix units $E_{kl}=v_k\ell_l$, with $E_{kl}E_{mn}=\delta_{lm}E_{kn}$, $E_{kk}=P_k$ and $H=\sum_k\omega_kE_{kk}$. The simple nilpotents $N_1=E_{12}$, $N_2=E_{23}$, $N_3=E_{34}$ square to zero, generate the algebra together with the subdiagonal units (their conjugate transposes in an orthonormal frame), and every composite is their consecutive product. The gauge $v_k\mapsto t_kv_k$ acts by $E_{kl}\mapsto(t_k/t_l)E_{kl}$, an effective torus $(\mathbb{C}^{\times})^3$ with one parameter $\lambda_k=t_k/t_{k+1}$ per simple nilpotent, subject to the cocycle that fixes the composites and to the relation $\lambda_1\lambda_2\lambda_3\lambda_4=1$. The frame modulo scale has $15$ complex dimensions, split as $12$ for the root and $3$ for the gauge; at $n=2$ this is $3=2+1$ complex, the $6=4+2$ real count of the biquaternion frame. An involution of the frame anticommuting with $H$ is a point of this gauge, not a second primitive. $$
\boxed{\ 15=12+3,\qquad N_k\longmapsto\lambda_kN_k,\qquad E_{kl}\longmapsto\lambda_k\cdots\lambda_{l-1}E_{kl},\qquad \lambda_1\lambda_2\lambda_3\lambda_4=1.\ }
$$An Explicit Gauge Transformation
Unit
Scale
$E_{12}$
$\lambda_1=2$
$E_{23}$
$\lambda_2=3$
$E_{34}$
$\lambda_3=5$
$E_{13}$
$\lambda_1\lambda_2=6$
$E_{24}$
$\lambda_2\lambda_3=15$
$E_{14}$
$\lambda_1\lambda_2\lambda_3=30$
$E_{21},E_{32},E_{43}$
$\lambda_1^{-1}, \lambda_2^{-1}, \lambda_3^{-1}$
$E_{31},E_{42},E_{41}$
$6^{-1}, 15^{-1}, 30^{-1}$
The Count of the Frame
An Anticommuting Involution Is a Gauge Point
Summary
Object
Dimension over $\mathbb{C}$
Frame $GL_4$
$16$
Frame modulo overall scale $PGL_4$
$15$
Root $PGL_4/T$
$12$
Gauge $(\mathbb{C}^{\times})^3$
$3$
Simple nilpotents
$3$
Independent gauge parameters
$3$
Summary of Notation
Symbol
Meaning
$v_k$
a nonzero vector in the spectral line $V_k$
$\ell_k$
the dual functional, $\ell_k(v_m)=\delta_{km}$
$E_{kl}=v_k\ell_l$
the matrix unit of the frame
$N_k=E_{k,k+1}$
the simple nilpotent, $N_k^2=0$
$t_k$
the gauge scale of $v_k$
$\lambda_k=t_kt_{k+1}^{-1}$
the gauge parameter on $N_k$
$T$
the torus of rescalings, effective $(\mathbb{C}^{\times})^3$
$D=\operatorname{diag}(t_1,t_2,t_3,t_4)$
the gauge matrix in the standard frame
Further Reading