Quadruple-Branch Transformers

• Quadruple-Branch Transformers is a unified analytic formalism for real det=1 ABCD matrices, parameterized continuously by the branch-parameter theta.
• It employs similarity transformations and an exponential mapping to convert matrices into one of four canonical Wigner forms, distinguishing rotation, squeeze, and shear behaviors.
• The approach is applied in paraxial optics, multilayer systems, and dynamical linear analyses, streamlining stability and propagation studies.

A quadruple-branch structure arises in the analysis of real $2 \times 2$ matrices of unit determinant—so-called ABCD matrices—which are fundamental in paraxial optics, multilayer systems, and a range of dynamical linear systems. Not all ABCD matrices are diagonalizable, but every real $\det=1$ ABCD matrix can be brought by real similarity transformation into one of four canonical Wigner forms. Remarkably, these four forms can be unified in a single analytic exponential family parameterized continuously by a “branch-parameter” $\theta$. The resulting quadruple-branch formalism provides a seamless analytic interpolation among elliptic (rotational), hyperbolic (squeezing), and two parabolic (shear) matrix forms, governed by the sign and value of $\cos 2\theta$ and by the trace or spectrum of the original matrix (Baskal et al., 2010).

1. Canonical Decomposition of Real $\det=1$ ABCD Matrices

Let $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$, with $A,B,C,D \in \mathbb{R}$ and $AD-BC=1$. Any such matrix can be brought, via a similarity transformation, into an equi-diagonal canonical form:

$[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$

where $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$ is a real rotation matrix, and $\det=1$0. The resulting $\det=1$1 has $\det=1$2 and determinant $\det=1$3. Thus, $\det=1$4 depends on two real parameters, and can be written as the exponential:

$\det=1$5

where $\det=1$6 and $\det=1$7. The generator $\det=1$8 is a real traceless matrix:

$\det=1$9

This unified exponential form underpins the quadruple-branch structure.

2. The Four Wigner Branches: Classification and Structure

The analytic family $\theta$0 decomposes into four qualitative “branches” determined by $\theta$1 or, equivalently, by the trace or eigenvalues of $\theta$2. The regime is dictated by the relative size of $\theta$3 vs. $\theta$4 and the special points $\theta$5.

Branch	$\theta$6 Condition	Canonical Form
Elliptic	$\theta$7	Rotation ($\theta$8)
Hyperbolic	$\theta$9	Squeeze ($\cos 2\theta$0)
Parabolic I	$\cos 2\theta$1	Upper shear ($\cos 2\theta$2)
Parabolic II	$\cos 2\theta$3	Lower shear ($\cos 2\theta$4)

Elliptic (Rotation) Branch

For $\cos 2\theta$5 ($\cos 2\theta$6):

• $\cos 2\theta$7, $\cos 2\theta$8
• $\cos 2\theta$9
• Eigenvalues: complex conjugate $\det=1$0, trace $\det=1$1, stable oscillatory propagation.

Hyperbolic (Squeeze) Branch

For $\det=1$2 ($\det=1$3):

• $\det=1$4, $\det=1$5
• $\det=1$6
• Eigenvalues: real and reciprocal $\det=1$7, trace $\det=1$8, exponentially growing/decaying (evanescent) modes.

Parabolic I (Upper-Shear) Branch

At $\det=1$9:

• $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$0, so $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$1
• Exponential truncates: $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$2
• Double eigenvalue $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$3, Jordan block structure.

Parabolic II (Lower-Shear) Branch

At $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$4:

• $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$5
• $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$6
• Double eigenvalue $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$7 or $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$8, Jordan block.

3. Branch Selection: Trace and Eigenvalue Criteria

The regime is determined by the eigenvalues $$ABCD \equiv \begin{pmatrix}A & B \ C & D\end{pmatrix}$$9 of $A,B,C,D \in \mathbb{R}$0, which always satisfy $A,B,C,D \in \mathbb{R}$1 since $A,B,C,D \in \mathbb{R}$2:

• Elliptic: $A,B,C,D \in \mathbb{R}$3 $A,B,C,D \in \mathbb{R}$4 eigenvalues $A,B,C,D \in \mathbb{R}$5, associated with stable, oscillatory propagation.
• Hyperbolic: $A,B,C,D \in \mathbb{R}$6 $A,B,C,D \in \mathbb{R}$7 eigenvalues $A,B,C,D \in \mathbb{R}$8, with exponential growth/decay.
• Parabolic: $A,B,C,D \in \mathbb{R}$9, double eigenvalue $AD-BC=1$0, corresponds to shear transformations.

The branch transitions occur smoothly as $AD-BC=1$1 approaches $AD-BC=1$2, but the underlying Jordan normal form changes discontinuously, reflecting the appearance of a nontrivial nilpotent structure.

4. Unified Exponential Representation and Analytical Continuity

All four branches are encompassed within the continuous analytic form:

$AD-BC=1$3

with analytic generator and branch parameter. The parabolic branches occur at the endpoints $AD-BC=1$4, where $AD-BC=1$5 and the exponential terminates at linear order. For intermediate values, the matrix exponential produces either trigonometric (elliptic) or hyperbolic (squeeze) forms, with explicit parameterizations for $AD-BC=1$6, $AD-BC=1$7, and $AD-BC=1$8 in terms of $AD-BC=1$9 and $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$0:

• Elliptic ($[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$1): $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$2, $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$3
• Hyperbolic ($[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$4): $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$5, $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$6

The similarity transform back to the original basis is $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$7.

5. Physical Realizations and Examples

The quadruple-branch formalism admits direct and explicit applications in optics:

Optical Activity with Differential Absorption

A chiral medium imposes rotation $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$8 per unit length and asymmetric attenuation $[abcd] = R(-\alpha)\,ABCD\,R(+\alpha)$9. The system is parameterized by $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$0, $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$1, $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$2, and the propagation is $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$3. The behavior is:

• $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$4: elliptic (net rotation with modulation).
• $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$5: hyperbolic (pure attenuation).
• $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$6: parabolic (shear).

Laser Cavity Round-Trip

For a symmetric two-mirror cavity, the round-trip ABCD matrix in appropriate coordinates is equi-diagonal. The quadruple-branch structure rationalizes the small-$$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$7 (mirror curvature) approximation and facilitates closed-form expressions for iteration and mode stability.

Multilayer Optics

In multilayered systems, single cycles combine boundary squeezes and phase-shift rotations. After suitable similarity reduction, these systems are parameterized by $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$8, and the central factor is exactly of $$R(\alpha) = \begin{pmatrix} \cos(\alpha/2) & -\sin(\alpha/2) \ \sin(\alpha/2) & \cos(\alpha/2)\end{pmatrix}$$9 form. The four-branch division controls whether the amplitude evolution is oscillatory, evanescent, or exhibits impedance-matching shear characteristics.

6. Significance and Unification

The quadruple-branch analytic form $\det=1$00, with explicit branch-selection via $\det=1$01 or trace, provides an exhaustive and continuous parameterization of real $\det=1$02 ABCD matrices (Baskal et al., 2010). This formalism unifies what would otherwise appear as disjoint regimes—bounded oscillatory, divergent, and degenerate (shear) optical behaviors—within a single, codified analytic structure. It streamlines the analysis of propagation, stability, and transfer in both classical paraxial optics and layered media. The smooth analytic interpolation among branches clarifies longstanding distinctions between rotation, squeezing, and shear in the Wigner-Mackey group-theoretic classification of linear systems and provides a powerful framework for practical design and analysis.