Linear Canonical Transformation¶
A symplectic-matrix-indexed family of wave operators that transforms functions while preserving the canonical structure of a conjugate phase plane.
Core Idea¶
A linear canonical transformation (LCT), in its standard real one-dimensional form, is a family of function operators indexed by real matrices \(M=\begin{pmatrix}a&b\\c&d\end{pmatrix}\) satisfying \(ad-bc=1\). The matrix acts linearly and symplectically on a conjugate position–momentum or time–frequency plane; a compatible metaplectic operator acts on the wavefield or signal. Fourier rotation, Fresnel propagation, scaling and quadratic-phase chirp multiplication are special cases. The \(b\ne0\) member has a quadratic-phase integral kernel, while \(b=0\) requires a pointwise scaling/chirp expression or distributional delta kernel. Operator phase/sign requires a consistent metaplectic lift; matrix multiplication alone does not globally fix it.[ref-eb81bdb6e66a][ref-48b962d9e69c]
Scope of Application¶
In coherent paraxial optics, an ideal first-order lens system has a ray-transfer matrix whose entries determine a corresponding input-to-output diffraction operator; Collins established this connection. In signal and image restoration, Barshan, Kutay and Ozaktas selected an LCT domain for filtering particular nonstationary or space/time-varying degradations. The same matrix/operator pairing is used in quadratic Hamiltonian settings. It is not a blanket claim about nonparaxial optics, arbitrary complex matrices or exact equivalence of a discrete implementation with the continuous unitary operator.[ref-788eec23e862][ref-b3f96aff8883][^ref-eb81bdb6e66a]
Clarity¶
The LCT separates a centered phase-plane map from its wave-function realization. \(M\) acts linearly, not as a generic affine translation; shifts and modulations belong to associated inhomogeneous operations. The distinction also prevents a common error at \(b=0\): the singularity of the usual integral formula is not the disappearance of identity, scaling or thin-lens chirp members. Finally, two operators can correspond to the same symplectic matrix with different metaplectic signs, so a phase-sensitive composition needs a declared convention.[ref-48b962d9e69c][ref-a70ef1deefff]
Manages Complexity¶
An ideal cascade of optical elements can be represented by a product of small ray matrices before choosing the wavefield operator branch. Similarly, a filter can be described by selecting a matrix-indexed analysis domain instead of inventing an unrelated transform for each parameter choice. This compression is useful only while retaining application boundaries: paraxial assumptions for optics, coherent-field phase for interference, and sampling/aliasing conditions for digital signals.[ref-788eec23e862][ref-eb81bdb6e66a]
Abstract Reasoning¶
Specify a conjugate coordinate pair and verify \(ad-bc=1\). Determine whether \(b\) is zero, use the integral or pointwise branch accordingly, and track the metaplectic lift when composing transforms. Compare a proposed use with the model's analytic and physical assumptions before inferring that a formal result applies. The framework organizes Fourier, Fresnel, scaling and chirp operations into one family; it does not prove that every quadratic-looking kernel is unitary or that every selected LCT filter improves restoration.[ref-eb81bdb6e66a][ref-48b962d9e69c][^ref-b3f96aff8883]
Knowledge Transfer¶
The optical and restoration cases share the same real symplectic parameter, phase-plane interpretation, lifted function operator and composition rule. Their purposes differ: one models wave propagation through an optical system; the other selects a representational domain for estimation. Performance and validity must be rechecked when transferring between them. The actual broad parent is Transformation; a still more general operator-lifting skeleton is a future-prime question, not a reason to promote this named LCT beyond its mathematical and physical scope. The live Integral Transform remains a related non-parent under its literal integral-kernel definition because \(b=0\) LCT members act pointwise.[ref-788eec23e862][ref-b3f96aff8883][^ref-eb81bdb6e66a]
[^ref-eb81bdb6e66a]: John J. Healy and Haldun M. Ozaktas, “Sampling and Discrete Linear Canonical Transforms,” in Linear Canonical Transforms: Theory and Applications (2016), §8.1.1. Author-hosted chapter. [^ref-48b962d9e69c]: Kurt Bernardo Wolf, “A Top-Down Account of Linear Canonical Transforms,” SIGMA 8 (2012), 033, §1. Journal PDF. [^ref-788eec23e862]: Stuart A. Collins, “Lens-System Diffraction Integral Written in Terms of Matrix Optics,” Journal of the Optical Society of America 60 (1970), 1168–1177, publisher abstract. Publisher page. [^ref-b3f96aff8883]: Billur Barshan, M. Alper Kutay and Haldun M. Ozaktas, “Optimal Filtering with Linear Canonical Transformations,” Optics Communications 135 (1997), 32–36. Author-hosted PDF. [^ref-a70ef1deefff]: Maurice A. de Gosson, “Paths of Canonical Transformations and their Quantization” (2015), §3. Original preprint.
Relationships to Other Abstractions¶
Current abstraction Linear Canonical Transformation Domain-specific
Parents (1) — more general patterns this builds on
-
Linear Canonical Transformation is a kind of Transformation Prime
An LCT transforms an input function under a rule with preserved canonical structure.
Hierarchy path (1) — routes to 1 parentless root
- Linear Canonical Transformation → Transformation → Function (Mapping)
Neighborhood in Abstraction Space¶
Linear Canonical Transformation sits in a sparse region of the domain-specific corpus (69th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Algebraic Structures, Groups & Operators (42 abstractions)
Nearest neighbors
- Hamiltonian Mechanics — 0.86
- Squeeze Mapping — 0.85
- Symplectic Structure — 0.84
- Vogel Plane — 0.84
- Drazin inverse — 0.83
Computed from structural-signature embeddings · 2026-10-08