Exact Self-Imaging with Arbitrary Revival Spacings

Technical Summary: Exact Self-Imaging with Arbitrary Revival Spacings

Problem Statement
Self-imaging, or the Talbot effect, is a fundamental phenomenon in paraxial wave propagation where a wavefield with transverse periodic structure undergoes exact reconstructions at specific propagation distances. Historically, since Talbot's observation in 1836 and Rayleigh's derivation in 1881, the recurrence planes of self-imaging have been understood to be uniformly spaced along the physical propagation axis ($z$) in free space. While numerous generalizations have expanded the effect to aperiodic structures, rotating planes, and different degrees of freedom (temporal, angular), the spacing of the recurrence sequence itself has remained fixed for a given system geometry. The central problem addressed by this work is the lack of a general principle allowing the physical recurrence planes to be prescribed as an arbitrary axial sequence (e.g., accelerating, decelerating, or following complex trajectories) while preserving the condition of exact self-imaging.

Methodology
The authors reformulate the general phenomenon of self-imaging within the framework of canonical phase-space geometry.

1. Theoretical Framework: They treat paraxial wave evolution as a canonical transformation on the transverse phase space $(x, k_x)$, governed by a quadratic Hamiltonian. Propagation through any first-order (ABCD) system is represented by a symplectic matrix $M$. The authors identify a specific off-diagonal matrix element, $B = M_{12}$, as the canonical recurrence coordinate. In this coordinate, the quadratic phase accumulated by transverse spatial-frequency components advances linearly.
2. Canonical vs. Physical Separation: The paper posits that exact self-imaging recurrences are intrinsically periodic and uniform in the canonical coordinate $B$, but the physical propagation axis $z$ is merely a physical embedding of this coordinate. The relationship is defined by a mapping $B = B(z)$.
3. Experimental Implementation: To demonstrate that the physical recurrence spacing can be arbitrarily programmed, the authors use a single programmable Spatial Light Modulator (SLM).
   • A monochromatic laser ($\lambda = 635$ nm) illuminates a periodic grating pattern encoded on the SLM.
   • The SLM imprints a controlled transverse phase profile $\phi(x)$ onto the field. This phase profile is calculated to assign specific order-dependent phases to the discrete spatial-frequency components of the Talbot spectrum.
   • By programming these phases, the authors effectively define the embedding function $B(z)$, mapping the uniform canonical recurrence lattice to a non-uniform physical sequence $z_\ell$.
   • The resulting field is propagated in free space, and the intensity distribution $I(x, y=0, z)$ is recorded via axial scanning with a CCD camera.

Key Contributions

• General Law of Self-Imaging: The paper establishes that the rigidity of uniform recurrence spacing is not fundamental to self-imaging but arises from the implicit assumption that the physical coordinate $z$ is the natural evolution parameter. The true invariant coordinate is the canonical $B$.
• Arbitrary Axial Control: The authors demonstrate that by treating $z$ as a physical coordinate where the canonical lattice is observed, rather than the recurrence coordinate itself, one can prescribe arbitrary axial trajectories for self-imaging planes.
• Experimental Realization: The work provides the first experimental demonstration of "sculpted" Talbot carpets where recurrence spacings accelerate, decelerate, and follow polynomial, exponential, and sinusoidal trajectories, all while maintaining exact self-imaging conditions.

Results
The experiments successfully verified the theoretical predictions across various embedding functions:

• Uniform Spacing: A standard grating with no imposed curvature ($B(z) = z$) produced uniformly spaced Talbot planes, serving as a baseline.
• Accelerating/Decelerating Spacings: By imprinting quadratic phases corresponding to positive and negative focal lengths, the authors induced nonlinear embeddings. Positive curvature compressed the recurrence intervals (accelerating Talbot effect), while negative curvature expanded them (decelerating effect).
• Complex Trajectories: Using arbitrary phase programming, the team realized recurrence planes following:
  • Quadratic and cubic acceleration.
  • Fractional-power acceleration ($z_\ell \propto \ell^{5/4}$).
  • Exponential acceleration ($z_\ell \propto e^{\alpha \ell}$).
  • Sinusoidal modulation of the spacing.
    In all cases, the measured positions of the self-imaging planes $z_\ell$ matched the predicted values derived from the inverse mapping $z_\ell = B^{-1}(\ell \Delta B)$, confirming that the underlying canonical recurrence remained uniform while the physical manifestation was fully programmable.

Significance
The paper claims to reveal a "hidden geometric freedom" in paraxial wave propagation. The significance lies in the decoupling of the invariant recurrence condition (rigid in canonical coordinates) from its physical observation (flexible in real space).

• Unification: This perspective unifies previously distinct phenomena, such as modified Talbot distances in curved wavefronts or non-Cartesian geometries, as different physical embeddings of the same underlying canonical recurrence.
• New Degrees of Control: It transforms self-imaging from a fixed diffraction effect into a versatile platform for exploring canonical geometry and programmable Hamiltonian evolution.
• Broader Applicability: The authors note that because the paraxial wave equation is mathematically equivalent to the 2D Schrödinger equation, these findings suggest routes for analog simulations of quantum revival engineering and potential applications in acoustics, electron optics, and spatiotemporal analogues of the Talbot effect. Furthermore, the ability to discretize and warp the canonical coordinate suggests potential connections to number theory and spectral engineering via optical implementations of arithmetic transforms.