View-Guided Beam Focusing

• View-guided beam-focusing is defined by prescribing an observation point or caustic trajectory to design aperture excitation for energy concentration.
• It integrates methodologies from metasurface antennas, XL-arrays, optical PICs, and plasma diagnostics, emphasizing spherical wave compensation and curvature control.
• The approach bridges near-field focusing and far-field steering through analytic scaling laws, addressing practical issues like phase quantization and hardware constraints.

View-guided beam-focusing can be understood as the class of beam-forming and wave-shaping strategies in which a prescribed observation point, range-angle location, propagation trajectory, or measurement geometry determines the aperture excitation, input field, or source geometry so that energy is concentrated at a finite distance or along a designed caustic. In the works considered here, this notion spans near-field focusing with spherical-wave compensation in 2D waveguide-fed metasurface antennas, focal-plane-based angular mapping in hybrid integrated metasurface-photonic circuits, curvature-controlled accelerating and abruptly autofocusing optical beams, turning-point-caustic focusing in Doppler backscattering, geometry-guided focusing of fast-electron beams in solids, and discrete-phase near-field beam-focusing in extremely large-scale arrays (XL-arrays) [(Gavriilidis et al., 5 May 2026); (He et al., 14 Apr 2026); (Goutsoulas et al., 2018); (Ruiz et al., 2024); (Scott, 2015); (Zhang et al., 2024); (Yang et al., 2024); (Li et al., 2012)].

1. Point focusing, steering, and the role of the prescribed “view”

A central distinction in this literature is between focusing at a specific finite-distance point and steering toward an asymptotic angle. In the 2D waveguide-fed metasurface setting, near-field beam focusing is defined as maximizing the radiated field at a finite-distance point $\mathbf s$, with particular emphasis on the on-axis target

$\mathbf s(R) = [0,0,R]^T.$

This differs from classical far-field beam steering because the design requires spherical phase compensation rather than a linear phase ramp, and because performance depends on the range $R$ as well as direction. The associated performance descriptor is therefore not only directionality but also the beam depth along the radial axis (Gavriilidis et al., 5 May 2026).

The same distinction appears in XL-array localization. There, the radiating field is modeled as spherical rather than planar, and the array is tuned so that its response is concentrated around a presumed user position rather than merely an arrival angle. The direct localization formulation exploits curvature-of-arrival (COA), so the “view” is the intended user position in the Fresnel region; analog coefficients are designed to sharpen the position-dependent likelihood surface around that point (Yang et al., 2024).

A related but not identical paradigm appears in the hybrid PIC–metasurface optical platform. That system implements focal-plane-based angular mapping: different emitters on the photonic integrated circuit illuminate different regions of the metasurface, and each illumination position maps to a different output angle. This is beam steering rather than finite-range focusing, but it is view-guided in the sense that the selected emitter and resulting beam footprint determine the output direction through a prescribed optical mapping (He et al., 14 Apr 2026).

2. Electromagnetic aperture formulations and near-field scaling laws

The most explicit aperture-level theory in this corpus is the coupled-dipole model for a 2D waveguide-fed metasurface antenna realized as a parallel-plate waveguide with a single embedded feed and a large number of subwavelength reconfigurable metamaterial elements on the top plate. Because the feed launches a guided wave whose amplitude decays with distance due to propagation, dielectric, and radiation effects, the aperture is naturally amplitude tapered rather than uniformly excited. Each element is described by a polarizability matrix $\mathbf A_n$, mutual interactions are represented by a Green-function coupling matrix $\bar{\mathbf G}$, and the induced magnetic moments satisfy

$$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$

For passive, lossless tuning, the element response is mapped through the Lorentzian phase control law

$$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$

with $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$ and $\varphi_n\in[0,2\pi]$ (Gavriilidis et al., 5 May 2026).

On the array normal, the scattered field simplifies, and the phase choice

$\varphi_n = kR_n(R) - \angle(h_{0,n}^y)$

aligns the coherent part of $\mathbf s(R) = [0,0,R]^T.$0 so that aperture contributions add constructively at the target point. The resulting power-normalized gain is defined by

$\mathbf s(R) = [0,0,R]^T.$1

with

$\mathbf s(R) = [0,0,R]^T.$2

The asymptotic scaling law is

$\mathbf s(R) = [0,0,R]^T.$3

where $\mathbf s(R) = [0,0,R]^T.$4. Hence $\mathbf s(R) = [0,0,R]^T.$5: the power-normalized gain scales linearly with the number of radiating elements. Under fixed current excitation and in the regime $\mathbf s(R) = [0,0,R]^T.$6, the field amplitude instead obeys $\mathbf s(R) = [0,0,R]^T.$7, but the paper identifies power-normalized gain as the physically meaningful metric for aperture comparison (Gavriilidis et al., 5 May 2026).

The same near-field perspective underlies XL-array beam-focusing with continuous and discrete phase shifters. For a target user at location $\mathbf s(R) = [0,0,R]^T.$8, the continuous-phase benchmark sets the analog beamformer equal to the near-field steering vector $\mathbf s(R) = [0,0,R]^T.$9 with element-wise phase

$R$0

When phase shifters are quantized, the quantized phase response $R$1 is expanded as a Fourier series, and the beam pattern becomes a sum over structured harmonics. This yields the paper’s central result that discrete phase shifters introduce additional grating lobes, while the main lobe still focuses at the intended user location. Its beam height is

$R$2

which increases with $R$3 and approaches $R$4 as $R$5, whereas the main-lobe beam-width and beam-depth are reported as unaffected by phase resolution (Zhang et al., 2024).

Near-field localization uses a related but estimator-centered formalism. In the multi-user uplink setting, the received signal is modeled through spherical-wave channel coefficients $R$6, and the maximum-likelihood objective is written in projection form:

$R$7

The analog combiner $R$8 is then optimized to maximize the objective at the true user positions, thereby coupling near-field beam focusing directly to localization accuracy. For hybrid arrays the phase-only solution is the average phase rule

$R$9

while DMAs impose the Lorentzian constraint

$\mathbf A_n$0

and are handled either by projection or by a Riemannian conjugate gradient method on a product of complex circles (Yang et al., 2024).

3. Beam depth, curvature, and caustic-guided focusing

Range selectivity in near-field focusing is quantified in the metasurface literature by a normalized beam-depth gain. If the aperture is designed for $\mathbf A_n$1 while the receiver lies at $\mathbf A_n$2, the normalized beam-depth gain is

$\mathbf A_n$3

Under dense-disk and large-argument Hankel approximations, this reduces to

$\mathbf A_n$4

where

$\mathbf A_n$5

The beam-depth law depends only on the normalized parameter $\mathbf A_n$6, and the far-field-like transition emerges when the upper bound $\mathbf A_n$7 becomes unbounded as

$\mathbf A_n$8

This yields an explicit bridge between near-field range focusing and far-field steering (Gavriilidis et al., 5 May 2026).

A different but structurally related framework appears in paraxial accelerating beams. Starting from the Fresnel diffraction integral with input field $\mathbf A_n$9, stationary-phase analysis gives the ray equation

$\bar{\mathbf G}$0

and the caustic trajectory

$\bar{\mathbf G}$1

For a prescribed convex trajectory $\bar{\mathbf G}$2, the input phase is designed so that tangent rays envelope onto that curve. Near the caustic, the field reduces to an Airy form, and the beam width is

$\bar{\mathbf G}$3

where $\bar{\mathbf G}$4. The result is notable because width depends only on trajectory curvature, while the peak amplitude along the caustic is independently controlled by the input amplitude

$\bar{\mathbf G}$5

For radially symmetric abruptly autofocusing beams, the focal position and peak field are likewise expressed in terms of curvature, slope, and the contributing input radius (Goutsoulas et al., 2018).

In Doppler backscattering, beam focusing is not imposed at the source but generated by propagation near a turning-point caustic in an inhomogeneous plasma. The Gaussian beam-tracing solution shows that the field envelope scales as $\bar{\mathbf G}$6, and the 1D DBS filter satisfies

$\bar{\mathbf G}$7

The scattering contribution is therefore enhanced where the beam width $\bar{\mathbf G}$8 is smallest. The analysis concludes that stronger DBS signal at small launch angle is due to beam focusing near the turning point rather than forward scattering, and that the resulting diagnostic can quantitatively measure the intermediate-to-high $\bar{\mathbf G}$9 spectral exponent but not the spectral peak corresponding to the driving scale of the turbulent cascade (Ruiz et al., 2024).

4. Architectures that realize guided focusing or guided angular control

The hybrid integrated metasurface-photonic platform provides a chip-scale route to guided beam control by concatenating a silicon PIC, freeform micro-optical reflectors, and an ultrawide-FOV metasurface. The conversion pathway is explicitly

$$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$0

The reflectors are fabricated by two-photon polymerization directly on the PIC, convert the guided TE mode into a vertically propagating Gaussian-like free-space beam, and achieve a waveguide-to-free-space coupling efficiency of $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$1 with equivalent insertion loss of $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$2 dB. The metasurface, a single-layer amorphous silicon metasurface on fused silica, is analytically optimized and has an effective focal length of $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$3 mm measured from the reflector top surface at $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$4 nm. Experimentally, the system achieves a measured steering angle of $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$5, corresponding to a $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$6 ultrawide 2D field of view, while maintaining diffraction-limited beam quality; the measured angular FWHM is $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$7 at $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$8 and $$\mathbf m = \left(\bar{\mathbf A}^{-1} - \bar{\mathbf G}\right)^{-1}\mathbf h_f I.$$9 at $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$0 (He et al., 14 Apr 2026).

In solid-density plasma, guided focusing is realized by geometry rather than phase shifters. A focused ultra-high-intensity laser interacting with a convex target front surface of matching radius of curvature launches fast electrons inward because the local $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$1 acceleration direction follows the local Poynting vector of the curved laser wavefront. In 2D PIC simulations with the EPOCH code, a target with radius of curvature $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$2 produces a fast-electron beam that focuses at about $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$3 depth, consistent with the statement that the beam focal length approximately equals the radius of curvature of the laser/target front. The fast electrons are defined as particles with kinetic energy $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$4, and the fast-electron flux is estimated to be $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$5 for a laser intensity of $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$6, giving an absorption fraction of roughly $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$7 (Scott, 2015).

These implementations differ in physical substrate and operating regime, but they share a common design logic: a guided mode, wavefront curvature, or prescribed illumination footprint is transformed into a desired spatial response. This suggests that “guided” focusing is less a single mechanism than a recurring systems principle linking source geometry to the eventual focal or steering behavior.

5. Misconceptions, ambiguities, and hardware-imposed deviations

Several papers in this area are organized around clarifying effects that can be misidentified if focusing is described only phenomenologically. In Doppler backscattering, the enhanced signal at small incident angle had been associated in earlier work with forward scattering, but the beam-tracing analysis shows that the enhancement is already recovered in a purely backscattering description once beam focusing is included. The decisive factor is the inverse-width scaling of the filter function, not an additional forward-scattering channel (Ruiz et al., 2024).

A parallel clarification appears in plasmonics. Surface plasmon polariton beams generated by non-perfectly-matched Bragg diffraction with lateral phase modulation

$$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$8

evolve from focusing-like behavior to an almost straight, nondiffracting-looking line as the phase type changes from $$\mathbf A_n = -\frac{1}{2C}\big(\jmath + e^{\jmath\varphi_n}\big)\mathbf I_2,$$9 to $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$0. The source-to-beam correspondence

$C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$1

shows why appearance can be misleading: for $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$2 the beams are better interpreted as linearly focusing beams rather than true nondiffracting beams, whereas only the special case $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$3 corresponds to the ideal straight-line mapping. The paper therefore resolves the “nondiffracting versus linear focusing” ambiguity by tying the observed beam to its source-to-beam geometry (Li et al., 2012).

Hardware constraints generate a different class of deviations. In XL-arrays with discrete phase shifters, quantization produces extra Fourier harmonics and therefore grating lobes. Type-I grating lobes ($C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$4) remain beam-focusing replicas with their own range-angle maxima, whereas Type-II lobes ($C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$5) behave more like beam-steering lobes without a special focusing range. Numerical results indicate that $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$6-bit and $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$7-bit phase shifters significantly degrade achievable rate, while $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$8-bit phase shifters achieve beam patterns and rate performance close to the continuous-phase case and $C = \frac{k^3}{3\pi}+\frac{k^2}{8h}$9-bit phase shifters are almost identical to continuous phase (Zhang et al., 2024).

Even when quantization is absent, focusing can lose its range selectivity. In the 2D waveguide-fed metasurface, the beam-depth analysis predicts a threshold $\varphi_n\in[0,2\pi]$0 beyond which positive radial shifts no longer significantly change the gain at level $\varphi_n\in[0,2\pi]$1; the system then behaves in a far-field-like way rather than as a range-focusing aperture. In the optical PIC platform, performance is also conditioned by practical limitations such as a total insertion loss excluding the metasurface of $\varphi_n\in[0,2\pi]$2 dB and partial restriction of the measurement range by the goniometer travel range (Gavriilidis et al., 5 May 2026, He et al., 14 Apr 2026).

6. Applications and cross-domain design implications

The application space represented in these works is broad. The hybrid PIC–metasurface platform is positioned for inter-satellite optical links, airborne LiDAR, point-to-point optical wireless communications, and collaborative robotic platforms, where wide 2D angular coverage and diffraction-limited beam quality are both required. The near-field localization framework targets 6G extremely large-scale arrays, where GPS may be unavailable and spherical-wave propagation creates new focusing degrees of freedom for direct localization. In plasma diagnostics, beam focusing changes the weighting of the DBS scattering kernel and therefore the interpretability of the measured turbulence spectrum. In ultra-high-intensity laser-solid interactions, convergent fast-electron beams are discussed in connection with fast ignition, where reduced beam radius and altered hotspot requirements are relevant but remain unproven at ignition scale (He et al., 14 Apr 2026, Yang et al., 2024, Ruiz et al., 2024, Scott, 2015).

Across these domains, several design implications recur. First, non-uniform excitation is not necessarily detrimental: in the waveguide-fed metasurface, the natural amplitude taper produced by guided-wave decay still yields linear power-normalized gain scaling with element count. Second, beam width or range selectivity is often controlled by geometric curvature, whether that curvature is the trajectory curvature of an accelerating beam, the turning-point structure of a plasma caustic, or the physical curvature of a laser wavefront and target surface. Third, implementation constraints determine whether the designed focus survives in practice: Lorentzian tuning constraints in DMAs, discrete phase resolution in XL-arrays, mutual coupling in metasurfaces, and collisions or target deformation in solid-density plasma all modify the realized beam (Gavriilidis et al., 5 May 2026, Goutsoulas et al., 2018, Yang et al., 2024, Zhang et al., 2024, Scott, 2015).

A plausible unifying implication is that view-guided beam-focusing is most naturally formulated as a geometry-constrained inverse problem. The prescribed “view” may be a focal point, a range-angle pair, a caustic trajectory, a user location, or a measurement-sensitive region; the design variables may be aperture phases, input amplitudes, source curvature, or analog combining coefficients; and the success criterion may be gain, beam depth, focal contrast, localization RMSE, or diagnostic selectivity. The cited works show that, once this geometry is made explicit, compact analytic laws often emerge for scaling, focal tolerance, or beam-width control, even in systems with non-uniform illumination, mutual coupling, or reduced-complexity hardware.