Polarization-Crafted Beams

• Polarization-crafted beams are optical fields with spatially varying polarization that enable precise control over wavefront shaping and diffraction.
• They are generated using methods like phase-only SLMs, q-plates, and liquid-crystal devices to tailor amplitude, phase, and polarization simultaneously.
• Applications span optical trapping, microscopy, and nonlinear optics, where customized polarization enhances light–matter interaction and beam propagation.

Polarization-crafted beams are optical fields whose polarization state is deliberately structured across the transverse plane of the beam, so that polarization becomes a primary degree of freedom for beam synthesis, wavefront control, and light–matter interaction. In this usage, the term encompasses vector beams, cylindrical vector beams, Poincaré beams, full Poincaré beams, and related fields with polarization singularities, including C-points and L-lines. Across the literature, polarization crafting appears in several complementary forms: spatially varying Jones-vector fields, superpositions of orthogonally polarized spatial modes, local Jones-matrix transformations, and geometric constructions in which the observable structure is encoded entirely in the polarization degree of freedom (Kottapalli et al., 2022, Chen et al., 2015, Gonzalez-Aceves et al., 13 Jun 2025).

1. Definition and conceptual scope

A structured polarization field is a beam whose polarization state varies across the transverse plane. Instead of a fixed Jones vector $\mathbf{e}_0$, one has a space-dependent Jones vector

$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$

where $(e_x,e_y)$ encodes a local polarization state whose orientation and ellipticity vary with $(x,y)$ (Kottapalli et al., 2022). In a more general non-separable representation, a vector beam may be written as

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$

so that spatial and polarization degrees of freedom are not factorizable into a purely spatial field and a fixed polarization vector (Gonzalez-Aceves et al., 13 Jun 2025).

The class includes several canonical families. Vector beams include radially and azimuthally polarized beams, where the local polarization direction depends on the azimuthal angle. Poincaré beams are beams whose local polarization states map out paths on the Poincaré sphere. Full Poincaré beams are fully correlated vector beams generated through the coaxial superposition of a Laguerre–Gauss and fundamental Gaussian mode with orthogonal polarizations, and contain all possible polarization states on the surface of the Poincaré sphere in a single beam (Kottapalli et al., 2022, Subith et al., 2022). Cylindrical vector beams are vector solutions of Maxwell’s equations that exhibit axial symmetry in amplitude, phase, and polarization, and are frequently written as superpositions of orthogonally polarized first-order Hermite–Gaussian modes (Alj et al., 2015).

A recurrent theme in this literature is the distinction between scalar and vector descriptions. In conventional wavefront shaping, one controls the phase of a scalar optical field and relies on scalar diffraction theory to determine propagation. In polarization-crafted beams, the field is genuinely vectorial, and the measured intensity may depend on projection onto an analyzer through interference of field components. This makes polarization not merely an auxiliary label, but a designable handle for diffraction, mode conversion, nonlinear propagation, and information encoding (Kottapalli et al., 2022, Bouchard et al., 2016).

2. Mathematical representations and polarization-space mappings

A common starting point is the Jones-vector propagation model. At the input plane,

$$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$

and a spatially varying polarization mask is modeled by a local Jones matrix

$$J(x,y)= \begin{pmatrix} J_{xx}(x,y) & J_{xy}(x,y)\ J_{yx}(x,y) & J_{yy}(x,y) \end{pmatrix},$$

so that

$$\mathbf{E}_{\text{mask}}(x,y,0)=J(x,y)\mathbf{E}_0(x,y).$$

Assuming negligible polarization-dependent propagation in free space, each component then propagates according to scalar diffraction, and an analyzer with Jones vector $\mathbf{a}$ produces the detected scalar field

$$E_{\text{det}}(x,y,z)=\mathbf{a}^\dagger\mathbf{E}(x,y,z), \qquad I_{\text{det}}(x,y,z)=|\mathbf{a}^\dagger\mathbf{E}(x,y,z)|^2.$$

This is the basis on which polarization structure can act as an effective diffraction mask after projection (Kottapalli et al., 2022).

A second formulation decomposes a vector beam into two orthogonally polarized complex fields. In a 4-$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$0 architecture with a phase-only SLM, the output field can be written as

$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$1

where the four independent spatially varying parameters are the overall amplitude $$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$2, the scalar phase $$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$3, and the two polarization parameters $$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$4 and $$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$5 that determine the local state of polarization on the Poincaré sphere (Chen et al., 2015).

The associated Stokes parameters are

$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$6

These relations make explicit that a fully coherent monochromatic vector beam in a plane has four real degrees of freedom, and that polarization crafting is mathematically inseparable from amplitude and phase control when the two complex field components are designed independently (Chen et al., 2015).

For full Poincaré beams, the field is often written as a coherent superposition of orthogonally polarized scalar modes. One representative form is

$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$7

or, in another widely used construction,

$$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$8

In these cases, the local polarization state is determined by the local complex ratio of the two orthogonally polarized modal amplitudes, and the beam sweeps the surface of the Poincaré sphere across its cross-section (Subith et al., 2022, Black et al., 2022).

A distinct geometric approach replaces modal superposition with coordinate geometry. Using a conformal map to define curvilinear coordinates, one constructs orthonormal polarization basis vectors $$\mathbf{E}(x,y) = \begin{pmatrix} E_x(x,y) \ E_y(x,y) \end{pmatrix} = A(x,y)e^{i\phi(x,y)} \begin{pmatrix} e_x(x,y) \ e_y(x,y) \end{pmatrix},$$9 and $(e_x,e_y)$0 from the coordinate tangents. In the circular basis,

$(e_x,e_y)$1

so the structure is entirely encoded in the spatially varying polarization angle $(e_x,e_y)$2. In this construction, the scalar amplitude may be almost trivial, while the Stokes field carries the full geometry of ellipses, parabolas, circles, or other conformal coordinate families (Gonzalez-Aceves et al., 13 Jun 2025).

3. Generation platforms and beam-synthesis strategies

Polarization-crafted beams have been generated by phase-only SLMs, q-plates, liquid-crystal devices, biaxial crystals, structured microcavities, and fixed curvilinear holograms. One branch of the field uses a single phase-only SLM in a 4-$(e_x,e_y)$3 system to create two first-order beams with independent amplitudes and phases, convert them into $(e_x,e_y)$4- and $(e_x,e_y)$5-polarized components with half-wave plates, and recombine them coaxially. Because the SLM controls both $(e_x,e_y)$6 and $(e_x,e_y)$7, this architecture permits independent and simultaneous tailoring of amplitude, phase, and complete polarization of vector beams (Chen et al., 2015).

A related but conceptually different SLM strategy uses the device as a polarization mask rather than as a scalar phase hologram. In the Young’s double-slit analog demonstrated in the polarization domain, two strips are assigned polarization rotated by $(e_x,e_y)$8 relative to the background. After propagation and projection by an analyzer, the detected intensity exhibits a double-slit–like interference pattern driven purely by polarization differences. Without the analyzer, the interference pattern disappears or drastically changes because orthogonally polarized components do not interfere in polarization-insensitive detection (Kottapalli et al., 2022). This establishes that a single structured polarization mask can effectively control the observed amplitude distribution after projection.

Liquid-crystal devices are a second major platform. A polar POLICRYPS structure, fabricated by photocuring a polymer–liquid crystal composite with concentric rings, yields annular liquid-crystal regions with radially oriented director. With circularly polarized Gaussian input at $(e_x,e_y)$9 from a He–Ne laser, the birefringent annuli act as a space-variant waveplate and generate a cylindrical vector beam. Because the polymer rings transmit an undisturbed Gaussian component while the liquid-crystal annuli generate the vector part, the output is a hybrid CVB rather than a pure CVB (Alj et al., 2015). The device is explicitly framed as a polarization shaper, or “polshape.”

Spin–orbit conversion in q-plates is another standard method. In warm rubidium experiments, a q-plate tuned to half-wave retardation produces vector vortex beams such as radial, azimuthal, and spiral polarization states, while quarter-wave retardation produces full Poincaré beams with lemon and star topologies. The general vector field is written in the circular basis as

$(x,y)$0

with the two polarization components carrying different Laguerre–Gaussian content (Bouchard et al., 2016). In fiber-based atomic-physics implementations, cylindrical vector beams are generated by exciting higher-order modes in a short section of Corning HI 1060 fiber at $(x,y)$1, followed by waveplate conversion to produce azimuthally varying ellipticity (Fatemi, 2011).

Biaxial-crystal conical refraction provides a passive bulk-optics route. When a focused Gaussian beam is directed along an optic axis of a biaxial crystal, the output field at the focal plane is described by a spatially varying Jones matrix whose scalar functions are Belsky–Khapalyuk–Berry integrals. For circularly polarized input the intensity is azimuthally symmetric, whereas for linearly polarized input the intensity forms a crescent with a dark point where the local polarization is orthogonal to the input. The output beams exhibit C-points, L-lines, and L-circles whose type and orientation are controlled by the input SOP and the parameter $(x,y)$2 (Turpin et al., 2014).

A source-level implementation appears in exciton-polariton microcavities. A polarization-selective subwavelength grating patterned into concentric circles acts as a cavity mirror with high reflectivity for locally radial polarization. In the strong-coupling regime, this selects a single-mode, radially polarized vector vortex beam from a vertical semiconductor microcavity. The emitted spin texture is characterized by polarization-resolved imaging and interferometry, and the device demonstrates direct generation of a radially polarized vector vortex beam without external polarization conversion (Hu et al., 2020).

Finally, a geometric generation route uses a single phase-only SLM programmed with the phase $(x,y)$3 associated with a conformal coordinate system, followed by a quarter-wave plate. This produces curvilinear vector beams whose polarization follows elliptical, parabolic, bipolar, or dipole coordinate lines, with Stokes polarimetry confirming agreement with theory (Gonzalez-Aceves et al., 13 Jun 2025).

4. Propagation, diffraction engineering, and nonlinear control

Polarization-crafted beams are not restricted to static transverse polarization patterns; they are also used to engineer diffraction and propagation. The most direct example is polarization-based diffraction control. If a spatially varying Jones transformation $(x,y)$4 is chosen such that $(x,y)$5 has the desired effective complex amplitude, then after projection through analyzer $(x,y)$6 the beam behaves as if it had been modulated in amplitude and phase. This suggests that polarization masks provide more design freedom than phase-only masks, since a lossless local Jones transformation has the degrees of freedom of $(x,y)$7, while a phase-only SLM provides one scalar degree of freedom per pixel (Kottapalli et al., 2022).

Polarization can also be controlled along the propagation axis. In the Frozen Waves framework, a vector beam is built as a superposition of co-propagating Bessel beams with coefficients chosen from closed-form expressions to enforce a prescribed longitudinal envelope $(x,y)$8. The resulting diffraction-attenuation-resistant beams permit arbitrary control of the state of polarization and the intensity along the propagation direction. Experimental demonstrations include a beam whose SoP evolves from horizontally polarized to radial polarization and then to vertical polarization while the beam intensity is held constant, as well as beams whose central ring is switched off over predefined space regions, generating multiple foci with different SoP and at different intensity levels (Corato-Zanarella et al., 2017).

Nonlinear propagation introduces a different use of polarization crafting. In a hot rubidium vapor cell with a saturable self-focusing nonlinearity, scalar Laguerre–Gaussian vortex beams exhibit azimuthal modulational instability and break into filaments, whereas radially symmetric vector beams and full Poincaré beams with lemon and star topologies do not exhibit beam breakup over the experimental propagation length while still showing nonlinear confinement and self-focusing. The propagation is modeled by coupled nonlinear Schrödinger-type equations for the two circular polarization components, with cross-phase modulation and saturation (Bouchard et al., 2016). This suggests that tailoring spatial polarization can effectively control modulational instability channels.

A complementary nonlinear result concerns optical rogue waves and nonlinear caustics. In a saturable self-focusing rubidium vapor at $(x,y)$9, a full-Poincaré lemon beam constructed from $$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$0 and $$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$1 with orthogonal circular polarizations suppresses nonlinear caustic formation relative to scalar and uniformly polarized comparison beams with the same phase noise. The field is written as

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$2

and the coupled propagation is governed by

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$3

with saturable susceptibility including cross-phase modulation. In the nonlinear regime at 50–90 mW, the full-Poincaré beam maintains statistics consistent with exponential speckle, while the scalar and false-Poincaré beams exhibit long-tailed intensity distributions with $$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$4 (Black et al., 2022).

Propagation can also reshape polarization statistics in free space. For Bessel–Poincaré beams, the two orthogonally polarized Bessel–Gauss components have the same far-field intensity distribution because the angular-spectrum ring radius is independent of azimuthal index. As a result, the far field becomes linearly polarized everywhere, and the beam behaves as a free-space polarization attractor transforming elliptical polarizations to linear polarizations (Lopez-Mago, 2019). In a different topological direction, tightly focused higher-order full-Poincaré beams and standing-wave configurations can be used to craft Bloch and Néel skyrmionic structures either in the electric spin vector for near-circular fields or in the polarization major-axis direction for near-linear fields (Gutiérrez-Cuevas et al., 2021).

5. Characterization, singularities, and statistical descriptors

The dominant characterization framework is Stokes polarimetry. Across the literature, the local polarization state is reconstructed from intensity measurements through sets of linear and circular analyzers. Representative formulas include

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$5

for cylindrical vector beams generated by POLICRYPS devices (Alj et al., 2015), and

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$6

for curvilinear vector beams measured with six analyzer settings (Gonzalez-Aceves et al., 13 Jun 2025). The local polarization angle may then be obtained from

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$7

or, in equivalent notation,

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$8

A second set of descriptors concerns overall vectorness. For locally fully polarized beams, the global degree of polarization and the summed Stokes variances satisfy

$$\mathbf{E}(\mathbf{r}) = \int d\boldsymbol{\alpha}\sum_{u=1,2} c_{\boldsymbol{\alpha},u}\,\psi_{\alpha}(\mathbf{r})\,\hat{e}_{u}(\mathbf{r}),$$9

where the $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$0 are intensity-weighted variances of the normalized Stokes parameters. This relation is used to classify full Poincaré, cylindrical vector, and Lambert–Poincaré beams, and it highlights, for example, that Lambert–Poincaré patterns have

$$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$1

consistent with uniform distribution of polarization states over the Poincaré sphere (Lopez-Mago, 2019).

The Vector Quality Factor is used in geometry-driven vector beams to quantify spatial inhomogeneity: $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$2 with $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$3. Parabolic and dipole systems reach $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$4, while elliptical and bipolar systems show a tunable decrease with semi-focal distance (Gonzalez-Aceves et al., 13 Jun 2025).

Polarization-crafted beams also support a rich singular-optics vocabulary. Full Poincaré beams exhibit C-points and L-lines; conical-refraction beams produce central C-points for circular input and L-lines for linear input; higher-order full-Poincaré beams exhibit multiple C-point pairs whose number scales with the vortex order; and in second-harmonic generation of FP beams the number of C-points and L-lines doubles because the vortex order doubles in the nonlinear process (Turpin et al., 2014, Subith et al., 2022). In full-Poincaré-beam SHG with two contiguous BIBO crystals, Stokes phases

$$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$5

and their arguments are used to identify Stokes vortices and to track the doubling of orbital angular momentum and polarization singularities (Subith et al., 2022).

Scattering introduces a statistical variant of polarization crafting. When Poincaré beams scatter from a ground glass plate, the resulting polarization speckles are locally fully polarized but spatially random. The scalar speckle size is characterized by the autocorrelation of $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$6,

$$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$7

while the polarization speckle size is defined through

$$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$8

The scalar speckle size is independent of vector-beam index, whereas the polarization speckle size decreases as the vector-beam index increases (Reddy et al., 2017).

6. Applications, limitations, and current directions

The application space is broad because polarization-crafted beams couple polarization and spatial structure at the source. Recurrent applications include optical trapping and micromanipulation, high-resolution microscopy, atom guiding, materials processing, and optical communications (Kottapalli et al., 2022, Corato-Zanarella et al., 2017). Cylindrical vector beams generated by POLICRYPS are linked to tighter focusing for radial polarization, donut-shaped focal distributions for azimuthal polarization, and particle trapping scenarios that depend on whether the trapped particles are metallic or have dielectric constant lower than the surrounding medium (Alj et al., 2015). In warm rubidium vapor, cylindrical vector beams serve as single-shot polarization-dependent probes and as pumps that write spatially varying spin patterns into the medium (Fatemi, 2011).

Nonlinear frequency conversion extends polarization crafting across wavelength. Using two contiguous BIBO crystals with orthogonal optic axes, full-Poincaré beams at 810 nm are frequency doubled to 405 nm with average power as high as 18.3 mW at a single-pass conversion efficiency of 2.19%. The second-harmonic beam remains a full Poincaré beam, while OAM, C-points, and L-lines double in a controlled way. The work also introduces a polarization-coverage metric based on occupancy of buckets in the $$\mathbf{E}_0(x,y)=A_0(x,y)e^{i\phi_0(x,y)}\mathbf{e}_0,$$9 plane, and finds that the SH FP beam has the highest polarization coverage for the pump beam having equal intensity weightage of the constituent beams (Subith et al., 2022).

Information encoding and hidden-structure readout are emerging directions. In partially polarized vector beams engineered to populate the equatorial disk of the Poincaré sphere, a carrier-dependent spatial mask can retrieve parametric curves encoded in the spatial–polarization mapping. The beam is described statistically by a local coherence matrix

$$J(x,y)= \begin{pmatrix} J_{xx}(x,y) & J_{xy}(x,y)\ J_{yx}(x,y) & J_{yy}(x,y) \end{pmatrix},$$0

with degree of polarization

$$J(x,y)= \begin{pmatrix} J_{xx}(x,y) & J_{xy}(x,y)\ J_{yx}(x,y) & J_{yy}(x,y) \end{pmatrix},$$1

This suggests a polarization-domain analogue of steganography in which information is not carried by scalar intensity or phase but by the engineered mapping between transverse coordinates and Stokes vectors (Tena-Piñon et al., 12 Mar 2026).

Several limitations recur across platforms. Many implementations are wavelength-specific because liquid-crystal retardance, metasurface response, and birefringent phase matching are chromatic (Alj et al., 2015, Kottapalli et al., 2022). Input polarization sensitivity is often critical, especially when the crafted field is produced by a local Jones transformation or spin–orbit converter (Kottapalli et al., 2022). Vector-field propagation and inverse design are computationally heavier than scalar phase retrieval because they double the field dimension and require enforcement of polarization constraints (Kottapalli et al., 2022). In nonlinear control applications, suppression of caustics or modulational instability is effective only within a medium-dependent power window (Black et al., 2022, Bouchard et al., 2016). In source-level devices such as the exciton-polariton microcavity, the realized mode family is fixed by the cavity design, so reconfigurability is more limited than in SLM-based systems (Hu et al., 2020).

A plausible implication is that the field is converging on two complementary design philosophies. One uses modal superposition and analyzer projection to convert vectorial degrees of freedom into effective scalar wavefront control; the other uses geometric or material patterning to encode the beam structure directly in the polarization degree of freedom. Together, these approaches define polarization-crafted beams as a general framework for engineering light fields in which the transverse polarization map is not secondary to amplitude and phase, but is itself the central synthesis variable (Kottapalli et al., 2022, Gonzalez-Aceves et al., 13 Jun 2025).