Polarization Index: Cross-Disciplinary Metrics

• Polarization index is a family of measurement constructs that compress complex polarization-relevant structures into scalar or low-dimensional diagnostics depending on the context.
• It is applied in singular optics, astrophysical emission, device photonics, ferroelectric mapping, and social-science opinion analysis to classify and quantify polarization phenomena.
• Key methodologies include topological winding numbers, intensity ratios, Bayesian inference, and transport distance metrics, each capturing distinct modal, spectral, and structural features.

Polarization index denotes a family of non-equivalent quantitative constructs rather than a single universal observable. In the literature, the term can refer to a topological winding number of a far-field polarization vortex, a Stokes-index associated with polarization singularities, a linear polarization degree or polarized spectral index in scattering and astrophysical measurements, a pseudosymmetry-sensitive confidence index for ferroelectric domain mapping, a chromospheric polar-network proxy for the solar polar magnetic field, or a formal measure of social and political polarization based on group separation, multimodality, or transport distance (Nazarov et al., 21 Jul 2025, Gangwar et al., 7 Sep 2025, Barreda et al., 2016, Jew et al., 2019, Griesbach et al., 14 Jan 2026, Mishra et al., 3 Feb 2025, Lee et al., 2024). The unifying feature is that each quantity compresses a polarization-relevant structure into a scalar or low-dimensional diagnostic, but the object being measured, the geometry of the state space, and the interpretation differ sharply across disciplines.

1. Terminological scope and taxonomic structure

Several papers explicitly show that “polarization index” is context dependent. In bound-state-in-the-continuum sensing, the relevant quantity is not a separately named index but the polarization winding number or topological charge of the far-field vortex (Nazarov et al., 21 Jul 2025). In low-index-contrast photonic crystals, the analysis is likewise not built around a standalone scalar index, but around TE/TM dispersion mismatch, modal-width mismatch, field-amplitude imbalance, and optical chirality enhancement (Barolak et al., 17 Jul 2025). In terahertz emission from Bi-2212 cross-whisker intrinsic Josephson junctions, the effective polarization-based mode indicator is the fitted field-amplitude ratio $E_{0x}/E_{0y}$, not a conventional degree-of-polarization index (Saito et al., 2022).

Domain	Quantity used as polarization index	Defining form
BIC and polarization singularities	Topological charge $q$, Stokes index $\sigma_{12}$	Phase winding; $\sigma_{12}=q-p$
Scattering and device photonics	$P_L(\theta)$, $S$, $E_{0x}/E_{0y}$	Polarization-degree or anisotropy metrics
Polarized astrophysical emission	$\beta$, $\alpha$, $\Pi$, $q$0	Spectral exponents or polarization fractions
Ferroelectrics and solar diagnostics	$q$1, PNI	Confidence/proxy indices
Social and political science	$q$2, $q$3, $q$4, $q$5	Separation, multimodality, or transport-based measures

This diversity is not merely terminological. It reflects different underlying mathematical objects: homotopy-like winding in momentum space, Stokes-field singularity indices, signal ratios, Bayesian posterior-shape functionals, transport distances on opinion distributions, and image-derived proxies for magnetic structure. A plausible implication is that cross-disciplinary use of the phrase requires immediate specification of the state space and the physical or statistical symmetry being probed.

2. Topological indices in singular optics and BIC photonics

In the BIC literature, the central polarization-based index is the topological charge of a far-field polarization vortex. For the one-dimensional periodic rod array studied in "Polarization Vortex for Enhanced Refractive Index Sensing" (Nazarov et al., 21 Jul 2025), the charge is defined from the phase winding of the far-field polarization vector: $q$6 The physical content is that the far-field polarization ellipses rotate around a singular point in momentum space, so the polarization angle accumulates a net $q$7 around a closed loop. For that structure, the allowed absolute value is restricted to $q$8; symmetry-protected and tunable off-$q$9 BICs carry $\sigma_{12}$0, whereas the at-$\sigma_{12}$1 BIC highlighted there is topologically trivial and can disappear under detuning (Nazarov et al., 21 Jul 2025).

The same paper ties the index to sensing. Changing the surrounding refractive index $\sigma_{12}$2 shifts the BIC vortex position in $\sigma_{12}$3-space, and near $\sigma_{12}$4 the angular displacement obeys

$\sigma_{12}$5

That square-root law is described as analogous to exceptional-point scaling, even though the system is not non-Hermitian in the usual EP sense. The reported angular sensitivity can substantially exceed the spectral sensitivity, with mode-II BIC($\sigma_{12}$6) angular sensitivity near $\sigma_{12}$7 tending to $\sigma_{12}$8, while the spectral response can include both conventional red shifts and an unusual blue shift depending on the BIC regime (Nazarov et al., 21 Jul 2025).

A related but distinct singular-optics usage appears in "Interferometric method to detect degenerate index states of polarization singularities" (Gangwar et al., 7 Sep 2025). There the polarization singular beam is modeled as

$\sigma_{12}$9

and the Stokes index is

$\sigma_{12}=q-p$0

For C-points, the polarization index is given by the winding of the polarization azimuth $\sigma_{12}=q-p$1, and the complex Stokes field $\sigma_{12}=q-p$2 has phase $\sigma_{12}=q-p$3, so phase singularities of $\sigma_{12}=q-p$4 correspond directly to polarization singularities (Gangwar et al., 7 Sep 2025). The paper’s central problem is index degeneracy: different $\sigma_{12}=q-p$5 pairs can yield the same polarization index. The proposed interferometric method resolves that ambiguity by reading the lower-order charge from the central spiral fringe, the charge difference $\sigma_{12}=q-p$6 from peripheral fork fringes, and the spin assignment from a $\sigma_{12}=q-p$7 projection onto the $\sigma_{12}=q-p$8 linear basis (Gangwar et al., 7 Sep 2025).

Taken together, these papers treat a polarization index as a winding invariant attached to a singularity rather than as a mere intensity-derived summary statistic. In this usage, the index classifies robustness, degeneracy, and motion of singular states under perturbation.

3. Spectral, scattering, and device-level polarimetric indices

In several experimental photonics and astrophysical settings, the relevant index is defined through polarization-resolved intensities or frequency scaling rather than topology. In the high-refractive-index dimer beam-splitter configuration, the governing quantity is the linear polarization degree

$\sigma_{12}=q-p$9

with $P_L(\theta)$0 and $P_L(\theta)$1 the scattered intensities for incident fields polarized perpendicular and parallel to the scattering plane, respectively (Barreda et al., 2016). At $P_L(\theta)$2, $P_L(\theta)$3 becomes the switching indicator: for the small-gap dimer at the switching frequency, $P_L(\theta)$4, $P_L(\theta)$5 is maximal, and $P_L(\theta)$6, which the paper interprets as a pure magnetic scatterer and a null-intensity off-state (Barreda et al., 2016).

In cosmic microwave and radio polarization analysis, “polarization index” can instead mean a polarized spectral index. In "The spectral index of polarized diffuse Galactic emission between 30 and 44 GHz" (Jew et al., 2019), the spectral index in each pixel is

$P_L(\theta)$7

where $P_L(\theta)$8 and $P_L(\theta)$9 are the true polarized intensities at the two frequencies. The inference is Bayesian, with a Rician likelihood for $S$0, an objective reference prior, and a maximum-entropy prior informed by total-intensity data and an assumed polarization fraction of $S$1 (Jew et al., 2019). The sky-average result is $S$2, while clustered detections yield $S$3; the northern Fermi bubble is much harder than the southern bubble, with a very large Bayes factor favoring different mean spectral indices (Jew et al., 2019).

A third device-level usage appears in terahertz emission from Bi-2212 cross-whisker intrinsic Josephson junctions. There the emission passed through a rotating wire-grid polarizer is fitted as

$S$4

and the field-amplitude ratio $S$5 acts as the polarization-based mode index (Saito et al., 2022). The reported values $S$6, $S$7, and $S$8 are used to assign TM$S$9 or TM$E_{0x}/E_{0y}$0 dominance, and the subsequent cavity-resonance analysis yields $E_{0x}/E_{0y}$1, larger than the conventional bulk value $E_{0x}/E_{0y}$2 (Saito et al., 2022).

These examples show that an index may encode mode selectivity, spectral hardness, or switching state without being topological. The common operation is reduction of polarization-resolved measurements to a parameter with direct interpretive utility.

4. Confidence indices and proxy indices in materials and solar physics

In ferroelectric EBSD, the polarization-related quantity is a classification confidence rather than a direct polarization magnitude. "Ferroelectric polarization mapping through pseudosymmetry-sensitive EBSD reindexing" (Griesbach et al., 14 Jan 2026) introduces a pseudosymmetry confidence index $E_{0x}/E_{0y}$3 to distinguish the six polarization directions in tetragonal ferroelectrics, despite the small $E_{0x}/E_{0y}$4 ratio and near-degenerate Kikuchi patterns. The method replaces ordinary similarity scoring by weighted cross correlation,

$E_{0x}/E_{0y}$5

with weights $E_{0x}/E_{0y}$6 emphasizing pixels where simulated variants differ (Griesbach et al., 14 Jan 2026). The penalty term

$E_{0x}/E_{0y}$7

measures inconsistency between theoretical and experimental variant neighborhoods, and the final index is

$E_{0x}/E_{0y}$8

The reported outcome is a practical domain-mapping tool: low $E_{0x}/E_{0y}$9 appears at boundaries and fine laminates, whereas domain interiors typically lie around $\beta$0 in BaTiO$\beta$1 and around $\beta$2 to $\beta$3 in PZT (Griesbach et al., 14 Jan 2026).

In solar physics, the Polar Network Index is a proxy for a different kind of polarization-related structure: the Sun’s polar magnetic field. "Ca II K Polar Network Index of the Sun" (Mishra et al., 3 Feb 2025) defines

$\beta$4

with the count taken over polar caps in the latitude range $\beta$5 after adaptive threshold segmentation of Ca II K images (Mishra et al., 3 Feb 2025). The threshold is

$\beta$6

and the resulting PNI from Kodaikanal Solar Observatory and Rome-PSPT data correlates strongly with direct Wilcox Solar Observatory polar-field measurements and with an Advective Flux Transport Model (Mishra et al., 3 Feb 2025). The paper uses the index to reconstruct the polar magnetic field continuously from 1904 to 2022, covering solar cycles 14–24 (Mishra et al., 3 Feb 2025).

The two cases share a structural property: the index is not itself the underlying physical polarization state. Rather, it is an estimator or confidence-weighted surrogate derived from indirect observables.

5. Social and political-science polarization indices

In social-science usage, a polarization index is usually an explicitly axiomatized function of a distribution over opinions, groups, or latent rater effects. "Measuring Political Polarization: Twitter shows the two sides of Venezuela" (Morales et al., 2015) defines perfect polarization as two equally sized groups with opposite opinions and constructs a network-based opinion model in which elite opinions are fixed at $\beta$7 and $\beta$8, listeners begin at $\beta$9, and opinions are iteratively averaged over incoming neighbors. From the induced density $\alpha$0, the polarization index is

$\alpha$1

where $\alpha$2 measures imbalance in group sizes and $\alpha$3 is the normalized distance between the gravity centers of the negative and positive poles (Morales et al., 2015). The index lies in $\alpha$4, equals $\alpha$5 for two balanced maximally separated camps, and distinguishes ideological polarization from mere network segregation (Morales et al., 2015).

A different statistical object is measured in "Mixture polarization in inter-rater agreement analysis: a Bayesian nonparametric index" (Mignemi et al., 2023). There the aim is to quantify latent disagreement among raters through the shape of the posterior distribution of random effects under a Dirichlet Process Mixture prior. The proposed index is

$\alpha$6

with $\alpha$7 the modes and $\alpha$8 the antimodes of a kernel density estimate of the posterior random-effects distribution (Mignemi et al., 2023). Large $\alpha$9 indicates clear multimodality with deep valleys between peaks. The paper’s explicit contrast is with the intra-class correlation coefficient,

$\Pi$0

which measures reliability rather than latent polarization (Mignemi et al., 2023).

In multidimensional political-space models, "Multidimensional Polarization Index and its Application to an Analysis of the Russian State Duma" (Aleskerov et al., 2016) generalizes the Aleskerov–Golubenko idea by representing each group $\Pi$1 as a weighted point $\Pi$2 with weight $\Pi$3, center of mass $\Pi$4, and polarization

$\Pi$5

Euclidean, Manhattan, and Chebychev versions are given, the maximum equals $\Pi$6 after normalization, and the empirical application finds the Russian State Duma least polarized in 1995 and most polarized in 2003 (Aleskerov et al., 2016).

Transport-based formalization appears in "The Wasserstein Bipolarization Index" (Lee et al., 2024). The observed opinion distribution $\Pi$7 is compared to a maximally separated distribution $\Pi$8 placing all mass at the two endpoints of the scale, and the basic measure is the $\Pi$9-Wasserstein distance

$q$00

After rescaling, the index becomes

$q$01

so values closer to $q$02 indicate higher bipolarization (Lee et al., 2024). The paper argues that variance and Sarle’s bimodality coefficient fail to satisfy one or more axioms, whereas the Wasserstein construction satisfies distributional invariance, affine equivariance, spread sensitivity, clustering sensitivity, and metric structure (Lee et al., 2024).

Finally, "Revisiting the Measurement of Polarization" (Crespo et al., 24 Nov 2025) reworks the Esteban–Ray framework for discrete populations. Polarization is written as

$q$03

and the main theorem states that Condition H holds if and only if

$q$04

with $q$05 and $q$06 continuous and non-decreasing, $q$07, $q$08 for $q$09 (Crespo et al., 24 Nov 2025). This broadens the admissible class beyond the classical linear-distance case and is reported to avoid counter-intuitive rankings in examples where the standard ER form can fail (Crespo et al., 24 Nov 2025).

6. Recurrent themes, misconceptions, and comparative interpretation

A recurring misconception is that polarization index always denotes a scalar measuring “how polarized” a system is in one generic sense. The surveyed literature does not support that view. Some papers explicitly reject a standalone scalar definition and instead use a related topological or anisotropy quantity: the BIC work identifies the polarization winding number $q$10 as the relevant measure (Nazarov et al., 21 Jul 2025), the low-index-contrast Bloch-surface-wave study characterizes polarization anisotropy through TE/TM mismatch and uses optical chirality

$q$11

as the practical figure of merit (Barolak et al., 17 Jul 2025), and the THz-emission work uses $q$12 as a polarization-based mode index rather than a degree-of-polarization metric (Saito et al., 2022).

A second misconception is that familiar dispersion measures can be substituted for polarization. In public-opinion work, variance and Sarle’s bimodality coefficient are explicitly criticized for failing to capture the joint role of extremity and clustering (Lee et al., 2024). In inter-rater analysis, ICC is said to summarize reliability or consistency, not latent subgroup polarization (Mignemi et al., 2023). In the Twitter study, network segregation alone is insufficient because polarization requires two opposing opinion camps, not merely disconnected communities (Morales et al., 2015).

The surveyed literature suggests that any rigorous use of the term should specify at least four elements: the state space in which polarization is defined, the mathematical operation used to compress structure into an index, the reference configuration relative to which the index is interpreted, and whether the quantity is intrinsic, proxy-based, or confidence-based. Without those specifications, the phrase “polarization index” remains formally ambiguous even within a single broad area such as optics or political science.