---
title: Two-Photon Polarization Correlation Tensor
url: https://www.emergentmind.com/topics/two-photon-polarization-correlation-tensor
type: topic
---

# Two-Photon Polarization Correlation Tensor

Searching arXiv for recent papers on two-photon polarization correlation tensors and closely related formulations.
The two-photon polarization correlation tensor is a tensorial representation of joint polarization structure in a biphoton system. In the Pauli-tensor formulation developed for quadratic quantum polarimetry, it is the \(4\times 4\) real tensor
\[
T_{ij}=\mathrm{Tr}\!\left[\rho\,(\sigma_i\otimes\sigma_j)\right],\qquad i,j=0,1,2,3,
\]
with \(\rho\) the two-photon density operator and \(\{\sigma_0,\sigma_1,\sigma_2,\sigma_3\}\) the identity and Pauli operators. Related literatures use closely allied four-index objects built from scattering amplitudes, field-correlation functions, or polarization-resolved \(g^{(2)}\) observables. In each case, the tensor carries the complete two-photon polarization information relevant to the chosen measurement model, and it is the natural object through which higher-order polarization effects, coincidence structure, and channel action are expressed [2604.09257].

## 1. Definitions and operator structure

In the quantum Stokes formalism, the single-photon polarization Hilbert space \(H_1\) is equipped with four Hermitian operators
\[
\sigma_0=I=|H\rangle\langle H|+|V\rangle\langle V|,\quad
\sigma_1=|H\rangle\langle V|+|V\rangle\langle H|,
\]
\[
\sigma_2=-i|H\rangle\langle V|+i|V\rangle\langle H|,\quad
\sigma_3=|H\rangle\langle H|-|V\rangle\langle V|,
\]
which satisfy \(\mathrm{Tr}[\sigma_i\sigma_j]=2\delta_{ij}\). For two photons, the tensor-product set \(\sigma_i\otimes\sigma_j\), \(i,j=0,1,2,3\), spans the \(4\times 4\) operator space on \(H_1\otimes H_1\). The corresponding correlation tensor is
\[
T_{ij}\equiv \mathrm{Tr}\!\left[\rho\,(\sigma_i\otimes\sigma_j)\right].
\]
Its \(3\times 3\) Pauli subtensor \(T_{ab}\), \(a,b=1,2,3\), contains all joint polarization correlations between the Pauli observables of photon 1 and photon 2, while \(T_{00}=1\) encodes normalization [2604.09257].

In the Pauli-basis expansion of the full two-photon density operator,
\[
\rho=\frac14\Bigl[I\otimes I+\sum_i a_i(\sigma_i\otimes I)+\sum_j b_j(I\otimes\sigma_j)+\sum_{i,j}T_{ij}(\sigma_i\otimes\sigma_j)\Bigr],
\]
the coefficients \(a_i\) and \(b_j\) are single-photon polarization vectors and \(T_{ij}\) are the polarization-correlation tensor components. Alexanian and Mkrtchian use this form for Landau’s two-photon spherical states, where the tensor is read off directly from the density matrix in the helicity or Pauli basis [2109.14984].

The term also appears in four-index forms. In polarization-resolved cavity photon statistics, Kass et al. define
\[
T_{\mu\mu';\nu\nu'}\equiv \sum_f
\langle0;\mathrm{in},\nu;\mathrm{in},\nu'|S^\dagger|f;\mu,\mu'\rangle
\langle f;\mu,\mu'|S|0;\mathrm{in},\nu;\mathrm{in},\nu'\rangle,
\]
a rank-\((2\otimes2)\times(2\otimes2)\) object indexed by two input and two output polarizations. In broadband spontaneous four-wave mixing, the measurable fourth-order correlation tensor is
\[
G_{ijkl}(\omega_s,\omega_i)
\equiv
\langle0|E_i^{(-)}(\omega_s)E_j^{(-)}(\omega_i)E_k^{(+)}(\omega_i)E_l^{(+)}(\omega_s)|0\rangle
\propto \Psi_{ij}(\omega_s,\omega_i)\Psi^*_{kl}(\omega_s,\omega_i),
\]
which is the field-correlation analogue of a polarization tensor [2606.11550].

| Setting | Tensor | Defining expression |
|---|---|---|
| Two-photon Pauli correlator | \(T_{ij}\) | \(\mathrm{Tr}[\rho(\sigma_i\otimes\sigma_j)]\) |
| Polarization-resolved scattering | \(T_{\mu\mu';\nu\nu'}\) | \(\sum_f \langle\cdots|S^\dagger|f;\mu,\mu'\rangle\langle f;\mu,\mu'|S|\cdots\rangle\) |
| Fourth-order field correlation | \(G_{ijkl}(\omega_s,\omega_i)\) | \(\langle E^{(-)}E^{(-)}E^{(+)}E^{(+)}\rangle\propto \Psi_{ij}\Psi_{kl}^*\) |

A recurrent point in the literature is that the phrase “two-photon polarization correlation tensor” does not denote a single universal indexed object. It denotes the tensorial carrier of two-photon polarization information appropriate to the representation being used: density operator, scattering operator, or field correlation function.

## 2. Quadratic channel action and Mueller congruence

The central structural result of quadratic quantum polarimetry is that a single-photon Mueller matrix acts bilinearly on the two-photon tensor when both photons traverse the same medium simultaneously. If the one-photon channel is a CPTP map with Kraus operators \(\{U_k\}\),
\[
\rho_{\mathrm{out}}=\sum_k U_k\,\rho_{\mathrm{in}}\,U_k^\dagger,
\]
then the joint two-photon map under identical action is
\[
\rho_{\mathrm{out}}=\sum_k (U_k\otimes U_k)\,\rho_{\mathrm{in}}\,(U_k^\dagger\otimes U_k^\dagger).
\]
Using the one-photon Mueller relation
\[
M_{i\ell}=\frac12\,\mathrm{Tr}\!\left[\sigma_i\Bigl(\sum_k U_k\,\sigma_\ell\,U_k^\dagger\Bigr)\right],
\]
or equivalently
\[
\sum_k U_k^\dagger \sigma_i U_k=\sum_\ell M_{i\ell}\sigma_\ell,
\]
the transformed tensor becomes
\[
T'_{ij}=\sum_{k,\ell}M_{ik}M_{j\ell}T_{k\ell}.
\]
In matrix notation,
\[
T'=MTM^T.
\]
This is the defining quadratic law of the formalism: a linear Mueller action on each photon induces a congruence transformation on the two-photon polarization correlations [2604.09257].

The derivation assumes that both photons experience the same set of Kraus operators \(\{U_k\}\), that these Kraus operators act independently on each photon copy, and that noise on different Kraus paths is uncorrelated aside from this identical action. Within those assumptions, the tensor furnishes an exact bridge between the Stokes-Mueller formalism and two-photon quantum channels.

A vectorized representation makes the quadratic structure explicit. If the \(4\times4\) tensor \(T\) is flattened into a 16-component column vector \(|T\rangle\), then
\[
|T'\rangle=(M\otimes M)\,|T\rangle.
\]
The \(16\times16\) Kronecker matrix \(M\otimes M\) has entries \(M_{ik}M_{j\ell}\), so the two-photon map is a quadratic mapping of the single-photon Mueller parameters. This Kronecker structure is the practical form used for numerical analysis and implementation [2604.09257].

## 3. Depolarization, purity, and entanglement decay

The same formalism gives a concrete quantitative distinction between one-photon probing and two-photon probing in a depolarizing medium. For the isotropic depolarizing Mueller matrix
\[
M=\mathrm{diag}(1,m,m,m),\qquad 0\le m\le 1,
\]
and the maximally entangled Bell state
\[
|\Psi^+\rangle=(|HV\rangle+|VH\rangle)/\sqrt2,
\]
the input correlation tensor is
\[
T_{\mathrm{in}}=\mathrm{diag}(1,-1,+1,+1).
\]
Under one-photon polarimetry, only one photon interacts, and the output purity is
\[
\gamma_{\mathrm{OPP}}=\mathrm{Tr}\,\rho_{\mathrm{out}}^2=\frac14(1+3m^2).
\]
Under two-photon polarimetry, both photons pass through the same depolarizing channel and
\[
T_{\mathrm{out}}=\mathrm{diag}(1,-m^2,m^2,m^2),
\qquad
\gamma_{\mathrm{TPP}}=\frac14(1+3m^4).
\]
Purity therefore scales quadratically in \(m\) for one-photon probing and quartically in \(m\) for two-photon probing [2604.09257].

The same contrast appears in entanglement degradation. For the output two-qubit state, the concurrence \(C\) decays as \(\propto m^2\) in OPP but \(\propto m^4\) in TPP. Mild depolarization thus produces a disproportionately large reduction of purity and entanglement in the two-photon probe. The abstract of the 2026 work states this as “quadratic degradation of entanglement and state purity,” while the detailed isotropic example exhibits quartic dependence on the depolarization factor in TPP because the tensor elements themselves acquire an \(m^2\) factor and quadratic state functionals then acquire \(m^4\) scaling [2604.09257].

This stronger scaling is also the source of enhanced sensitivity. If the depolarization factor depends on a sample parameter \(\eta\), \(m=m(\eta)\), then one may quantify sensitivity by \(d\gamma/d\eta\). The extra power of \(m\) in TPP leads to an enhanced slope and hence higher sensitivity to changes in \(\eta\). In the experimental realization with polarization-entangled photon pairs transmitted through controlled scattering media, the predicted response was confirmed and enhanced sensitivity to polarization scrambling relative to single-photon probing was observed [2604.09257].

## 4. Scattering tensors, \(g^{(2)}\), and symmetry-resolved coincidences

In cavity quantum materials, the relevant observable is the polarization-resolved second-order coherence function with explicit polarization indices,
\[
g^{(2)}_{\alpha\beta,\gamma\delta}(0)=
\frac{\langle b_\gamma(0)b_\delta(0)b^\dagger_\beta(0)b^\dagger_\alpha(0)\rangle}
{\langle b_\gamma(0)b^\dagger_\alpha(0)\rangle
 \langle b_\delta(0)b^\dagger_\beta(0)\rangle}.
\]
Using input-output relations, Kass et al. express the observable in terms of the two-photon scattering operator \(S\) as
\[
g^{(2)}_{\mu\mu'}(\omega_{\mathrm{in}};\nu\nu')
=
\frac{\sum_f|\langle f;\mu,\mu'|S|0;\mathrm{in},\nu;\mathrm{in},\nu'\rangle|^2}
{\sum_f|\langle f;\mu|S|0;\mathrm{in},\nu\rangle|^2\cdot
 \sum_f|\langle f;\mu'|S|0;\mathrm{in},\nu'\rangle|^2}.
\]
The associated tensor
\[
T_{\mu\mu';\nu\nu'}
\]
is therefore a scattering-probability tensor rather than a Pauli-moment tensor, but it plays the same organizational role for two-photon polarization information [2606.11550].

In a single-mode cavity there are only three orthogonal two-photon output states in a given polarization basis: \(HH\), \(VV\), and \((HV\equiv VH)\), or in the circular basis \(LL\), \(RR\), and \((LR\equiv RL)\). Accordingly, \(T\) has only three physical components per input-output pairing. The familiar autocorrelation is \(T_{\mu\mu;\nu\nu}\), whereas the rotated or cross-polarized coincidence is \(T_{\mu\mu;\nu\nu'}\) with \(\mu\perp\nu\), \(\mu'\perp\nu'\) [2606.11550].

At weak light-matter coupling \(D\), the unrotated coincidence channel probes a two-point material correlator:
\[
g_u^{(2)}(\omega_{\mathrm{in}};1,1)
=
1+\int d\omega\;
\frac{4(-\omega/2)\chi^{11}(\omega)}
{((\omega_{\mathrm{in}}-\omega_c)^2+\gamma^2)\,((\omega_{\mathrm{in}}-\omega_c-\omega)^2+\gamma^2)},
\]
with
\[
\chi^{11}(\omega)=\int dt\,e^{i\omega t}\langle0|D^{11}(t)D^{11}(0)|0\rangle,
\]
identified as the Raman dynamical structure factor in the “11” polarization channel. In contrast, the rotated channel \(\mu\perp\nu\) has no elastic background and yields a fourth-order expansion probing the four-point kernel
\[
K_{11}^{22}(\omega_1,\omega_2,\omega_3).
\]
Selecting orthogonal polarizations thus isolates higher-order light-matter scattering processes that probe higher-order material correlations [2606.11550].

The tensor also carries symmetry information. In the Kitaev-Heisenberg stripy phase (\(C_2\)), the unrotated \(g_u^{(2)}\) exhibits a two-fold pattern in the linear-polarization angle \(\alpha\); in the antiferromagnetic (\(C_3\)) phase it exhibits six-fold symmetry for linear polarization or continuous rotational symmetry for circular polarization. The rotated \(g_r^{(2)}\), being purely inelastic, can be tuned into strongly bunched or antibunched regimes not visible in the unrotated channel [2606.11550].

## 5. Nonlinear generation and time-resolved reconstruction

In broadband spontaneous four-wave mixing in diamond, the two-photon polarization tensor is built from the \(\chi^{(3)}\) tensor. In the low-gain regime, the biphoton amplitude is
\[
\Psi_{ij}(\omega_s,\omega_i)\propto
\sum_{mn}\chi^{(3)}_{imnj}(\omega_s,\omega_p,\omega_p,-\omega_i)\,E_{p,m}E_{p,n}\,\Phi(\Delta k),
\]
with \(\Phi(\Delta k)=\mathrm{sinc}[(\Delta k\cdot L)/2]e^{i(\Delta k\cdot L)/2}\). The measurable fourth-order correlation tensor factorizes as
\[
G_{ijkl}(\omega_s,\omega_i)\propto \Psi_{ij}(\omega_s,\omega_i)\Psi^*_{kl}(\omega_s,\omega_i).
\]
In diamond,
\[
\chi^{(3)}_{ijkl}=\chi^{(3)E}_{ijkl}+\chi^{(3)R}_{ijkl}(\omega_s-\omega_p),
\]
with a nonresonant electronic part and a resonant Raman part. For the vertical-pump geometry, the nonzero tensor elements include \(\chi^{(3)}_{VVVV}\) and \(\chi^{(3)}_{HVVH}\), and the biphoton state reduces to
\[
|\Psi(\Delta)\rangle
=
\chi_{VVVV}^{(3)}(\Delta)|VV\rangle+\chi_{HVVH}^{(3)}(\Delta)|HH\rangle.
\]
Observable polarization correlations are then derived from \(G_{ijkl}\), used to construct \(P(\theta_s,\theta_i;\Delta)\), the correlation coefficient \(E(\theta_1,\theta_2;\Delta)\), the CHSH combination \(S_{\Psi^\pm}(\Delta)\), and the concurrence
\[
C(\Delta)=\frac{2|A_{VV}(\Delta)A_{HH}(\Delta)|}{|A_{VV}(\Delta)|^2+|A_{HH}(\Delta)|^2}.
\]
Over a broad band \((|\Delta|\lesssim 1250\,\mathrm{cm}^{-1})\), the measured \(S_{\Psi^+}(\Delta)\) violates the CHSH bound \(S\le2\) by up to \(8.7\,\sigma\) [2408.11477].

A distinct reconstruction protocol appears in continuous-wave quantum-dot cascade tomography. Tur et al. define the delay-dependent tensor
\[
T_{ij}(\tau)\equiv \langle \sigma_i\otimes\sigma_j\rangle_\tau
=\mathrm{Tr}\{\rho^{(2)}(\tau)(\sigma_i\otimes\sigma_j)\},\qquad i,j=1,2,3,
\]
and recover it from 36 measured time-resolved polarization correlations \(g^{(2)}_{P_1,P_2}(\tau)\). Each analyzer setting corresponds to a projector
\[
\Pi_P=\frac12\bigl[I+n_x\sigma_x+n_y\sigma_y+n_z\sigma_z\bigr],
\]
and the tensor components are obtained from
\[
T_{ij}(\tau)=\sum_{m,n=\pm1}mn\,P_{mn}^{ij}(\tau)
=
g_{++}^{ij}(\tau)+g_{--}^{ij}(\tau)-g_{+-}^{ij}(\tau)-g_{-+}^{ij}(\tau).
\]
Because \(P_1\) and \(P_2\) run over the six canonical bases \(H/V\), \(D/\bar D\), and \(R/L\), the 36 measured \(g^{(2)}\) functions suffice to invert a fixed \(9\times36\) linear map and extract the full \(3\times3\) real matrix \(T(\tau)\) [2503.03268].

The time dependence is modeled with a Lindblad master equation for the four-level biexciton-exciton cascade. The tensor components inherit the Lindbladian eigenstructure and take the form
\[
T_{ij}(\tau)=\sum_k \alpha_k^{(ij)}e^{-\lambda_k|\tau|}
+\beta^{(ij)}e^{-(G+1/\tau_X)|\tau|}\cos(\Delta\tau/\hbar).
\]
The reported fit parameters include \(\Delta=29\,\mu\mathrm{eV}\), oscillation period \(T_0=\hbar/\Delta=140\pm10\,\mathrm{ps}\), \(\tau_H=1180\pm10\,\mathrm{ps}\), \(\tau_V=990\pm10\,\mathrm{ps}\), and \(G^{-1}=8.0\pm0.5\,\mathrm{ns}\). At \(\tau=0\), the diagonal components are approximately \(T_{zz}(0)=+0.62\pm0.02\), \(T_{xx}(0)=+0.31\pm0.03\), and \(T_{yy}(0)=+0.28\pm0.03\), and the nine functions agree with theory at the few-percent level after convolution with the \(42\,\mathrm{ps}\) detector response [2503.03268].

## 6. Angular-momentum-resolved and atomic realizations

For Landau’s two-photon spherical states in the center-of-mass frame, the tensor is indexed by total angular momentum \(J\), magnetic quantum number \(M\), and parity \(P\):
\[
T_{ij}(J,M,P)=\mathrm{Tr}\!\bigl[\rho(J,M,P)\,\sigma_i\otimes\sigma_j\bigr].
\]
Alexanian and Mkrtchian find two families. For \(|A|=0\) states, the density matrices are
\[
\rho^{(P=\pm)}_{J,M}
=\frac14\Bigl[I\otimes I+\sigma_3\otimes\sigma_3
\pm(\sigma_1\otimes\sigma_1-\sigma_2\otimes\sigma_2)\Bigr],
\]
so that
\[
T_{33}=+1,\qquad T_{11}=\pm1,\qquad T_{22}=\mp1,\qquad T_{i\neq j}=0.
\]
For \(|A|=2\) states, one obtains
\[
T_{11}=T_{22}=-\chi_{JM}(\theta),\qquad T_{33}=0,
\]
with single-photon terms \(T_{30}=-T_{03}=\zeta_{JM}(\theta)\) and the remaining mixed components \(T_{31}=T_{32}=T_{13}=T_{23}=0\). The corresponding coincidence laws are
\[
W_\pm(\phi,\phi')=\frac14\bigl[1\pm\cos2(\phi-\phi')\bigr]
\]
for the \(|A|=0\) states and
\[
W_{e/o}(\phi,\phi';\theta)=\frac14\bigl[1+\chi_{JM}(\theta)\cos2(\phi+\phi')\bigr]
\]
for the even-\(J\) Landau states. The appearance of \(\cos2(\phi+\phi')\) terms is the distinctive new feature emphasized in that analysis [2109.14984].

In two-photon decay of hydrogen-like ions, Fratini and Surzhykov formulate the polarization tensor as the joint two-photon spin density matrix projected onto a chosen polarization basis:
\[
C^{(2)}_{\alpha\beta;\gamma\delta}(\omega_1,\omega_2,\theta)
=
\sum_{\lambda_1\lambda_2\lambda'_1\lambda'_2}
[\epsilon_\alpha(k_1)]_{\lambda_1}[\epsilon_\beta(k_2)]_{\lambda_2}
\rho_{\lambda_1\lambda_2;\lambda'_1\lambda'_2}(k_1,k_2)
[\epsilon_\gamma(k_1)]^*_{\lambda'_1}[\epsilon_\delta(k_2)]^*_{\lambda'_2}.
\]
The underlying density matrix is built from second-order Dirac transition amplitudes and a multipole expansion involving reduced radial matrix elements, Wigner \(D\)-matrices, and \(3j\), \(6j\), and \(9j\) symbols. In the \(2E1\) approximation for the \(2s_{1/2}\to1s_{1/2}\) transition, the normalized coincidence law is
\[
\Phi^{2E1}_{\chi_1,\chi_2}(\theta)
=
\frac{1}{1+\cos^2\theta}
\bigl[\sin\chi_1\sin\chi_2+\cos\chi_1\cos\chi_2\cos\theta\bigr]^2,
\]
equivalently corresponding to the pure state
\[
|\Psi^{2E1}_{2\gamma}\rangle
=
\frac{1}{\sqrt{1+\cos^2\theta}}
\bigl(|yy\rangle+\cos\theta\,|xx\rangle\bigr).
\]
For heavy ions, inclusion of \(M1\), \(E2\), \(M2\), and related higher multipoles opens small “forbidden” orthogonal-plane coincidences. At energy sharing \(x=1/16\) and \(\theta=90^\circ\), the reported values are \(\Phi(0^\circ,90^\circ)=0\) and \(\Phi(90^\circ,90^\circ)=1\) for H in full \(2E1\), \(\Phi(0^\circ,90^\circ)\approx0.02\) and \(\Phi(90^\circ,90^\circ)\approx0.98\) for \(\mathrm{Xe}^{53+}\), and \(\Phi(0^\circ,90^\circ)\approx0.15\) and \(\Phi(90^\circ,90^\circ)\approx0.85\) for \(\mathrm{U}^{91+}\) [1006.4799].

## 7. Scope, operational meaning, and recurring points of interpretation

The operational content of the tensor is especially transparent in coincidence-based Bell and nonlocality experiments. In the coherence-optics Franson-type experiment analyzed by Ham, the bipartite polarization-qubit tensor is defined as
\[
T_{ij}=\langle \sigma_i^A\otimes \sigma_j^B\rangle,\qquad i,j\in\{x,y,z\},
\]
with experimental reconstruction through
\[
T_{ij}=
\frac{N_{++}^{ij}+N_{--}^{ij}-N_{+-}^{ij}-N_{-+}^{ij}}{N_{\rm tot}^{ij}}.
\]
For the two-photon state
\[
|\Psi\rangle=\frac1{\sqrt2}\Bigl(e^{i\phi_+}|H\rangle_A|V\rangle_B+e^{i\phi_-}|V\rangle_A|H\rangle_B\Bigr),
\qquad \Delta\phi=\phi_+-\phi_-,
\]
the nonzero components are
\[
T_{xx}=\cos\Delta\phi,\quad
T_{yy}=-\cos\Delta\phi,\quad
T_{xy}=\sin\Delta\phi,\quad
T_{yx}=-\sin\Delta\phi,\quad
T_{zz}=-1,
\]
with all other components vanishing. In the quantum-eraser setting, the normalized correlation function becomes
\[
E(\phi_A,\phi_B)=\sin 2(\phi_A+\phi_B),
\]
and the CHSH parameter reaches \(S=2\sqrt2\) [2206.04251].

A common misconception is that the two-photon polarization correlation tensor is a single fixed object independent of experimental context. The cited literature shows instead that the same phrase covers several tightly connected constructions: Pauli-moment tensors for density operators, rank-\((2\otimes2)\times(2\otimes2)\) scattering tensors for polarization-resolved \(g^{(2)}\), and fourth-order field-correlation tensors in nonlinear optics. Another common misconception is that the tensor is merely a redundant repackaging of the density matrix. The cited works use it to expose channel congruence \(T'=MTM^T\), higher-order sensitivity to depolarization, symmetry fingerprints in cavity quantum materials, explicit analyzer-angle laws involving either \(\phi-\phi'\) or \(\phi+\phi'\), and time-resolved decoherence under Lindbladian evolution [2604.09257].

A plausible implication is that the most useful version of the tensor depends on what is being characterized. For polarization channels, the Pauli-tensor form is the natural vehicle for the quadratic Mueller action. For photon scattering, the four-index coincidence tensor is the natural object because it resolves both input and output polarizations. For source engineering and spectroscopy, the fourth-order field-correlation tensor directly links material response functions or nonlinear susceptibilities to measurable polarization coincidences. Across these formulations, the invariant theme is the same: the tensor is the minimal structured object that retains the joint polarization content of a two-photon process while remaining directly connected to experimentally accessible coincidences.

Source: https://www.emergentmind.com/topics/two-photon-polarization-correlation-tensor