Unified Depolarizing-Dephasing Noise Model

• The unified depolarizing-dephasing channel model is a comprehensive framework that describes both isotropic depolarization and axis-specific dephasing in single-qubit systems.
• It employs multiple representations, including the process-matrix, centered ellipsoid, and tetrahedron parametrizations, to link theoretical constructs with practical optical implementations.
• The model integrates non-Markovian extensions and microscopic dynamics, facilitating improvements in quantum key distribution, quantum process tomography, and higher-dimensional channel analysis.

“Unified depolarizing-dephasing channel model” denotes a common description of two archetypal unital noise processes on qubits or polarization qubits: depolarization, which drives states toward the maximally mixed state, and dephasing, which suppresses selected coherences while preserving populations in a preferred basis. In the cited literature, this unification appears in several closely related forms: a Pauli-channel process-matrix representation, a centered-ellipsoid map of the Poincaré or Bloch sphere, optical channels obtained by tracing out temporal or spatial degrees of freedom, and composite effective channels with separate depolarization and decoherence parameters for free-space links (Shaham et al., 2019, Shaham et al., 2010, Peng et al., 2 Sep 2025).

1. Channel-theoretic definition

At the single-qubit level, a unified treatment begins with the general process-matrix representation

$$\mathcal{E}(\hat{\rho})=\sum_{m,n}\chi_{mn}\sigma_m\hat{\rho}\sigma_n^\dagger,$$

where $\sigma_0$ is the identity and $\sigma_{1,2,3}$ are the Pauli matrices. The class most directly relevant here is the class of unital channels, defined by $\mathcal{E}(\hat{I})=\hat{I}$. In Shaham et al., all unital single-qubit channels map the Poincaré sphere into a centered ellipsoid; dephasing and isotropic depolarization are specific points or subfamilies inside that larger geometry (Shaham et al., 2019).

Within this framework, the isotropic depolarizing channel is written as

$$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$

while the dephasing channel used in the same work is

$$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$

The first contracts the Bloch sphere uniformly; the second suppresses phase coherence along a selected axis. In an equivalent Bloch-vector description, the depolarizing channel acts as $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$, making its isotropy explicit (Shaham et al., 2019, Tiago et al., 19 Dec 2025).

In optical polarization language, the same distinction is expressed on the Poincaré sphere through the Stokes vector $\overline S=(S_1,S_2,S_3)$ and the degree of polarization

$$D=\sqrt{S_1^2+S_2^2+S_3^2}=\sqrt{1-4\det(\hat{\rho})}.$$

A single birefringent crystal aligned along a fixed axis acts as a dephasing channel, whereas appropriate sequences of crystals and wave plates realize more general depolarizing maps, including anisotropic and isotropic cases (Shaham et al., 2010).

2. Unified parametrizations and formal variants

A common parametrization of unital single-qubit channels uses the tetrahedron representation, in which the channel is specified by $\vec D=\{D_1,D_2,D_3\}$, the eigenvalues of a real matrix associated with the channel’s Choi matrix. This representation is central in the four-crystal optical realization, where almost every physically allowed point in the tetrahedron can be reached by adjusting half-wave plate angles $\sigma_0$0 (Shaham et al., 2019).

The literature also uses other unified forms, depending on the operational setting. Representative constructions are summarized below.

Representation	Channel form	Salient parameters
Unital single-qubit map	centered ellipsoid in the unital tetrahedron	$\sigma_0$1
Noisy unitary mixture	$\sigma_0$2	$\sigma_0$3, choice of $\sigma_0$4
Composite turbulence channel	depolarization plus off-diagonal shrinkage plus loss	$\sigma_0$5

These parametrizations are not interchangeable in purpose, but they are structurally aligned. The tetrahedral form is geometric and channel-complete for unital qubit maps; the noisy-unitary mixture is tailored to learning, storage, and retrieval tasks; the composite turbulence channel is designed to plug directly into MDI-QKD security formulas (Shaham et al., 2019, Pavličko et al., 2022, Peng et al., 2 Sep 2025).

A further unified formulation appears in the probabilistic storage-and-retrieval setting, where noisy phase gates are modeled as

$\sigma_0$6

with $\sigma_0$7 taken either as the depolarizing channel $\sigma_0$8 or the dephasing channel $\sigma_0$9. This places depolarizing and dephasing noise inside the same convex-combination architecture, while preserving distinct operational consequences (Pavličko et al., 2022).

3. Optical realizations and channel engineering

The most explicit laboratory realization of a nearly universal unital channel for polarization qubits is the four-crystal scheme of Shaham et al. The setup comprises four birefringent calcite crystals interleaved with three half-wave plates. The outer crystals $\sigma_{1,2,3}$0 have thickness $\sigma_{1,2,3}$1 mm, the inner crystals $\sigma_{1,2,3}$2 have thickness $\sigma_{1,2,3}$3 mm, and the fast axes of $\sigma_{1,2,3}$4 and $\sigma_{1,2,3}$5, together with the slow axes of $\sigma_{1,2,3}$6 and $\sigma_{1,2,3}$7, are parallel. Each crystal delays $\sigma_{1,2,3}$8 and $\sigma_{1,2,3}$9 by different amounts; when the delay $\mathcal{E}(\hat{I})=\hat{I}$0, polarization becomes entangled with temporally distinguishable modes, and slow detectors trace out the temporal degree of freedom, leaving a mixed polarization state. The half-wave plate angles control the temporal-mode populations and hence the implemented unital map (Shaham et al., 2019).

For dephasing, the prescription

$\mathcal{E}(\hat{I})=\hat{I}$1

yields a one-parameter family with

$\mathcal{E}(\hat{I})=\hat{I}$2

Varying $\mathcal{E}(\hat{I})=\hat{I}$3 from $\mathcal{E}(\hat{I})=\hat{I}$4 to $\mathcal{E}(\hat{I})=\hat{I}$5 tunes the channel continuously from no dephasing to complete dephasing. The geometry of the accessible region implies that pure dephasing channels on a tetrahedron edge are approximated rather than reached exactly, but the approximation can be within $\mathcal{E}(\hat{I})=\hat{I}$6 process fidelity. Full quantum process tomography, based on the input states $\mathcal{E}(\hat{I})=\hat{I}$7, gave an average process fidelity of $\mathcal{E}(\hat{I})=\hat{I}$8 with theory. The same work reports that the scheme is wavelength insensitive and independent of the fine temporal properties of the input light, in contrast to a Soleil-Babinet compensator dephasing channel, which depends on photon temporal properties (Shaham et al., 2019).

Earlier birefringent-crystal schemes already exhibited the same unifying principle in less general form. In particular, a Lyot depolarizer made from two crystals of lengths $\mathcal{E}(\hat{I})=\hat{I}$9 and $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$0 rotated by $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$1 achieves complete isotropic depolarization, while other two-crystal and quarter-wave-plate arrangements allow continuous tuning from anisotropic dephasing-like behavior to isotropic depolarization. For one such scheme, the channel is completely isotropic at $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$2, only one basis is completely depolarized at $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$3, and two states are equivalently depolarized at $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$4. Process fidelities were reported as above $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$5 for the tested configurations (Shaham et al., 2010).

A distinct optical route uses spin-orbit modes and vector beams. In that setting, any single-qubit channel can be decomposed into a convex combination of two quasi-extreme channels using one ancillary qubit, two CNOTs, and four single-qubit gates, following the Solovay-Kitaev decomposition for quantum channels. The system is encoded in polarization and the ancilla in Hermite-Gaussian transverse modes. The same framework can represent both dephasing and depolarizing channels by parameter adjustment, while a compact linear optical circuit based on a maximally nonseparable spin-orbit mode realizes programmable depolarization directly through partial tracing over the spatial degree of freedom. Reported fidelities exceed $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$6 for the compact circuit, whereas the more complex SK implementation showed $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$7 fidelity for fully mixed states but about $$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$8 fidelity for a pure initial vertical state (Tiago et al., 19 Dec 2025).

4. Non-Markovian and microscopic extensions

Unified depolarizing-dephasing descriptions are not restricted to memoryless channels. An explicit non-Markovian construction modifies the Kraus weights of qubit Pauli channels and identifies non-Markovianity with the breakdown of CP-divisibility, that is, the appearance of a not-completely-positive intermediate map. For dephasing, the Kraus operators are

$$\mathcal{E}(\hat{\rho})=(1-p)\hat{\rho}+\frac{p}{3}\sum_{i=1}^3 \sigma_i\hat{\rho}\sigma_i,$$9

with $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$0 and $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$1. For depolarizing noise, the analogous construction is

$$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$2

In the dephasing case, the Choi eigenvalues of the intermediate map cross at a singular point $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$3, corresponding to momentary non-invertibility and a singular canonical decoherence rate; beyond that point the rate becomes negative. Non-Markovianity is quantified both by a normalized HCLA/RHP construction and by the BLP distinguishability measure, with $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$4 in this model (Shrikant et al., 2018).

Microscopic models provide a different unification axis. For depolarizing noise, one class of models couples a spin-one-half system to harmonic oscillators or collective spin systems, producing an isotropic contraction of the Bloch vector with a time-dependent depolarizing parameter $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$5. A classical counterpart uses random magnetic fields with Gaussian statistics. For phase damping, the interaction is via $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$6 only, yielding exponential or non-exponential decay of off-diagonal density-matrix elements through a decoherence function $$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$7. Markovian and non-Markovian versions of both channels are thereby derived from explicit Hamiltonians or stochastic fields (Romero et al., 2012).

A central limitation is also explicit in this microscopic literature: no single microscopic Hamiltonian is constructed that unifies depolarizing and phase-damping channels into one process. The combination is straightforward at the level of maps or time-local master equations, but not supplied as a derived microscopic model. In the notation summarized in that work, a combined map can be written as

$$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$8

or, at the generator level, by a master equation carrying both depolarizing and dephasing terms (Romero et al., 2012).

5. Task-specific unified models

A unified model becomes operationally significant when the two noise types are processed differently by a protocol. In optimal probabilistic storage and retrieval of phase gates, noisy inputs are modeled as convex combinations of the unitary target channel and either depolarizing or dephasing noise. For dephasing,

$$\mathcal{E}(\hat{\rho})=(1-P)\hat{\rho}+P\sigma_3\hat{\rho}\sigma_3.$$9

and the retrieved channel preserves exactly the same form,

$$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$0

with success probability

$$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$1

For depolarizing noise,

$$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$2

the retrieved channel becomes

$$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$3

with

$$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$4

Thus the retrieval stage leaves dephasing unchanged but converts depolarizing noise into a less noisy dephasing admixture, with $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$5 for $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$6 (Pavličko et al., 2022).

A more explicitly named unified depolarizing-dephasing model appears in the free-space MDI-QKD turbulence literature. There the physical channel composes phase perturbations, Gaussian beam spreading, beam drift, aperture truncation, and scintillation into three effective parameters: depolarization factor $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$7, decoherence factor $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$8, and detection probability $$\vec r\mapsto \left(1-\frac{4\lambda}{3}\right)\vec r$$9. For post-selected detection events, the polarization state is modeled as

$\overline S=(S_1,S_2,S_3)$0

This separates isotropic mixing from off-diagonal attenuation while preserving a direct link to physically measurable channel coefficients (Peng et al., 2 Sep 2025).

The same framework maps turbulence-induced polarization rotation to a von Mises-Fisher or Watson distributed SU(2) rotation, with phase fluctuations described by the structure function

$\overline S=(S_1,S_2,S_3)$1

or equivalently

$\overline S=(S_1,S_2,S_3)$2

Local depolarization is

$\overline S=(S_1,S_2,S_3)$3

and local off-diagonal shrinkage is

$\overline S=(S_1,S_2,S_3)$4

The resulting analytic SKR formulas are compatible with existing MDI-QKD security analyses, and the paper reports numerical agreement across clear, overcast, and hazy weather conditions for ground-to-satellite links (Peng et al., 2 Sep 2025).

6. Higher-dimensional generalizations and interpretive caveats

The unification of depolarizing and dephasing behavior becomes subtler beyond qubits. One result from higher-dimensional channel theory is that gate fidelity loses discriminative power: for dimensions $\overline S=(S_1,S_2,S_3)$5, there exist non-depolarizing quantum channels whose gate fidelity with respect to the identity is constant across all pure inputs, exactly as for a depolarizing channel. Moreover, as $\overline S=(S_1,S_2,S_3)$6, the gate fidelity of any channel concentrates around its mean and converges, in probability and in $\overline S=(S_1,S_2,S_3)$7, to that of the depolarizing channel with the same average fidelity. The variance bound

$\overline S=(S_1,S_2,S_3)$8

makes the concentration explicit. A common misconception is therefore corrected: a flat gate-fidelity profile does not uniquely identify depolarization (Magesan, 2010).

A different higher-dimensional extension is the SU(2)-invariant random-unitary model of depolarization for quantum states of light. There the channel is generated by random SU(2) rotations,

$\overline S=(S_1,S_2,S_3)$9

and, under Markovian and time-continuous assumptions, yields the SU(2)-invariant master equation

$$D=\sqrt{S_1^2+S_2^2+S_3^2}=\sqrt{1-4\det(\hat{\rho})}.$$0

In this model, depolarization is described as simultaneous “dephasing” along all three orthogonal axes, ensuring isotropy. For a qubit the model reduces to the standard depolarizing channel, but for higher photon numbers the evolution generally cannot be written in that standard form. The asymptotic steady state is the identity in each photon-number subspace, and higher-order polarization measures show that nonclassical states such as NOON and twin-photon-number states depolarize faster than SU(2) coherent states (Rivas et al., 2013).

Taken together, these results indicate that the phrase “unified depolarizing-dephasing channel model” does not designate a single canonical formula across the literature. Instead, it denotes a class of representations in which depolarizing and dephasing effects are embedded in a common formal structure, with the preferred representation determined by the problem: universal unital control, microscopic dynamics, non-Markovian diagnostics, noisy-unitary processing, or turbulence-aware QKD analysis.