Sub-Ohmic Spin-Boson Model

• The sub-Ohmic spin-boson model is a quantum two-level system coupled to a bosonic environment whose low-frequency spectral density scales as ω^s (0<s<1), leading to pronounced memory effects.
• It exhibits rich phenomena including delocalized-localized quantum phase transitions, non-Markovian relaxation, and hidden pseudo-coherent dynamics under various conditions.
• The model serves as a sensitive benchmark for diverse numerical and analytical methods, with practical applications in quantum simulation and understanding decoherence in solid-state devices.

The sub-Ohmic spin-boson model describes a quantum two-level system coupled linearly to a bosonic environment whose low-frequency spectral density scales as $J(\omega)\propto \omega^s$ with $0$s=0$ and $s=-1$ (Shen et al., 2023, Abdi et al., 2018). The model has therefore become a standard setting for dissipative quantum criticality, non-Markovian relaxation, and the analysis of decoherence by low-frequency environments (Kirchner et al., 2011, Noh et al., 2014).

1. Canonical formulation and spectral classification

A standard spin-boson Hamiltonian is

$$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$

where $\varepsilon$ is the bias, $\Delta$ is the tunneling amplitude, and the bath enters through the mode frequencies $\omega_k$ and couplings $\lambda_k$ (Shen et al., 2023). The bath is encoded by a spectral density $J(\omega)$; one common convention is $0Shen et al., 2023, Yang et al., 2020).

The exponent $0Shen et al., 2023). Smaller $0Kirchner et al., 2011). Recent work often separates the sub-Ohmic interval into a deep regime $0Shen et al., 2023).

The model is also studied in effective and extended forms. A mechanically engineered analog simulator based on a color center in a free-standing h-BN membrane realizes $s=0$0 with $s=0$1, with the cases $s=0$2 and $s=0$3 corresponding to $s=0$4-type and white-noise environments (Abdi et al., 2018). This suggests that the conventional sub-Ohmic family is best viewed as one part of a broader infrared-dominated dissipative landscape.

2. Zero-temperature phases and quantum criticality

At zero bias and zero temperature, the canonical sub-Ohmic model has a transition between a delocalized phase with vanishing order parameter and a localized phase with finite magnetization $s=0$5 (Kirchner et al., 2011, Chin et al., 2011). In a variational Asymmetrically Displaced Oscillator treatment, the transition is continuous for $s=0$6, with

$s=0$7

and

$s=0$8

in the scaling limit (Chin et al., 2011). That same work emphasized that the localized phase retains finite coherence $s=0$9, and interpreted localization as the emergence of a self-consistent low-frequency bath bias rather than a trivial freezing of all tunneling processes (Chin et al., 2011).

The universality class has been controversial. Bosonic NRG work on critical susceptibilities reported an interacting quantum critical point with $s=-1$0 and $s=-1$1 scaling across $s=-1$2, and attributed the failure of the quantum-to-classical mapping to a Berry-phase term in the continuum spin path integral (Kirchner et al., 2011). A displaced-Fock-states expansion instead reported $s=-1$3 for the whole interval $s=-1$4 and interpreted this as evidence that the system is always above its upper critical dimension (He et al., 2014). A later multiple-polaron numerical variational study found mean-field criticality for $s=-1$5, non-mean-field $s=-1$6-dependent criticality for $s=-1$7, and stated that the quantum-to-classical correspondence is fully confirmed over the entire sub-Ohmic range (Shen et al., 2023).

These competing claims define a central interpretive fault line in the literature. A plausible implication is that the sub-Ohmic model is not only a problem in dissipative dynamics, but also a sensitive benchmark for how infrared bath physics, bosonic truncation, and continuum limits are handled numerically.

3. Nonequilibrium dynamics, transient phase diagrams, and hidden dynamical structure

Real-time nonequilibrium studies have shown that the dynamical organization of the sub-Ohmic model is richer than the equilibrium phase diagram alone. A numerically exact inchworm-QMC study extracted transient phase boundaries from the post-quench polarization

$s=-1$8

using $s=-1$9 as a localization diagnostic and $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$0, $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$1 as coherence diagnostics (Goulko et al., 2024). In that formulation, localization is associated with $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$2, while coherence loss can occur either through a smooth overdamping crossover $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$3 or through a sharp frequency collapse $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$4 (Goulko et al., 2024).

This dynamical viewpoint does not coincide exactly with the equilibrium one. The same work reported that the transient localization threshold $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$5 agrees well with equilibrium expectations only for sufficiently small $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$6, while at larger $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$7 the transient and equilibrium critical couplings and apparent critical exponents diverge (Goulko et al., 2024). A later finite-temperature extension found that increasing temperature weakens localization and coherence in different ways: the incoherent region expands toward weaker coupling, whereas the transient localization threshold shifts only weakly and, where resolvable, tends to move to higher coupling at small $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$8 (Goulko et al., 2 Sep 2025). That study also emphasized that coherence and localization are distinct dynamical properties, since a state can be delocalized yet incoherent (Goulko et al., 2 Sep 2025).

A further refinement is the “hidden” or pseudo-coherent phase. Using QUAPI/TEMPO, strong-coupling sub-Ohmic dynamics was shown to contain three regimes—coherent, incoherent, and pseudo-coherent—rather than a simple coherent/incoherent dichotomy (Otterpohl et al., 2022). In the pseudo-coherent regime the polarization shows a single minimum without a subsequent maximum, and the timescale of that feature scales as $$\hat{H}=\frac{\varepsilon}{2}\sigma_z-\frac{\Delta}{2}\sigma_x+\sum_k \omega_k b_k^\dagger b_k+\frac{\sigma_z}{2}\sum_k \lambda_k(b_k^\dagger+b_k),$$9, indicating that the spin is slaved to oscillatory bath dynamics rather than executing intrinsic coherent oscillations (Otterpohl et al., 2022).

Initial preparation also matters sharply. A time-dependent Davydov-$\varepsilon$0 study at $\varepsilon$1 found that under a polarized bath initial condition coherent oscillations persisted up to $\varepsilon$2, whereas under a factorized initial condition a coherent-incoherent transition occurred at $\varepsilon$3 (Wu et al., 2013). This establishes that “strong-coupling coherence” in the deep sub-Ohmic regime is not a preparation-independent statement.

Recent entanglement-based work further complicates the picture. A TTN-TDVP-PS study found that the population-based coherent, incoherent, and pseudo-coherent regimes do not map one-to-one onto distinct stationary entanglement phases; instead, the stationary spin entropy defines a single entropy ridge in the $\varepsilon$4 plane, which follows the population-based boundary only at small $\varepsilon$5 and remains single-valued inside the incoherent region at larger $\varepsilon$6 (Gong et al., 18 Jun 2026).

4. Decoherence, finite temperature, and equilibrium dynamical response

Sub-Ohmic environments are exceptionally effective at dephasing because the infrared sector couples strongly to the nonoscillatory parts of the reduced dynamics. A consistent perturbative treatment showed that the standard Markov approximation fails for the kernel $\varepsilon$7 in the sub-Ohmic regime, replacing simple exponential dephasing by a nonexponential factor $\varepsilon$8 and removing the spurious instantaneous-dephasing pathology of earlier weak-coupling formulas (Noh et al., 2014). In the $\varepsilon$9, $\Delta$0 limit this treatment gives

$\Delta$1

so arbitrarily small $\Delta$2 ratios arise as the bath coupling becomes predominantly longitudinal (Noh et al., 2014). For $\Delta$3, the off-diagonal coherence decays as $\Delta$4, i.e. with nonanalytic stretched-exponential-type behavior rather than a simple exponential (Noh et al., 2014).

Equilibrium finite-temperature dynamics adds another layer. Full-density-matrix NRG calculations of the symmetrized correlation function $\Delta$5 found a thermal peak at

$\Delta$6

with $\Delta$7 merging with the $\Delta$8 result for $\Delta$9 and deviating strongly from the zero-temperature $\omega_k$0 laws for $\omega_k$1 (Yang et al., 2020). The same work interpreted finite temperature as a strong infrared regulator that can mask the zero-temperature crossover scale near criticality (Yang et al., 2020).

Bias does not remove all universal low-frequency structure. For the biased sub-Ohmic model, bosonic NRG found that

$\omega_k$2

throughout the biased parameter space, including the biased strong-coupling regime, except at the special point $\omega_k$3, where

$\omega_k$4

(Zheng et al., 2017). The same study also supported the generalized Shiba relation

$\omega_k$5

over a wide range of parameters (Zheng et al., 2017).

Equilibrium dynamics near criticality has also been analyzed through the full set of spin components. A combined bosonic-NRG and Majorana-diagrammatics study concluded that the bosonic self-energy is essential for the description of critical fluctuations, but that many-body vertex corrections must also be included to obtain quantitative agreement with numerical simulations (Florens et al., 2011). This work suggested that long-time out-of-equilibrium dynamics beyond the Bloch-Redfield regime deserves reconsideration in dissipative critical systems (Florens et al., 2011).

5. Numerical and analytical methodologies

Because the sub-Ohmic model combines strong infrared singularity, large bosonic displacements, and long memory times, no single method is uniformly reliable across all observables and parameter regimes. The main methodological families are summarized below.

Method family	Representative papers	Main use
Bosonic NRG / FDM-NRG	(Kirchner et al., 2011, Yang et al., 2020, Zheng et al., 2017)	Critical susceptibilities, equilibrium spectra, biased and unbiased low-frequency laws
Inchworm QMC	(Goulko et al., 2024, Goulko et al., 2 Sep 2025)	Transient dynamical phase diagrams and finite-temperature real-time polarization
Variational coherent-state methods	(Chin et al., 2011, Shen et al., 2023)	Ground-state criticality, critical couplings, mean-field versus non-mean-field behavior
MPS/DMRG-type approaches	(Frenzel et al., 2012, Zhao et al., 2014)	Localized phase with large occupations, chain representations, multi-bath generalizations
Path-integral and tensor-network real-time methods	(Otterpohl et al., 2022, Gong et al., 18 Jun 2026)	Hidden dynamical phases and entanglement structure

Several method-specific developments are particularly tied to the sub-Ohmic problem. An MPS representation without explicit local Hilbert-space truncation was introduced specifically to cope with the very large boson occupations in and beyond the localized phase, and it reproduced the mean-field exponent $\omega_k$6 while allowing extrapolation of infinite-chain critical couplings (Frenzel et al., 2012). A single-mode approximation constructed from a rotating-wave transformation and NRG-inspired transformations yielded the classical exponents

$\omega_k$7

for $\omega_k$8, and was used to argue that the original bosonic NRG mishandles the crossover temperature $\omega_k$9 in that regime (Liu et al., 2012).

On the dynamical side, a variational surface-hopping algorithm built on the Davydov $\lambda_k$0 ansatz was designed to treat coherent and incoherent population evolution within a single framework, and its hopping rates follow Marcus-theory-like $\lambda_k$1 scaling more closely than conventional semiclassical surface hopping (Yao et al., 2013). This suggests that the sub-Ohmic model has also become a testbed for hybrid quantum-classical algorithm design, not only for critical phenomenology.

6. Generalizations, engineered realizations, and broader significance

Several extensions move beyond the canonical one-bath, one-coupling-direction problem. A two-bath model with one bath coupled diagonally to $\lambda_k$2 and another off-diagonally to $\lambda_k$3 was shown, by DMRG with optimized boson basis, to exhibit a second-order phase transition in the deep sub-Ohmic regime between two different doubly degenerate phases rather than the standard delocalized-localized transition (Zhao et al., 2014). In a related generalized model with both diagonal and off-diagonal couplings, a Davydov-$\lambda_k$4 analysis found a discontinuous first-order transition between a zero-magnetization state and a finite-magnetization state, while the special case $\lambda_k$5 admits a continuous crossover from a single localized phase to a doubly degenerate localized phase (Lu et al., 2013).

The most explicit physical implementation to date is the analog quantum-simulation proposal based on color centers in free-standing h-BN membranes. There the spin-motion coupling is generated by a magnetic-field gradient, and the membrane geometry and tensile strain control the bath exponent. The proposal accesses $\lambda_k$6, with $\lambda_k$7 realizing $\lambda_k$8 noise and $\lambda_k$9 realizing white noise; in those regimes the calculated dynamics shows coherence revivals at periods set by the bath characteristic frequency and coherent localization of the spin polarization (Abdi et al., 2018). Because $J(\omega)$0 noise and white noise are among the most important decoherence sources in solid-state qubits, this platform was proposed not only as a simulator of dissipative many-body physics but also as a testbed for decoherence mechanisms in solid-state devices (Abdi et al., 2018).

Taken together, these developments show that the sub-Ohmic spin-boson model is not merely a special case of the spin-boson Hamiltonian. It is the regime in which infrared bath structure becomes decisive: for criticality, for the distinction between localization and incoherence, for the validity or failure of Markovian approximations, and for the design of numerical and analog tools capable of resolving strong non-Markovian quantum dissipation (Kirchner et al., 2011, Goulko et al., 2024).