Temperature-Dependent Geometric Phase

• Temperature-dependent geometric phase is a phenomenon where the accumulated phase in quantum systems is explicitly modified by temperature through Boltzmann weights, environmental coupling, and decoherence.
• This effect is analyzed via thermal mixed-state approaches, adiabatic system–environment models, and open-system dynamics to reveal how thermal effects alter observable quantum phases.
• Imaginary-time path integrals and topological considerations demonstrate that temperature can induce phase transitions and modify thermodynamic properties in systems with conical intersections.

Temperature-dependent geometric phase denotes a class of geometric or interferometric phase phenomena in which the accumulated phase depends explicitly on temperature, either because the quantum state itself is thermal or mixed, because adiabatic coupling to an equilibrated environment modifies the Berry connection, because nonunitary open-system dynamics makes the phase a functional of temperature-dependent decoherence rates, or because finite-temperature imaginary-time paths sample topologically nontrivial loops such as windings around conical intersections. Across these settings, the core geometric structure remains a holonomy associated with parallel transport or an induced gauge potential, while temperature enters through Boltzmann weights, effective thermal occupations, thermally modified wavefunctions, or thermal path statistics (Ben-Aryeh, 2023, Wang, 29 Apr 2026, Liu, 27 Mar 2026).

1. Conceptual scope and definitions

The zero-temperature reference point is the standard Berry phase for an adiabatically transported eigenstate, or more generally the Aharonov–Anandan and Pancharatnam–Samuel–Bhandari constructions for cyclic or noncyclic pure-state evolution. In the adiabatic setting, the geometric phase is the loop integral of a gauge potential, $\gamma[C] = \oint_C \mathbf{A}(R)\cdot dR$, with $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$ in a Born–Oppenheimer-type reduction (Wang, 29 Apr 2026). In the mixed-state interferometric formulation, the geometric phase is obtained from a trace overlap after imposing parallel transport conditions that remove dynamical phases, $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$ (Ben-Aryeh, 2023).

Temperature dependence arises through several distinct mechanisms. In thermal mixed-state approaches, temperature appears through Gibbs weights in the initial density matrix. In adiabatic system–environment constructions, temperature changes the environment wavefunction and therefore the induced Abelian gauge potential. In open quantum systems, temperature enters dissipative coefficients such as $\Sigma$ and $\Upsilon$, thereby changing the Bloch trajectory and the resulting mixed-state geometric phase. In imaginary-time path integrals, temperature controls the inverse-time length $\beta$, so ring-polymer paths can acquire nontrivial winding around conical intersections, making the geometric phase relevant for thermodynamic observables (Ben-Aryeh, 2023, Capolupo et al., 2015, Liu, 27 Mar 2026).

A recurring structural distinction is between temperature dependence of the geometric object itself and temperature dependence of its observability. In some driven-dissipative oscillator settings, the geometric phase is temperature-independent while thermal noise controls phase diffusion and thus experimental resolution; in others, temperature directly modifies the phase through state populations, thermal gauge connections, or non-equilibrium occupation numbers (2002.04492, Amooghorban et al., 2024).

2. Thermal mixed states and interferometric formulations

A canonical example is the thermal spin-$1/2$ system in a constant magnetic field along $z$ with additional short-time $z$-directed magnetic-field pulses generating an $SU(2)$ evolution. The zeroth-order Hamiltonian is $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$0, and the initial state is the Gibbs mixture

$$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$1

with

$$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$2

In this construction, temperature enters exclusively through the Boltzmann weights $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$3 in the initial density matrix (Ben-Aryeh, 2023).

After imposing the mixed-state parallel transport condition on each eigenvector component, the geometric phase becomes a Pancharatnam phase for the mixed state. For the thermal two-level model, the resulting phase is

$$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$4

with $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$5 and $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$6 determined by the additional controlled fields. This expression makes explicit that the phase is a temperature-weighted version of the pure-state phase: the factor $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$7 interpolates between a fully polarized low-temperature regime and a completely mixed high-temperature regime (Ben-Aryeh, 2023).

The limiting behavior is especially transparent. As $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$8, the state becomes pure and the mixed-state geometric phase reduces to the pure-state Pancharatnam phase, $$\mathbf{A}(R)=i\langle \phi(R,r)|\nabla_R \phi(R,r)\rangle$$9 modulo $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$0. As $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$1, the state approaches $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$2, the two spin contributions cancel, and the geometric phase vanishes. The same trace overlap also determines the reduced interference visibility, so thermal mixing suppresses both phase and contrast (Ben-Aryeh, 2023).

A related but distinct mixed-state generalization appears in $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$3-symmetric quantum mechanics, where the interferometric geometric phase for thermal states is built from complex pure-state geometric phases $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$4. For a closed loop,

$$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$5

Here the imaginary part of $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$6 modifies the thermal distribution itself, producing effective temperatures and allowing finite-temperature geometric phase transitions with discrete jumps at critical points (Wang et al., 2024).

3. Adiabatic system–environment coupling and temperature-dependent gauge potentials

A second major route to temperature dependence preserves a pure adiabatic system state but couples it to a fast environment treated in a Born–Oppenheimer-like approximation. The total Hamiltonian is

$$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$7

with slow coordinates $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$8 and fast environment coordinates $$\gamma_{\text{geom}}(t)=\arg\,\mathrm{Tr}[\rho_0 U(t)]$$9. For fixed $\Sigma$0, one solves the environment eigenproblem and obtains a parametric environment state $\Sigma$1 and eigenvalue $\Sigma$2 (Wang, 29 Apr 2026).

Temperature is introduced by assuming that the environment relaxes rapidly and is in local thermal equilibrium. In the semiclassical ansatz $\Sigma$3, the amplitude is taken to satisfy a Boltzmann distribution,

$\Sigma$4

so the environment wavefunction becomes explicitly temperature dependent. The effective slow-system Hamiltonian is then

$\Sigma$5

with induced Abelian gauge potential

$\Sigma$6

and geometric phase

$\Sigma$7

Because $\Sigma$8 depends on the environment temperature, both the gauge connection and the effective potential are temperature dependent (Wang, 29 Apr 2026).

The model example is the $\Sigma$9 ion, where the nuclei form the slow subsystem and the electron is the fast environment. In that treatment, the electronic orbitals are given a temperature-dependent form,

$\Upsilon$0

with $\Upsilon$1. The resulting gauge potential $\Upsilon$2 and geometric phase $\Upsilon$3 decrease with temperature, while the effective potential minimum shifts from approximately $\Upsilon$4 a.u. at $\Upsilon$5 to approximately $\Upsilon$6 a.u. at $\Upsilon$7, indicating a slight increase in equilibrium bond length with temperature (Wang, 29 Apr 2026).

This framework differs conceptually from thermal mixed-state geometric phases. Temperature does not merely reweight a family of pure-state phases; rather, it alters the adiabatic environment state from which the Berry connection is computed. A plausible implication is that thermal geometry can be embedded directly into effective slow-coordinate dynamics through minimal coupling, not only into interference observables (Wang, 29 Apr 2026).

4. Open-system, dissipative, and non-equilibrium temperature dependence

In open quantum systems, the relevant phase is typically defined for mixed states undergoing nonunitary, noncyclic evolution. For a two-level atom weakly coupled to a field or thermal bath, the reduced density matrix follows a Lindblad-type master equation, and the Wang–Liu kinematic phase yields

$\Upsilon$8

All temperature dependence enters through the environment-dependent coefficients $\Upsilon$9 and $\beta$0, which determine the Bloch-vector damping and hence the path in state space (Capolupo et al., 2015).

For an atom in a thermal bath, these coefficients are

$\beta$1

As temperature increases, decoherence and relaxation accelerate, altering the geometric phase and enabling interferometric thermometry through the phase difference $\beta$2 (Capolupo et al., 2015).

A more spatially structured non-equilibrium setting is the geometric phase of a two-level atom near a dielectric nanosphere whose temperature $\beta$3 differs from the free-space temperature $\beta$4. In that case, the rates can be written as

$\beta$5

where $\beta$6 is the partial local density of photonic states and

$\beta$7

The correction to the geometric phase is therefore controlled jointly by local photonic structure and an effective non-equilibrium thermal parameter. The out-of-equilibrium geometric phase is always bounded between the thermal-equilibrium counterparts, and the temperature difference matters primarily at moderate atom–surface distances where medium and vacuum contributions are comparable (Amooghorban et al., 2024).

Temperature dependence also distinguishes different decoherence mechanisms. In a Davies weak-coupling qubit model, pure dephasing preserves a monotonic, antisymmetric dependence of the geometric phase on the Bloch polar angle, whereas dissipative coupling produces non-monotonic behavior with local extrema and a pronounced temperature dependence near the poles. This sensitivity has been proposed as a diagnostic of the qubit–environment interaction mechanism (Dajka et al., 2010).

5. Thermodynamic and imaginary-time geometric phase effects

In molecular systems with conical intersections, the geometric phase is a topological sign change of adiabatic electronic wavefunctions around a loop enclosing the intersection. In an imaginary-time path-integral formulation, finite temperature enters through the inverse temperature $\beta$8 and therefore through the length and statistics of ring-polymer paths. The multi-electronic-state path integral (MES-PI) formulation expresses the partition function as an integral over nuclear ring polymers multiplied by an electronic trace,

$\beta$9

where $1/2$0 is the overlap matrix between adiabatic electronic states at adjacent beads. The geometric phase is naturally included in this overlap-product trace (Liu, 27 Mar 2026).

For a path that winds around a conical intersection, the ordered product of overlaps acquires a topological factor determined by the winding number. In the Jahn–Teller two-state single-CI model, the winding-number-induced phase factor is simply $1/2$1. The authors introduce a geometric signature matrix and an ad hoc GP-excluded MES-PI by multiplying the standard MES-PI weight by the inverse topological factor, thereby isolating the purely geometric contribution to thermodynamic observables (Liu, 27 Mar 2026).

This construction reveals a pronounced low-temperature effect. At $1/2$2 r.u., the winding-number statistics are approximately $1/2$3 for $1/2$4, $1/2$5 for $1/2$6, $1/2$7 for $1/2$8, and $1/2$9 for $z$0, showing that odd windings—and thus nontrivial geometric phases—occur with substantial probability. As temperature decreases, ring polymers become long enough to encircle the conical intersection more frequently, so geometric phase effects become thermodynamically significant (Liu, 27 Mar 2026).

The clearest manifestation is the heat capacity. The GP-excluded MES Hamiltonian produces an artificial low-temperature heat-capacity peak, whereas the GP-included MES formulation gives the correct low-temperature behavior. At higher temperature, both geometric-phase effects and nuclear quantum effects diminish, and the classical-nuclei MES limit becomes accurate. This identifies temperature-dependent geometric phase as a genuine thermodynamic effect when the thermal de Broglie wavelength becomes comparable to the geometric length scale needed for an imaginary-time path to encircle a conical intersection (Liu, 27 Mar 2026).

6. Criticality, phase transitions, and broader implications

Temperature-dependent geometric phase is closely linked to critical phenomena in several settings. In a central-spin model coupled to an antiferromagnetic environment, the off-diagonal coherence decays as $z$1, with decoherence time $z$2 determined by thermal magnon populations. As temperature increases, $z$3 decreases and the geometric phase is monotonically suppressed. When the external magnetic field approaches the antiferromagnetic spin-flop critical field, the decoherence time collapses and the geometric phase changes abruptly to zero, providing a witness of the many-body phase transition (Yuan et al., 2010).

A different type of criticality appears in $z$4-symmetric quantum mechanics. Because the pure-state phase $z$5 is complex, its imaginary part reweights the thermal distribution. In a two-level model, the generalized interferometric geometric phase exhibits finite-temperature geometric phase transitions: the returning amplitude vanishes at critical points where the magnitudes of the two contributions match and their real parts differ by an odd multiple of $z$6, producing discrete $z$7 jumps in the geometric phase (Wang et al., 2024).

Temperature dependence also constrains implementation in quantum-control settings. In trapped-ion geometric phase gates with Markovian dissipation, finite temperature enters through the thermal occupation $z$8 and stochastic trajectory corrections. The average gate fidelity acquires leading corrections controlled by $z$9, so the geometric phase remains structurally area-like but visibility and motional closure are degraded by thermal fluctuations (Müller et al., 2019). In spin-torque oscillators, by contrast, the geometric phase itself depends only on deterministic parameter cycling, while temperature enters through Langevin noise and phase diffusion; the phase should remain experimentally observable at room temperature if $z$0 is at least $z$1, corresponding to an effective number of spins of order $z$2 for $z$3 (2002.04492).

A persistent misconception is that geometric phases are either strictly temperature-independent or always reduced merely by thermal averaging. The surveyed literature shows both statements are incomplete. In some formulations, temperature enters only through statistical weights and suppresses or cancels the net phase; in others, temperature directly modifies the Berry connection, the effective gauge potential, or the topology sampled by imaginary-time paths; in still others, the geometric phase is unchanged but its observability is temperature-limited by noise and decoherence (Ben-Aryeh, 2023, Wang, 29 Apr 2026, 2002.04492).

Taken together, these results support a broad taxonomy of temperature-dependent geometric phase. One branch concerns mixed-state interferometric phases of thermal ensembles; another concerns pure-state adiabatic phases induced by coupling to equilibrated environments; a third concerns open-system geometric phases shaped by thermal or non-equilibrium reservoirs; and a fourth concerns topological thermodynamic effects in imaginary time. This suggests that “temperature-dependent geometric phase” is not a single formalism but a family of geometrical phenomena unified by the survival of holonomy under finite-temperature state preparation, evolution, or statistical sampling (Capolupo et al., 2015, Liu, 27 Mar 2026).