---
title: Quasi-Adiabatic Criterion
url: https://www.emergentmind.com/topics/quasi-adiabatic-criterion
type: topic
---

# Quasi-Adiabatic Criterion

In the cited literature, the expression **quasi-adiabatic criterion** does not denote a single universal formula. It denotes a family of quantitative conditions that separate an adiabatic, quasi-static, or magnetized limit from regimes with finite-rate, stochastic, nonadiabatic, or irreversible effects. Depending on the field, the criterion is formulated as a stochasticity parameter, a constant-adiabaticity condition, a Landau–Zener bound, a relative-entropy benchmark, a superadiabatic threshold, or a finite-memory convergence condition [2009.05644] [1411.5783] [2505.20042] [1607.02193] [1808.03704]. This suggests that quasi-adiabaticity is best understood relationally: it is defined with respect to a chosen adiabatic reference and to a specified mechanism by which that reference fails.

## 1. General conceptual structure

A recurrent feature of quasi-adiabatic criteria is that they interpolate between two limiting descriptions. In plasma shocks, the interpolation is between magnetized heating and demagnetizing stochasticity; in quantum control, between infinitely slow adiabatic following and finite-time transfer; in thermodynamics, between reversible quasi-static adiabatic change and strongly irreversible evolution; in thermal-state preparation, between exact Gibbs structure and finite-time unitary processing [2009.05644] [1411.5783] [1209.5740] [2505.20042].

The criteria also differ in whether they are **local** or **global**. Local criteria compare an instantaneous driving rate to an instantaneous protective scale such as a spectral gap, an electric-field-gradient scale, or an adiabatic slope. Global criteria compare an entire final state to a reference state through quantities such as relative entropy, residual energy, variance, or persistent amplitude [2010.04210] [2510.13555] [1705.00259]. In several papers, the quasi-adiabatic regime is explicitly not the strict adiabatic limit; rather, it is the regime in which the leading non-adiabatic correction remains controlled [1212.4815].

A further conceptual caution appears in the quantum-adiabatic literature. The comment on the “traditional quantitative adiabatic condition” argues that the standard matrix-element-over-gap condition,
\[
\left| \frac{\langle m(t)|\dot n(t)\rangle}{E_m(t)-E_n(t)} \right| \ll 1,
\]
is not established as a general necessary condition for adiabatic approximation, because proofs of necessity can rely on hidden “over-strong” assumptions on the correction sector [1104.0277]. This is relevant because many later quasi-adiabatic criteria use operational diagnostics rather than claiming universal necessity.

## 2. Collisionless shocks and the demagnetization threshold

In collisionless-shock physics, the quasi-adiabatic criterion is formulated through the species-dependent heating function
\[
\chi_j(t,\mathbf r)=\frac{m_j}{q_j B^2}\,\mathrm{div}(\mathbf E_\perp),
\]
where \(m_j\) and \(q_j\) are the mass and charge of species \(j\), \(B=|\mathbf B|\), and \(\mathbf E_\perp\) is the electric field perpendicular to \(\mathbf B\). The parallel electric field is excluded because it does not directly produce the orbit stochasticity of interest. The criterion is explicit: for \(|\chi_j|<1\), particles remain magnetized or adiabatic; for \(|\chi_j|\gtrsim 1\), gyromotion is destabilized and nonadiabatic stochastic heating becomes possible [2009.05644].

For quasi-perpendicular shocks, electrons are in the quasi-adiabatic regime when
\[
|\chi_e|<1.
\]
In that case, conservation of the magnetic moment,
\[
\mu=\frac{m v_\perp^2}{2B},
\]
or equivalently \(T_\perp/B=\mathrm{const}\), supplies the perpendicular energy gain, while wave scattering redistributes that gain into the parallel degree of freedom. Starting from
\[
3\,dT = 2\,T B^{-1}dB,
\]
integration gives the central quasi-adiabatic heating law
\[
\frac{T}{B}=\frac{T_0}{B_0}\left(\frac{B_0}{B}\right)^{1/3},
\qquad
T=T_0\left(\frac{B}{B_0}\right)^{2/3}.
\]
The diagnostic signature is a dip in \(T/B\) at the magnetic overshoot or ramp maximum. In the same events, ions satisfy \(\chi_p\sim 10\)–\(100\), so they lie in the demagnetized stochastic regime instead [2009.05644].

For quasi-parallel shocks, the same stochasticity threshold separates electron quasi-adiabatic and stochastic heating:
\[
|\chi_e|<1 \quad \text{versus} \quad |\chi_e|>1.
\]
The quasi-adiabatic isotropic law is written as
\[
\frac{T}{B}=\frac{T_0}{B_0}\left(\frac{B_0}{B}\right)^{\alpha}.
\]
Here the paper finds that \(\alpha=2/3\) fits the observations better than \(\alpha=1/3\). The modified balance is
\[
6dT=2TB^{-1}dB,
\]
which incorporates an additional energy sink into wave production. In the stochastic regime, the same paper associates bulk electron heating with waves at \((0.4-5)f_{ce}\) and tail acceleration with \(f>5f_{ce}\) [2010.04210].

Taken together, these studies define quasi-adiabaticity as an intermediate regime in which adiabatic invariance remains the source of energization, but waves isotropize or redistribute the gained energy before demagnetization occurs. This suggests that, in shock physics, the quasi-adiabatic criterion is simultaneously a **magnetization criterion** and a **heating-regime classifier**.

## 3. Quantum control, adiabatic passage, and non-Hermitian dynamics

In finite-time quantum control, one common quasi-adiabatic criterion is to keep a standard adiabaticity measure constant in time. In FAQUAD, for a Hamiltonian \(H[\lambda(t)]\), the defining condition is
\[
\hbar \left |\frac{ \langle \phi_1(t)|\partial_t \phi_2(t)\rangle}{ E_1(t)-E_2(t)} \right |
=
\hbar \left |\frac{ \langle \phi_1(t)|\frac{\partial H}{\partial t}| \phi_2(t)\rangle}{ [E_1(t)-E_2(t)]^2} \right |
= c.
\]
This yields
\[
\dot \lambda
=
\mp\frac{c}{\hbar}\left | \frac{ [E_1(\lambda)-E_2(\lambda)]^2}{ \langle \phi_1(\lambda)|\frac{\partial H}{\partial \lambda}|\phi_2(\lambda)\rangle} \right |,
\]
so the protocol automatically slows down near small gaps and speeds up where the gap is large [1411.5783].

A two-parameter generalization appears in path-optimized FAQUAD. For controls \((\kappa,\delta)\), the local adiabaticity parameter is
\[
A = \left| \frac{\langle \psi_2 | \partial_x \psi_1 \rangle}{k_2 - k_1} \right|.
\]
With
\[
\vec{p}(x)=(\kappa(x),\delta(x)), \qquad
\vec E =
\left(
\frac{\langle \psi_2 | \partial_\kappa \psi_1 \rangle}{k_2-k_1},
\frac{\langle \psi_2 | \partial_\delta \psi_1 \rangle}{k_2-k_1}
\right),
\]
the criterion becomes
\[
A=|\vec E\!\cdot\!\vec v|,
\qquad
\int_{x_i}^{x_f}A\,dx=\int_C |\vec E\!\cdot d\vec p|.
\]
The method first minimizes \(\int_C |\vec E\!\cdot d\vec p|\) over paths \(C\), and then imposes constant \(A(x)\) along the chosen path [2511.01697].

A stricter one-parameter criterion is used in SIQUAD. For the Landau–Zener Hamiltonian, the proposed quasiadiabatic parameter is
\[
s'=\frac{1}{2}\frac{\dot{\delta}(t)}{\delta(t)^2+\Omega^2}\ll 1.
\]
Setting \(s'\) constant yields
\[
\delta(t)=\Omega \tan\!\left[\left(\frac{2t}{T}-1\right)\arctan(\delta_m/\Omega)\right].
\]
The method is called state-independent because the practical protocol depends only on the spectral gap and one control parameter, not on explicit adiabatic eigenstate engineering [1809.07579].

In phononic mode conversion, the operative criterion is Landau–Zener-like:
\[
\eta = 1 - e^{-2\pi g^{2}/\left|\kappa \frac{dw}{dx}\right|},
\qquad
\kappa = \frac{d(\beta_r - \beta_l)}{dw}.
\]
High conversion requires
\[
\left|\kappa \frac{dw}{dx}\right| \ll 2\pi g^2.
\]
This criterion governs adiabatic following through an avoided crossing between quasi-Rayleigh and quasi-Love modes [2202.06770].

The same general theme extends to more specialized settings. In quasi-adiabatic WKB for Hamiltonian Grover search, the small parameter is \(\epsilon=1/(\mu t_f)\), and the approximation is found to be useful only when the schedule slows quadratically with the gap, \(g(r)\propto \Delta(r)^{-2}\), corresponding to the Roland–Cerf schedule [1703.10201]. In non-Hermitian exceptional-point encircling, the paper explicitly states
\[
\varepsilon(t)=\left|\frac{f(t)}{2\lambda(t)}\right|\ll 1
\]
as the quasi-adiabatic condition, but the actual switch occurs only after delayed loss of stability, at the departure time determined asymptotically by
\[
|\Delta e^{\Psi(t_+)}|=1
\]
rather than at the instant where \(\Im\lambda\) changes sign [1410.1882].

Across these formulations, quasi-adiabaticity is finite-time and constructive. The criteria do not enforce exact adiabatic following; they distribute or bound nonadiabaticity so that adiabatic-like transfer remains possible on experimentally relevant timescales.

## 4. Many-body, thermal, and imaginary-time criteria

In imaginary-time dynamics, quasi-adiabaticity is defined perturbatively. For a ramp \(\lambda(\tau)\), adiabatic perturbation theory gives
\[
\alpha_n\approx
v_\lambda \frac{\langle n|\partial_\lambda|0\rangle}{(\mathcal E_n-\mathcal E_0)^r}
=
-
v_\lambda \frac{\langle n|\partial_\lambda \mathcal H|0\rangle}{(\mathcal E_n-\mathcal E_0)^{r+1}},
\]
so the quasi-adiabatic regime is the regime where these amplitudes are small. For linear ramps,
\[
\left| v_\lambda \frac{\langle n|\partial_\lambda \mathcal H|0\rangle}{(\mathcal E_n-\mathcal E_0)^2}\right| \ll 1.
\]
Near a quantum critical point, the local gap criterion is replaced by finite-size Kibble–Zurek scaling,
\[
v_{crit} \sim L^{-(zr + 1/\nu)},
\]
and for \(r=1\),
\[
vL^{z+1/\nu}=\text{const}
\quad\text{or smaller}.
\]
In QAQMC this becomes a practical choice of operator-string length,
\[
M\sim L^{d+z+1/\nu}
\]
for linear ramps [1212.4815].

Finite-temperature unitary adiabatic processing uses different diagnostics. In QATE, the ideal adiabatic reference state is
\[
\rho_{\mathrm{min}}
=
\frac{1}{Z}\sum_k e^{-\beta E_k^i}\,|E_k^f\rangle\langle E_k^f|.
\]
The proposed benchmarks are: diagonality in the \(H_f\) eigenbasis, the energy difference
\[
\Delta E_{\mathrm{QATE}}
=
\mathrm{Tr}(\rho(T)H_f)-\mathrm{Tr}(\rho_{\mathrm{min}}H_f),
\]
the variance difference
\[
\Delta \mathrm{Var}_{\mathrm{QATE}}
=
\mathrm{Var}(\rho(T))-\mathrm{Var}(\rho_{\mathrm{min}}),
\]
and off-diagonality measures such as
\[
\mathrm{COD}
=
-\frac{\mathrm{Tr}\!\left([\rho,H_f]^2\right)}{\mathrm{Tr}(\rho^2)}.
\]
For the noncritical transverse-field Ising model with linear ramp, the paper finds
\[
\mathrm{COD},\ \Delta E_{\mathrm{QATE}},\ \Delta \mathrm{Var}_{\mathrm{QATE}}
=
\mathcal{O}\!\left(\frac{N}{T^2}\right)
\]
[2505.20042].

A thermodynamic-limit formulation appears in quasi-adiabatic thermal ensemble preparation. There the success criterion is the specific relative entropy
\[
s_N(\rho\|\rho_{\mathrm g}(\beta))
=
\frac{1}{N}\operatorname{Tr}\!\left[\rho(\ln\rho-\ln\rho_{\mathrm g}(\beta))\right],
\]
with the operational condition
\[
s(\rho\|\rho_{\mathrm g}(\beta))=0
\]
in the thermodynamic limit, because this implies agreement of local observables. In nonintegrable systems at high temperature, one homogeneous parameter can suffice, whereas in the integrable transverse-field Ising model an extensive set of parameters tied to local conserved quantities is generally necessary, and the operation time must scale as \(\tau\propto N\) [2510.13555].

These works treat quasi-adiabaticity as controlled deviation from an ideal mixed-state or thermal target. The target is not always the exact Gibbs state; it may instead be the minimum-energy state compatible with entropy conservation or with the spectrum of the initial density matrix.

## 5. Classical thermodynamics, warm inflation, and superadiabatic convection

In classical thermodynamics, one explicit scalar criterion is the reversibility parameter \(r\) for adiabatic piston expansion:
\[
\frac{T_f}{T_i}=\left(\frac{V_i}{V_f}\right)^{rR/C_V},
\qquad
r=
\sqrt{\frac{54}{\pi}\int_\beta^\infty (\beta-a)^2 e^{-3a^2/2}\,da},
\qquad
\beta=\frac{w}{v_{\rm rms}}.
\]
Here \(r=1\) is the quasi-static reversible adiabat, \(r=0\) is free expansion, and intermediate values quantify the degree of irreversibility [1209.5740].

For reversible ideal-gas paths in the \(P\!-\!V\) plane, the local adiabatic-point criterion is
\[
\delta Q=0
\quad\Longleftrightarrow\quad
\frac{dP}{dV}=-\gamma\frac{P}{V}.
\]
The paper further suggests a local quasi-adiabatic reading when
\[
\left|\frac{dP}{dV}+\gamma\frac{P}{V}\right|
\]
is small. It also distinguishes true adiabaticity from **pseudoadiabatic** processes, for which
\[
\int_a^b \delta Q = 0
\]
but \(\delta Q\) is not identically zero along the path [1801.07302].

In warm inflation, the quasi-stable radiation criterion is the slow-roll approximation
\[
\dot\rho_r \ll 4H\rho_r,\ \Gamma\dot\phi^2,
\qquad
4H\rho_r \simeq \Gamma\dot\phi^2.
\]
The paper argues that this is only a zeroth-order slow-roll statement. If exact adiabatic particle production \(\dot s_I=0\) is imposed together with exact balance, one is driven to the unphysical conclusion that \(\dot n_r\simeq0\), \(\dot T_r\simeq0\), \(\dot\rho_r\simeq0\), and \(\dot p_r\simeq0\). The quasi-stable condition is therefore approximate rather than exact [2112.10813].

In compressible Rayleigh–Bénard convection, the relevant notion is superadiabaticity. The conductive base state is \(T_b(z)=1-az\), while the adiabatic reference profile satisfies
\[
\frac{\mathrm{d} T_a}{\mathrm{d} z} = - \frac{\mathcal{D}\,\alpha_a T_a}{c_{pa}}.
\]
The superadiabatic temperature difference is
\[
\Delta T_{SA}
=
\Delta T-\Delta T_{ad},
\]
and the onset threshold is written in terms of the critical superadiabatic Rayleigh number
\[
Ra_{SA,c} = \frac{27\pi^4}{4} + dRa_{SA}.
\]
For the ideal gas,
\[
\Delta T_{SA}=a-\mathcal D.
\]
This makes the near-adiabatic criterion explicit: the system is close to the adiabatic profile when \(\Delta T_{SA}\) is small, and instability requires positive superadiabaticity [1607.02193].

These classical and continuum formulations show that quasi-adiabaticity need not refer to quantum adiabatic theorems. It can instead denote closeness to a reversible adiabat, to zero local heat flux, to slow-roll balance, or to an adiabatic stratification.

## 6. Operational and numerical criteria

Some literatures use quasi-adiabaticity as an operational or numerical control concept rather than as a physical threshold. In the exactly solvable dragged-oscillator open system, the central diagnostic is the persistent amplitude
\[
\langle E_\mu|{\rm T}\{e^{-\frac{i}{\hbar}\int_0^Tdt\,\hat H_I(t)}\}|E_\mu\rangle.
\]
The dynamics is quasi-adiabatic when its modulus remains close to a pure phase. For cyclic uniform dragging at zero temperature, the condition is
\[
\frac{4k^2v^2}{\hbar} \int d\lambda\, \frac{\kappa_\lambda^2}{|\eta_+(\omega_\lambda)|^2\omega_\lambda} \sin^4\frac{\omega_\lambda T}{4} \ll 1,
\]
while the asymptotic scaling regime is
\[
v\to 0,\qquad T\to\infty,\qquad v^2T=\text{finite}.
\]
At finite temperature, the criterion is tightened by the factor \(\coth(\beta\hbar\omega_\lambda/2)\) [1705.00259].

In QUAPI, by contrast, “quasi-adiabatic” refers to a numerical approximation in which bath memory is finite on the discretized time grid. The operative conditions are: a sufficiently small Trotter step \(\delta t\), with global error \(O(t\,\delta t^2)\), and a sufficiently large memory time
\[
\tau_{\rm mem}=\Delta j_{\max}\,\delta t
\]
so that truncating the discrete influence functional beyond \(\tau_{\rm mem}\) no longer changes observables. The practical convergence criterion is achieved when results remain unchanged upon decreasing \(\delta t\) and increasing \(\tau_{\rm mem}\) further [1808.03704].

This usage suggests a broad editorial point. In some areas, quasi-adiabaticity describes a physical regime of a system under slow driving; in others, it describes a controlled approximation scheme whose validity is established by asymptotic smallness or by numerical convergence.

The term **quasi-adiabatic criterion** therefore has a stable encyclopedic meaning only at a higher level of abstraction. In every domain represented here, it denotes a quantitative test for remaining close to an adiabatic reference while admitting finite-rate, finite-memory, stochastic, or irreversible corrections. What varies is the reference state, the small parameter, and the observable signature of failure.

Source: https://www.emergentmind.com/topics/quasi-adiabatic-criterion