The Journal of Chemical Physics

On the transition from nonadiabatic to adiabatic rate kernel: Schwinger’s stationary variational principle and Padé approximation

Minhaeng Cho, R. Silbey

Published Feb 15, 1997• 22 citations

Abstract

For a two state system coupled to each other by a nonzero matrix element Δ and to the bath arbitrarily, the generalized master equation is derived by applying the well-known projection operator techniques to the quantum Liouville equation. The time-dependent rate kernel is expressed by an infinite summation of the perturbative terms in Fourier–Laplace space. The Schwinger’s stationary variation principle in Hilbert space is extended to Liouville space and then applied to the resummation of the rate kernel. The Cini–Fubini-type trial state vector in Liouville space is used to calculate the variational parameters. It is found that the resulting stationary value for the rate kernel in Fourier–Laplace space is given by the [N,N−1]–Padé approximants, in the N-dimensional subspace constructed by the N perturbatively expanded Liouville space vectors. The (first-order) simplest approximation satisfying the variational principle turns out to be equal to the [1,0] Padé approximant instead of the second-order Fermi golden rule expression. Two well-known approximations, the noninteracting blip approximation (NIBA) and nonadiabatic approximation, are discussed in the context of the [1,0] Padé approximants, based on the variational principle. A higher-order approximation, [2,1] Padé approximant, is also briefly discussed.

DOI: 10.1063/1.473412

Related papers

Padé approximants to the evolution operator through the Lippmann Schwinger variational principle

Fernando Calamante, Horacio Grinberg

Abstract Using the connection between the evolution operator and the stationary value of the Lippmann–Schwinger functional, approximations to this operator are obtained using diagonal Padé approximants. A harmonic oscillator with a non‐hermitean perturbation proportional to powers of the bosonic creation operator is considered and its evolution operator is evaluated. The poles of the spectral representation obtained by this method are compared to both: the ones of the usual perturbative expansion and those of the exact solution. Extensions to Hermitian Hamiltonians are discussed, involving the necessity of inverting more complex operators in the calculation of the Fourier transform. However, the approximation obtained by this procedure becomes exactly unitary. © 1995 John Wiley & Sons, Inc.

International Journal of Quantum ChemistryMay 5, 19950 citations

Adiabatic and nonadiabatic contributions to the energy of a system subject to a time-dependent perturbation: Complete separation and physical interpretation

Anirban Mandal, Katharine L. C. Hunt

When a time-dependent perturbation acts on a quantum system that is initially in the nondegenerate ground state ∣0> of an unperturbed Hamiltonian H(0), the wave function acquires excited-state components ∣k> with coefficients c(k)(t) exp(-iE(k)t/ℏ), where E(k) denotes the energy of the unperturbed state ∣k>. It is well known that each coefficient c(k)(t) separates into an adiabatic term a(k)(t) that reflects the adjustment of the ground state to the perturbation--without actual transitions--and a nonadiabatic term b(k)(t) that yields the probability amplitude for a transition to the excited state. In this work, we prove that the energy at any time t also separates completely into adiabatic and nonadiabatic components, after accounting for the secular and normalization terms that appear in the solution of the time-dependent Schrödinger equation via Dirac's method of variation of constants. This result is derived explicitly through third order in the perturbation. We prove that the cross-terms between the adiabatic and nonadiabatic parts of c(k)(t) vanish, when the energy at time t is determined as an expectation value. The adiabatic term in the energy is identical to the total energy obtained from static perturbation theory, for a system exposed to the instantaneous perturbation λH'(t). The nonadiabatic term is a sum over excited states ∣k> of the transition probability multiplied by the transition energy. By evaluating the probabilities of transition to the excited eigenstates ∣k'(t)> of the instantaneous Hamiltonian H(t), we provide a physically transparent explanation of the result for E(t). To lowest order in the perturbation parameter λ, the probability of finding the system in state ∣k'(t)> is given by λ(2) ∣b(k)(t)∣(2). At third order, the transition probability depends on a second-order transition coefficient, derived in this work. We indicate expected differences between the results for transition probabilities obtained from this work and from Fermi's golden rule.

The Journal of Chemical PhysicsOct 26, 201212 citations

Convergence of high order memory kernels in the Nakajima-Zwanzig generalized master equation and rate constants: Case study of the spin-boson model

Meng Xu, Yaming Yan, Yanying Liu +1 more

The Nakajima-Zwanzig generalized master equation provides a formally exact framework to simulate quantum dynamics in condensed phases. Yet, the exact memory kernel is hard to obtain and calculations based on perturbative expansions are often employed. By using the spin-boson model as an example, we assess the convergence of high order memory kernels in the Nakajima-Zwanzig generalized master equation. The exact memory kernels are calculated by combining the hierarchical equation of motion approach and the Dyson expansion of the exact memory kernel. High order expansions of the memory kernels are obtained by extending our previous work to calculate perturbative expansions of open system quantum dynamics [M. Xu et al., J. Chem. Phys. 146, 064102 (2017)]. It is found that the high order expansions do not necessarily converge in certain parameter regimes where the exact kernel show a long memory time, especially in cases of slow bath, weak system-bath coupling, and low temperature. Effectiveness of the Padé and Landau-Zener resummation approaches is tested, and the convergence of higher order rate constants beyond Fermi's golden rule is investigated.

The Journal of Chemical PhysicsApr 24, 201831 citations

Variance of the energy of a quantum system in a time-dependent perturbation: Determination by nonadiabatic transition probabilities

Anirban Mandal, Katharine L. C. Hunt

For a quantum system in a time-dependent perturbation, we prove that the variance in the energy depends entirely on the nonadiabatic transition probability amplitudes bk(t). Landau and Lifshitz introduced the nonadiabatic coefficients for the excited states of a perturbed quantum system by integrating by parts in Dirac’s expressions for the coefficients ck(1)(t) of the excited states to first order in the perturbation. This separates ck(1)(t) for each state into an adiabatic term ak(1)(t) and a nonadiabatic term bk(1)(t). The adiabatic term follows the adiabatic theorem of Born and Fock; it reflects the adjustment of the initial state to the perturbation without transitions. If the response to a time-dependent perturbation is entirely adiabatic, the variance in the energy is zero. The nonadiabatic term bk(1)(t) represents actual excitations away from the initial state. As a key result of the current work, we derive the variance in the energy of the quantum system and all of the higher moments of the energy distribution using the values of |bk(t)|2 for each of the excited states along with the energy differences between the excited states and the ground state. We prove that the same variance (through second order) is obtained in terms of Dirac’s excited-state coefficients ck(t). We show that the results from a standard statistical analysis of the variance are consistent with the quantum results if the probability of excitation Pk is set equal to |bk(t)|2, but not if the probability of excitation is set equal to |ck(t)|2. We illustrate the differences between the variances calculated with the two different forms of Pk for vibration–rotation transitions of HCl in the gas phase.

The Journal of Chemical PhysicsMar 12, 20202 citations

Decoherence and quantum-classical master equation dynamics

Robbie Grunwald, Raymond Kapral

The conditions under which quantum-classical Liouville dynamics may be reduced to a master equation are investigated. Systems that can be partitioned into a quantum-classical subsystem interacting with a classical bath are considered. Starting with an exact non-Markovian equation for the diagonal elements of the density matrix, an evolution equation for the subsystem density matrix is derived. One contribution to this equation contains the bath average of a memory kernel that accounts for all coherences in the system. It is shown to be a rapidly decaying function, motivating a Markovian approximation on this term in the evolution equation. The resulting subsystem density matrix equation is still non-Markovian due to the fact that bath degrees of freedom have been projected out of the dynamics. Provided the computation of nonequilibrium average values or correlation functions is considered, the non-Markovian character of this equation can be removed by lifting the equation into the full phase space of the system. This leads to a trajectory description of the dynamics where each fictitious trajectory accounts for decoherence due to the bath degrees of freedom. The results are illustrated by computations of the rate constant of a model nonadiabatic chemical reaction.

arXiv:cond-mat/0612203The Journal of Chemical PhysicsMar 20, 200723 citations

Quantum transition probabilities during a perturbing pulse: Differences between the nonadiabatic results and Fermi’s golden rule forms

Anirban Mandal, Katharine L. C. Hunt

For a perturbed quantum system initially in the ground state, the coefficient ck(t) of excited state k in the time-dependent wave function separates into adiabatic and nonadiabatic terms. The adiabatic term ak(t) accounts for the adjustment of the original ground state to form the new ground state of the instantaneous Hamiltonian H(t), by incorporating excited states of the unperturbed Hamiltonian H0 without transitions; ak(t) follows the adiabatic theorem of Born and Fock. The nonadiabatic term bk(t) describes excitation into another quantum state k; bk(t) is obtained as an integral containing the time derivative of the perturbation. The true transition probability is given by bk(t)2, as first stated by Landau and Lifshitz. In this work, we contrast bk(t)2 and ck(t)2. The latter is the norm-square of the entire excited-state coefficient which is used for the transition probability within Fermi’s golden rule. Calculations are performed for a perturbing pulse consisting of a cosine or sine wave in a Gaussian envelope. When the transition frequency ωk0 is on resonance with the frequency ω of the cosine wave, bk(t)2 and ck(t)2 rise almost monotonically to the same final value; the two are intertwined, but they are out of phase with each other. Off resonance (when ωk0 ≠ ω), bk(t)2 and ck(t)2 differ significantly during the pulse. They oscillate out of phase and reach different maxima but then fall off to equal final values after the pulse has ended, when ak(t) ≡ 0. If ωk0 < ω, bk(t)2 generally exceeds ck(t)2, while the opposite is true when ωk0 > ω. While the transition probability is rising, the midpoints between successive maxima and minima fit Gaussian functions of the form a exp[−b(t − d)2]. To our knowledge, this is the first analysis of nonadiabatic transition probabilities during a perturbing pulse.

The Journal of Chemical PhysicsMay 17, 20188 citations