Post-pulse dipole instability in adiabatic TDDFT: fact or artifact?

Davood B. Dar, Dhyey Ray, Neepa T. Maitra · 2026

Read Post-pulse dipole instability in adiabatic TDDFT: fact or artifact? by Davood B. Dar, Dhyey Ray, Neepa T. Maitra (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-08691.txt.

Introduction

Davood Dar, Dhyey Ray, and Neepa Maitra probe a strange late wobble in molecule dipoles. The wobble shows up long after a pulse of hard light ends. It shows up in a much used time based density method. The method uses a shortcut for how states swap. The team argues the wobble is an artifact. On its own, that way of stepping in time pumps up small, pure swings.

A newer form of the same theory shows no such growth. Edge rules for the grid play the key role, just as the claim says. Tests on nitrogen, with a range of swap rules, back the verdict. Chemists who compute gain a warning. Odd late acts may be a flaw of the scheme, not the molecule.

Catalog record

CREATORDavood B. Dar, Dhyey Ray, Neepa T. Maitra
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.08691

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Post-pulse dipole instability in adiabatic TDDFT: fact or artifact? Davood B. Dar, Dhyey Ray, Neepa T. Maitra Abstract Recent real-time TDDFT calculations have reported an unexpected delayed growth of molecular dipole oscillations some time after an extreme-ultraviolet (XUV) pulse is applied. We show that numerical and analytical arguments suggest that this instability is an artifact of an incorrect non-linearity introduced by the computational approach: Propagation with an adiabatic exchange-correlation approximation within the time-dependent Kohn-Sham equations of time-dependent density functional theory (TDDFT) tends to amplify initially small and pure sinusoidal oscillations in a system. On the other hand, when this same adiabatic approximation is used within the recent response-reformulated RR-TDDFT, the instability is absent. The absorbing boundary condition plays a crucial role consistent with our argument. We demonstrate this explicitly on the N 2 molecule subject to an XUV pulse, with a range of adiabatic functionals. I Introduction Predictive computational modeling of molecular processes far from their equilibrium requires a reliable method to capture both correlated electronic motion as well as coupling to ions. At the shortest timescales the electronic dynamics is of prime importance, and significantly influences the subsequent motion of the heavier nuclei. Even neglecting the nuclear motion, solving the full time-dependent Schrödinger equation (TDSE) for the many-electron wavefunction is impossible beyond a few electrons, and one typically resorts to reduced descriptions and approximations of interaction terms. Time-dependent density functional theory (TDDFT) has emerged as the most practical approach for such problems, which in principle captures the exact time-evolving density of the physical system via a set of non-interacting Kohn-Sham (KS) orbitals evolving in a one-body potential 58 ; 69 ; 40 . A key component in this potential is the exchange-correlation potential, v_{\scriptscriptstyle\rm XC}[n;\Psi(0),\Phi(0)]({\bf r},t) , a functional of the one-body density n({\bf r},t^{\prime}<t) (including its past evolution), the interacting initial state \Psi(0) and the choice of initial Kohn-Sham state \Phi(0) . This is usually approximated by an adiabatic approximation, within which the instantaneous density is input in a chosen ground-state approximation, v_{\scriptscriptstyle\rm XC}^{\rm A}[n;\Psi(0),\Phi(0)]({\bf r},t)=v_{\scriptscriptstyle\rm XC}^{\rm g.s.}[n(t)]({\bf r}) , thus neglecting all the memory-dependence. When orbital-dependent ground-state functionals such as hybrid functionals are used, the corresponding potential depends on the instantaneous orbital, and so has a limited sense of memory. The resulting adiabatic time-dependent Kohn-Sham (TDKS) equations i\,\partial_{t}\phi_{i}(\mathbf{r},t)=\left[-\tfrac{1}{2}\nabla^{2}+v_{\scriptscriptstyle\rm S}(\mathbf{r},t)\right]\phi_{i}(\mathbf{r},t), (1) with v_{\scriptscriptstyle\rm S}=v_{\rm ext}+v_{\scriptscriptstyle\rm H}[n]+v_{\scriptscriptstyle\rm XC}^{\rm A}[n,\Psi(0),\Phi(0)] as the sum of the external, Hartree, and adiabatic exchange-correlation potential, have been applied to simulate a range of photo-excited and laser-driven phenomena in molecules and materials at the atto- and femto-timescales 36 ; 60 ; 62 ; 3 ; 68 ; 40 ; 34 . One such domain of application is the use of ultrashort XUV pulses to investigate detailed time-resolved ultrafast dynamics in molecules and clusters, with the promise to reveal the role of correlations and mechanisms in a number of intriguing processes, from Auger decay to charge-migration 8 ; 73 . A recent series of papers have uncovered an unusual but persistent phenomenon predicted by these simulations 55 ; 56 ; 29 ; 28 ; 12 : a resurgence of a dipole signal some femtoseconds after an ultrafast XUV pulse centered at an off-resonant frequency is turned off. This signal, which has been termed a “dipole-instability”, appears after a period of quiet, almost zero, dipole, that begins to grow with an exponential envelope before reaching a maximum, decaying, and repeating. Such an instability could be measured in principle by time-resolved photo-emission spectroscopy, with a delayed ionization peak corresponding to the resurgence of the dipole. The functional approximation used in the calculations satisfies several exact conditions, including the zero-force theorem, and is energy-conserving, and the dynamics are robust with different choices of numerical parameters. The phenomenon persists across a range of molecules, field strengths, frequencies, and orientations and survives with nuclear motion 29 ; 56 . The authors have highlighted that the key role of the pulse is in creating a population inversion in the system, and demonstrated that the instability also arises from instead creating an instantaneous hole in a deep-lying state; the dependence on the degree of depletion of this state has also been studied 29 ; 56 ; 12 . The population inversion (even partial) has been central in the analyses, where coupling of the upper state to many modes (continuum) is argued to result in a spontaneous symmetry-breaking, backed up by the presence of maxima in phase-space potential energy surfaces near the hole state 12 . While significant insight has been gained, particularly in the drilled hole scenario, the question has remained whether the observed instability is in fact real, or whether it is an artifact of the approximations inherent in the adiabatic TDKS equations. Here, we answer this question by considering two different formulations of TDDFT for real-time non-equilibrium dynamics: one is the traditional TDKS approach of Eq. 1 , and the other is the recent response-reformulated TDDFT (RR-TDDFT) 10 in which all exchange-correlation quantities needed can be obtained from ground-state or response calculations. While both are exact and would give identical results in the formal case where the exact functionals are used, RR-TDDFT has been demonstrated to be more accurate when the adiabatic approximation is used. Using finite-basis set calculations, we reproduce the instability in the case where it develops after an XUV pulse is applied to the molecule in the TDKS simulation, and we find it is absent in the RR-TDDFT simulation. We begin in Section II with a recall of the RR-TDDFT approach. In Sec III , we study the dipole dynamics of N 2 , motivated by the case studied in Ref. 56 , using both TDKS and RR-TDDFT, keeping all other aspects of the calculation the same as much as possible, i.e. the same functional, the same basis set, and approximately the same treatment of continuum states. Consistently, the TDKS calculation yields the dipole-instability, while RR-TDDFT does not; this holds over a range of different field parameters and classes of functional approximations. We provide an argument for how to understand how the instability observed in the TDKS calculation is onset in Sec. III.3 . Finally, we conclude in Sec. IV . II Response-Reformulated TDDFT RR-TDDFT proceeds by propagating a set of ordinary differential equations (ODEs) for the coefficients of the time-dependent many-body state expanded in a basis of field-free many-body states \{\Psi_{m}\} , with a key point being that these many-body states are never explicitly constructed 10 . All quantities entering the ODEs can be obtained from ground-state DFT and response TDDFT, and all properties directly related to the density (i.e. representable by one-body multiplicative operators) can in principle be exactly obtained. Like the traditional TDKS formalism, RR-TDDFT is formally exact and its foundation relies on the Runge–Gross theorem 58 , which proves a one-to-one mapping between the time-evolving density and a one-body multiplicative external potential. But unlike TDKS, the exchange-correlation terms for RR-TDDFT need to be evaluated only near ground-state densities, not on the fully non-equilibrium density, which makes the approach much more accurate when adiabatic functionals are used. The time evolution of the expansion coefficients is governed by i\,\dot{C}_{m}(t)=E_{m}\,C_{m}(t)+\sum_{n}V^{\mathrm{app}}_{mn}(t)\,C_{n}(t), (2) where the matrix elements of the applied potential are V^{\mathrm{app}}_{mn}(t)=\langle\Psi_{m}|V^{\mathrm{app}}(t)|\Psi_{n}\rangle=\int d^{3}r\;v^{\mathrm{app}}(\mathbf{r},t)\,\rho_{mn}(\mathbf{r}). (3) For the common case

Read the complete text