Time-integration of Gaussian variational approximation for the magnetic Schrödinger equation
Malik Scheifinger, Kurt Busch, Marlis Hochbruck, Caroline Lasser · 2025
Read Time-integration of Gaussian variational approximation for the magnetic Schrödinger equation by Malik Scheifinger, Kurt Busch, Marlis Hochbruck, Caroline Lasser (2025) 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-2504-03407.txt.
Introduction
Malik Scheifinger and co-authors present a 2025 study on tracking charged particles in magnetic fields with quantum math. The governing equation is hard to solve in full, so the team follows one Gaussian wave packet as a stand-in. A classic variational rule turns the packet's traits into a set of motion equations. Fast, structure-keeping solvers are the team's focus.
They extend a second-order solver from classical mechanics to the quantum case. They also adapt a fourth-order method for the same task. Tests probe how errors shrink as settings tighten and whether key shapes of the motion stay intact. A Penning trap case, which has an exact known answer, serves as the yardstick. Such solvers help physicists model traps and beams where full quantum runs cost too much.
Catalog record
| CREATOR | Malik Scheifinger, Kurt Busch, Marlis Hochbruck, Caroline Lasser |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-03407
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Time-integration of Gaussian variational approximation for the magnetic Schrödinger equation Malik Scheifinger, Kurt Busch, Marlis Hochbruck, Caroline Lasser Abstract In the present paper we consider the semiclassical magnetic Schrödinger equation, which describes the dynamics of charged particles under the influence of a electro-magnetic field. The solution of the time-dependent Schrödinger equation is approximated by a single Gaussian wave packet via the time-dependent Dirac–Frenkel variational principle. For the approximation we use ordinary differential equations of motion for the parameters of the variational solution and extend the second-order Boris algorithm for classical mechanics to the quantum mechanical case. In addition, we propose a modified version of the classical fourth order Runge–Kutta method. Numerical experiments explore parameter convergence and geometric properties. Moreover, we benchmark against the analytical solution of the Penning trap. Journal: J. Comput. Phys. 1 Introduction In the present paper we study the numerical time-integration for charged quantum particles that are subjected to external magnetic and electric fields. The dynamics is governed by the semiclassical magnetic Schrödinger equation \mathrm{i}\varepsilon\partial_{t}\psi(t)=H(t)\psi(t),\hskip 10.00002pt\psi(0)=\psi_{0},\hskip 10.00002ptt\in\mathbb{R}, (1.1a) on \mathbb{R}^{d} with magnetic Hamiltonian H(t)=\frac{1}{2}\bigl(-\mathrm{i}\varepsilon\nabla_{x}-{A}(t,\cdot)\bigr)^{2}+\phi(t,\cdot), (1.1b) and initial value \psi_{0}\in L^{2}(\mathbb{R}^{d}) with semiclassical parameter 0<\varepsilon\ll 1 . Here, {A} is a divergence-free magnetic vector potential, and \phi is the electric potential. From a numerical point of view, solving this time-dependent partial differential equation raises three major problems. First, it is a high-dimensional problem, since the space dimension is typically given by d=3N , where N is the number of quantum particles in the system. Further, the computational domain \mathbb{R}^{d} is naturally unbounded, and thus most numerical methods require truncation before discretization. For the method of lines (first discretize space, then time), high dimension combined with an unbounded domain leads to inadequately if not intractably large systems that have to be integrated in time. Another challenge is given by the high oscillations induced by the small semiclassical parameter \varepsilon . For standard time integration schemes severe step-size restrictions have to be imposed and leave these methods impracticable. We consider the case that the initial condition \psi_{0} is strongly localized and given by a Gaussian wave packet, and investigate the numerical time-integration of an approximate solution {u}(t)\approx\psi(t) , which is a Gaussian wave packet {u}(t,x)=\exp{\Bigl(\frac{\mathrm{i}}{\varepsilon}\bigl(\frac{1}{2}(x-{q}_{t})^{\top}{\mathcal{C}}_{t}(x-{q}_{t})+(x-{q}_{t})^{\top}{p}_{t}+\zeta_{t}\bigr)\Bigr)}. (1.2) The parameters to be computed are the packet’s position and momentum center {q}_{t},{p}_{t} , the complex width matrix {\mathcal{C}}_{t} of the envelope, and the complex phase and weight parameter \zeta_{t} . These parameters evolve according to ordinary differential equations, which are systematically derived by the Dirac–Frenkel time-dependent variational principle. By the approximation ansatz, high oscillations in time and space are captured and thus eliminated for the numerical time integration. For {A}=0 it is well established that variational Gaussian wave packets offer reasonable mesh-free approximations at low computational cost, see for example [ 1 , 2 ] . More recently, they have also been proposed for magnetic quantum dynamics [ 3 ] . 1.1 Contributions of the paper As our main contribution we derive two fast algorithms to solve the equations of motion for the parameters of a variational Gaussian wave packet approximation such that norm and energy conservation of the quantum solution are echoed by the time integrator. On a standard laptop with a non-optimized Jupyter Notebook 1 1 1 Codes available at https://gitlab.kit.edu/malik.scheifinger/magnschroedti . , these algorithms enable us to compute approximations within minutes. First, we suggest an extension of the Boris algorithm [ 4 , 5 ] , which is a standard integrator for the classical equations of motion for a charged particle system in plasma physics, to the quantum mechanical setting. We furthermore modify the well-known Runge–Kutta 4 method, such that it conserves the L^{2} -norm of a Gaussian wave packet at every time step. Numerical experiments in two and three space dimensions underline the efficiency and expected accuracy of the proposed algorithms. 1.2 Outline of the paper The paper is structured as follows. In Section 2 , we discuss the quantum dynamics of an electron and a proton in a hyperbolic Penning trap as a guiding example for a magnetic Schrödinger equation in the semiclassical regime. In Section 3 , we compare variational and traditional Gaussian wave packet approximation and review the known error estimates. In Section 4 , we transform the system of ordinary differential equations, that determines the variational parameter evolution, in a form that features averaged magnetic momenta and contains the magnetic vector field {{B}}=\nabla\times{A} on the right hand side. In particular, we derive an equation for the imaginary part of the phase parameter \zeta guaranteeing the preservation of the L^{2} -norm. In Section 5 , we briefly introduce the Boris algorithm used and derive our two algorithms for solving the parameters ODEs for the approximating Gaussian wave packet. Finally, in Section 6 , we present numerical experiments for a two-dimensional magnetic system with trigonometric potentials and for the three-dimensional Penning trap. Appendix A gives formulas for the magnetic energy and other averages of Gaussian wave packets. 1.3 Notation For a scalar function a\colon\mathbb{R}^{d}\to\mathbb{R} we denote the Hessian matrix by \nabla^{2}a(x) and for a vector field A\colon\mathbb{R}^{d}\to\mathbb{R}^{d} we denote the Jacobi matrix by J_{A}(x)=(\partial_{\ell}A_{k}(x))_{k,\ell=1}^{d} . For a function W\colon\mathbb{R}^{d}\to\mathbb{R}^{L} , L\geq 1 , and more generally a linear operator \mathbf{A} acting on L^{2}(\mathbb{R}^{d}) we define the averages \displaystyle\langle W\rangle_{{u}}\coloneqq\langle{u}|W{u}\rangle,\hskip 10.00002pt\langle\mathbf{A}\rangle_{{u}} \displaystyle\coloneqq\bigl\langle{u}|\mathbf{A}{u}\bigr\rangle, (1.3) if the inner products exist. We follow the convention that inner products are anti-linear in the first component. We also use the dot product of vectors v,w\in\mathbb{C}^{L} as v\cdot w\coloneqq v^{\top}w=v_{1}w_{1}+\cdots+v_{L}w_{L} and the square v^{2}\coloneqq v\cdot v . For complex matrices {\mathcal{C}}\in\mathbb{C}^{d\times d} we denote the component-wise real and imaginary parts by {{{\mathcal{C}}}_{\textup{R}}},{{{\mathcal{C}}}_{\textup{I}}}\in\mathbb{R}^{d\times d} , respectively. 2 Penning trap To illustrate semiclassical scaling for magnetic quantum dynamics we consider a charged microscopic particle, e.g., an electron or a proton, in a macroscopic hyperbolic Penning trap. Such a trap consists of an arrangement of magnetic coils and electrodes which feature a static magnetic field along the x_{3} -direction and a quadrupole-like static electric field that is rotationally symmetric around the x_{3} -axis, see, e.g., [ 6 ] . The corresponding vector and scalar potentials, respectively, read as \displaystyle{A}(x)=\frac{1}{2}B_{0}\begin{pmatrix}-x_{2}\\ x_{1}\\ 0\end{pmatrix}\hskip 10.00002pt\text{and}\hskip 10.00002pt\phi(x)=\frac{\phi_{0}}{2\delta^{2}}\left(x_{3}^{2}-\frac{1}{2}\left(x_{1}^{2}+x_{2}^{2}\right)\right), (2.1) see Table 1 for typical trap parameters B_{0},\phi_{0},\delta for protons and electrons. A classical point particle (mass m , charge q_{e} ) moving in such an electromagnetic field configuration executes an oscillatory motion along the x_{3} -axis (angular frequency \omega_{3} ), while simultaneously executing an epicyclic motion in the x_{1}x_{2} -plane where the low-frequency magnetron orbit (magnetron frequency \omega_{-} ) is overlayed with high-frequency cyclotron orbits (angular frequency \omega_{+} ), cf. Figure 1 . In terms of the particle and trap parameters, these frequencies are given by \displaystyle\omega_{\pm}=\frac{1}{2}\left(\omega_{c}\pm\Omega\right),\hskip 10.00002pt\omega_{3}=\sqrt{\frac{|q_{e}|\phi_{0}}{m\delta^{2}}},\hskip 10.00002pt\text{where}\hskip 10.00002pt\omega_{c}=\frac{|q_{e}|B_{0}}{m},\hskip 10.00002pt\Omega=\sqrt{\omega_{c}^{2}-2\omega_{3}^{2}}. (2.2) The associated classical trajectories x_{c}(t) may be obtained via Newtonian [ 6 ] or Hamiltonian [ 7 ] approaches. Quantity Electron Proton \delta (Trap Size) 0.00335 m 0.00112 m B_{0} (Magnetic Field) 5.872 T 5.050 T \phi_{0} (Electrode Potential) 10.22 V 53.10 V \nu_{+}=\frac{\omega_{+}}{2\pi} (Corrected Cyclotron Frequency) 164.38 GHz 76.299 MHz \nu_{3}=\frac{\omega_{3}}{2\pi} (Axial Frequency) 63.698 MHz 10.134 MHz \nu_{-}=\frac{\omega_{-}}{2\pi} (Magnetron Frequency) 12.341 kHz 672.93 kHz Table 1: Typical frequencies for the undulatory motion of electrons and protons in a Penning trap, cf. [ 6 , Tables I and II] .