Classical limit for the varying-mass Schrödinger equation with random inhomogeneities

Shi Chen, Qin Li, Xu YangJournal of Computational Physics 2021 · DOI

TL;DR

This paper derives a radiative transfer equation as the asymptotic limit of the varying-mass Schrödinger equation (VMSE) in the semiclassical regime with random inhomogeneities. This is achieved by systematically applying the Wigner transform and expanding the Wigner equation. The results are numerically validated, showing good agreement between the VMSE solutions and their limiting radiative transfer equation.

What Problem Does This Paper Address?

Simulating the varying-mass Schrödinger equation (VMSE), particularly in the classical regime where the Planck constant is small, presents significant computational challenges. This is especially true when the material properties, represented by the varying mass term, are random and highly oscillatory. Standard numerical solvers struggle to resolve the small wavelengths, and handling randomness requires computationally expensive methods like Monte Carlo sampling over many realizations. This paper aims to bridge this gap by deriving a simpler, asymptotic limit equation – a radiative transfer equation – that can guide the design of more efficient numerical schemes and provide theoretical understanding of the system's behavior in the classical limit.

Methods

The core of this work lies in the application of the Wigner transform. This technique maps quantum mechanical wave functions to a phase-space representation, which is crucial for analyzing the classical limit. The authors systematically expand the Wigner equation derived from the VMSE in powers of the rescaled Planck constant $\varepsilon$. By neglecting higher-order terms in $\varepsilon$, they derive a radiative transfer equation as the asymptotic limit. For numerical validation, they employ:

  • Finite Difference methods for solving the Schrödinger equation, with mesh sizes and time steps adapted to resolve the small $\varepsilon$ scales.
  • Spectral methods for efficient computation of spatial derivatives.
  • Monte Carlo sampling combined with Karhunen-Loève (KL) decomposition to handle the random inhomogeneities in the effective mass.
  • Particle methods and typical ODE solvers for the deterministic Liouville equation (the limit of the Wigner equation without randomness).

Datasets

Not discussed in the paper. The authors use synthetic initial conditions and parameters for their numerical simulations.

Key Insights and Ideas

  • Bridging Quantum and Classical Regimes: The paper demonstrates a rigorous mathematical framework for transitioning from a quantum mechanical description (VMSE) to a classical one (radiative transfer equation) in the presence of random material properties.
  • Wigner Transform as a Key Tool: The Wigner transform proves to be a powerful tool for analyzing the semiclassical limit of quantum equations, especially when dealing with complex phenomena like varying mass and random inhomogeneities.
  • Radiative Transfer as a Useful Limit: The derived radiative transfer equation offers a computationally tractable alternative for understanding the macroscopic behavior of systems described by VMSE in the semiclassical limit, particularly when randomness is present.
  • Impact of Randomness Scale: The numerical results suggest that the scale of the random perturbation in the effective mass plays a critical role in the resulting dynamics and the convergence to the radiative transfer limit.

Key Results

  • The paper successfully derives a radiative transfer equation as the asymptotic limit of the VMSE with random inhomogeneities in the semiclassical regime ($\varepsilon \ll 1$).
  • Numerical simulations confirm that the solutions of the VMSE and its limiting radiative transfer equation agree well in the classical limit.
  • For the deterministic case (Theorem 2.4), numerical results show that the errors in particle density and flux decay at a rate of $O(\varepsilon^2)$.
  • For the random case (Main Result 3.1), numerical results show a good agreement between the expected values of the particle density and current density from VMSE and the radiative transfer equation. The errors in these quantities are observed to decay at a rate of $O(\varepsilon)$, which is stronger than theoretically predicted for $W^{(1)}$ but suggests a higher-order convergence.
  • The computational cost of solving VMSE increases significantly with decreasing $\varepsilon$, highlighting the practical advantage of using the derived radiative transfer equation for large-scale simulations.

Limitations and Future Work

  • Rigorous Justification: While the derivation of the asymptotic limit is presented formally, the rigorous mathematical justification for the convergence of the Wigner transform in the presence of random potentials is left for future research. This involves conditions on the Fourier transform of the random mass being a Markov process.
  • Randomness Scale: The paper explores different scales of random perturbation ($O(\sqrt{\varepsilon})$ and $O(\varepsilon)$). The $O(\sqrt{\varepsilon})$ scale is critical for the radiative transfer limit. The behavior for other scales, like $O(\varepsilon^{0.4})$, shows significantly different scattering, suggesting that the precise scaling of randomness is crucial.
  • Computational Cost: The numerical simulations for VMSE become prohibitively expensive for very small $\varepsilon$, reinforcing the need for asymptotic-preserving schemes or direct use of the limit equation.
  • Future Work: The authors suggest extending the rigorous justification of the results and exploring the impact of different scales of random perturbations. Further investigation into the statistical properties of the solutions, such as standard deviation and covariance, is also suggested.

Summary updated

BibTeX
@article{chen2021classical,
  author = {Shi Chen and Qin Li and Xu Yang},
  title = {{Classical limit for the varying-mass Schrödinger equation with random inhomogeneities}},
  journal = {Journal of Computational Physics},
  year = {2021},
  doi = {10.1016/j.jcp.2021.110365},
}
Figure