Accelerating Atomistic Quantum Transport Simulations

23 concepts9 papers1 proposal

As semiconductor transistors shrink to the nanometer scale, classical physics models lose their predictive power. Engineering next-generation electronics requires quantum mechanical simulations that model devices atom by atom. However, simulating quantum transport at this scale introduces a massive computational bottleneck.

The standard mathematical framework for these simulations is the Non-Equilibrium Green’s Function (NEGF) formalism. When particles move freely through a perfect crystal lattice (ballistic transport), NEGF calculations are manageable. But real-world nanodevices are full of inelastic interactions: electrons scatter off phonons (lattice vibrations), and phonons scatter off each other (anharmonic scattering).

To model these interactions accurately, researchers traditionally rely on the Self-Consistent Born Approximation (SCBA). The SCBA is an iterative method that repeatedly solves massive matrix equations to account for all possible scattering events until the system reaches equilibrium. Because these matrix operations scale terribly with device size, full SCBA calculations are often too computationally expensive for practical engineering. As noted by Samuel Poncé, shifting quantum transport from an academic exercise to an industrial tool requires physics-informed numerical acceleration and shortcuts that bypass this computational overhead (Special issue on ab initio simulations of electron and phonon transport in nanostructures).

A powerful strategy has emerged to solve this problem: replacing the infinite loops of the SCBA with stabilized, low-order mathematical approximations that deliver near-identical accuracy at a fraction of the cost.

Bypassing the Iterative Loop: The Lowest Order Approximation

Instead of iterating indefinitely to capture every possible scattering diagram, what if we only computed the very first few?

This is the premise of the Lowest Order Approximation (LOA). By truncating the Born series expansion to its lowest-order terms, researchers can avoid the expensive outer loops of the SCBA entirely. However, a naive truncation introduces a severe mathematical problem: polynomial power series expansions tend to diverge or lose accuracy when interactions become strong. A simple low-order truncation fails to capture the true physics of a highly scattering environment.

To overcome this, Y. Lee et al. introduced a method that pairs the Born series expansion with Padé approximants (Efficient quantum modeling of inelastic interactions in nanodevices).

Standard SCBA:  [Initial Guess] ──> [Compute Scattering] ──> [Check Convergence] ── (Repeat 10-100x)
                                                                   │
                                                                   └──> [Final Solution]

LOA + Padé:     [Lowest-Order Terms] ──> [Padé Rational Function] ──> [Final Solution] (Single Step)

Intuitively, while a standard polynomial expansion tries to approximate a complex function using a straight line or simple curve, a Padé approximant represents the series as a rational function (a ratio of two polynomials). This mathematical shift acts as an analytical continuation. It allows the algorithm to accurately approximate the behavior of the full, infinite SCBA series using only the information contained in the lowest-order terms, resulting in massive computational speedups.

Enforcing Physical Laws via Rescaling and Extrapolation

In quantum transport, approximations cannot just be fast; they must also obey fundamental physical laws, such as the conservation of electrical current and energy. Truncating a scattering series often violates these conservation laws, leading to unphysical results.

To resolve this, Y. Lee et al. developed a rescaling technique that extracts current-conserving LOA Green's functions directly from the intermediate steps of the SCBA framework (Quantum treatment of phonon scattering for modeling of three-dimensional atomistic transport). By dividing terms in the self-energy calculation by powers of an artificial scaling factor, the algorithm systematically suppresses higher-order, non-conserving terms during the calculation. This allows the simulation to maintain strict physical conservation laws without executing full matrix iterations.

To further refine accuracy in highly scaled devices like nanowire field-effect transistors (FETs), Youseung Lee et al. combined the LOA-Padé framework with Richardson extrapolation (Quantum Treatment of Inelastic Interactions for the Modeling of Nanowire Field-Effect Transistors). Richardson extrapolation takes solutions calculated at different scaling parameters and mathematically cancels out the leading error terms.

This combined approach yields an order-of-magnitude reduction in simulation time while maintaining greater than 90% accuracy compared to full SCBA. Furthermore, the authors demonstrated that this framework is highly versatile: it works equally well for simulating thermal transport, specifically modeling anharmonic phonon-phonon scattering in complex nanostructures (Anharmonic phonon-phonon scattering modeling of three-dimensional atomistic transport: An efficient quantum treatment).

Alternative Mathematical Shortcuts

The LOA-Padé framework belongs to a broader paradigm of choosing the optimal mathematical representation for a given physical problem. Depending on the specific regime, other shortcuts can be deployed:

By combining these structural reformulations with mathematical series acceleration techniques like LOA-Padé, modern transport simulators can bypass traditional computational walls, making the design of billion-atom nanodevices practically feasible.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_lowest_order_approximation_cb80a96a,
  title = {lowest order approximation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/lowest-order-approximation/rdr_cb80a96a8242409aa14b12ff078c1305},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_lowest_order_approximation_cb80a96a_v1,
  title = {lowest order approximation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/lowest-order-approximation/rdr_cb80a96a8242409aa14b12ff078c1305/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake