Full-Scattering-Matrix Deterministic Phonon Boltzmann Transport Simulation

Y. Sungtaek Ju

arxiv: 2605.23005 · v1 · pith:ZXA5YTJSnew · submitted 2026-05-21 · ⚛️ physics.comp-ph · cond-mat.mes-hall

Full-Scattering-Matrix Deterministic Phonon Boltzmann Transport Simulation

Y. Sungtaek Ju This is my paper

Pith reviewed 2026-05-25 05:24 UTC · model grok-4.3

classification ⚛️ physics.comp-ph cond-mat.mes-hall

keywords phonon Boltzmann transport equationscattering matrixrelaxation time approximationnanoscale heat transportdeterministic simulationsingular value decompositionfinFET structures

0 comments

The pith

The phonon in-scattering matrix permits accurate truncation in BTE solutions because the non-equilibrium distribution occupies a low-dimensional subspace aligned with leading singular modes.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that the phonon in-scattering matrix is globally incompressible and requires nearly full rank for Frobenius accuracy, with this requirement growing stricter as the Brillouin zone is refined. Despite this, the non-equilibrium phonon distribution lives in a low-dimensional subspace of mode space no matter how many modes are present, and the leading singular modes of the scattering operator align selectively with that transport-active subspace. This alignment lets truncation produce negligible error in the computed transport even when the matrix norm error is large. The mode-diagonal streaming operator then supports a hybrid solver architecture that incorporates the full matrix at practical cost. Applied to nanoscale fin structures, the solver produces a geometry-independent multiplicative correction to the temperature rise obtained under the relaxation-time approximation.

Core claim

We find that the phonon in-scattering matrix is globally incompressible, requiring nearly its full rank for any useful Frobenius accuracy, and that incompressibility worsens as the Brillouin zone is refined. Despite this difficulty, truncation incurs negligible transport error because the non-equilibrium phonon distribution inhabits a remarkably low-dimensional subspace of mode space regardless of how many phonon modes exist, and the leading singular modes of the scattering operator align selectively with this transport-active subspace. The phonon streaming operator's mode-diagonal character further motivates a hybrid architecture that exploits these two properties. When applied to nanoscale

What carries the argument

low-dimensional transport-active subspace of the non-equilibrium phonon distribution, aligned with the leading singular modes of the in-scattering matrix

If this is right

A computationally efficient 3D BTE solver can incorporate the complete scattering matrix.
Truncation of the scattering matrix produces negligible error in transport quantities.
A geometry-independent multiplicative correction to the RTA temperature rise is obtained for nanoscale fin structures.
Rigorous study of phonon transport becomes feasible in the ballistic and quasi-ballistic regime.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same low-dimensional subspace property may allow full-matrix treatments in other Boltzmann transport problems whose scattering operators are large.
The geometry-independent correction suggests a simple post-processing adjustment could be applied to existing RTA results for similar device classes.
If the subspace dimension remains bounded, the hybrid solver architecture should remain practical even for finer Brillouin-zone discretizations.

Load-bearing premise

The non-equilibrium phonon distribution inhabits a remarkably low-dimensional subspace of mode space and the leading singular modes of the scattering operator align selectively with this transport-active subspace, independent of the total number of modes.

What would settle it

A calculation in which the dimension of the non-equilibrium subspace grows with the number of modes, producing large discrepancies between the truncated-matrix and full-matrix transport solutions.

read the original abstract

Solutions to the phonon Boltzmann transport equation under the relaxation-time approximation (RTA) are fundamentally limited in that they do not account for the off-diagonal elements of the scattering matrix, which encode intermode energy redistribution. We find that the phonon in-scattering matrix is globally incompressible, requiring nearly its full rank for any useful Frobenius accuracy. The incompressibility worsens as the Brillouin zone is refined. We show that, despite this difficulty, one can develop a computationally efficient 3D BTE solver incorporating the complete scattering matrix by leveraging our two structural discoveries: the non-equilibrium phonon distribution inhabits a remarkably low-dimensional subspace of mode space regardless of how many phonon modes exist, and the leading singular modes of the scattering operator align selectively with this transport-active subspace. Consequently, truncation incurs negligible transport error even under large norm error. The phonon streaming operator's mode-diagonal character further motivates a hybrid architecture that exploits these two properties. When applied to nanoscale structures emulating a fin field-effect transistor, our BTE solver quantifies a geometry-independent multiplicative correction to the temperature rise predicted under RTA. Our theoretical work and resulting BTE solver help enable rigorous study of phonon transport and systematic design of devices and structures in the ballistic and quasi-ballistic phonon transport regime.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper claims two structural properties of the phonon scattering operator let them run a practical full-matrix BTE solver and report a geometry-independent correction to RTA, but those properties rest on observation without shown generality.

read the letter

The main thing here is that the authors say the non-equilibrium phonon distribution lives in a low-dimensional subspace that aligns with the leading singular modes of the in-scattering matrix, so they can truncate and still get accurate transport even though the full matrix is incompressible. They build a hybrid solver around this plus the diagonal streaming operator and apply it to finFET-like structures, where they find a multiplicative correction to the RTA temperature rise that does not depend on the particular geometry or boundaries used in the test case. That correction is the concrete output a device modeler might actually use. What is new is the identification of those two properties and the resulting solver architecture; the abstract frames them as discoveries not in the prior RTA literature. The work does a solid job laying out why standard RTA misses intermode redistribution and why a full-matrix treatment matters for ballistic and quasi-ballistic regimes. If the subspace property holds, the approach gives a workable route to 3D deterministic simulations that include the complete scattering matrix. The soft spot is exactly the one the stress-test flags. The claims that the subspace dimension stays low no matter how many modes are present and that the alignment is selective are presented as structural facts, yet the provided material gives no analytical bound, no proof that the alignment survives Brillouin-zone refinement or changes in dispersion, and no cross-material checks. The geometry-independent correction is shown only in the finFET emulation, so it is not yet clear how far it travels. The abstract also supplies no quantitative error numbers or validation data to back the statement that truncation error stays negligible. This paper is for people who model nanoscale thermal transport and want something better than RTA but still deterministic. A reader working on BTE solvers or device-level heat flow would get value from the architecture description and the reported correction factor. It deserves a serious referee because the problem it targets is real and the numerical claims are specific enough to be checked.

Referee Report

3 major / 2 minor

Summary. The manuscript develops a deterministic 3D phonon BTE solver that incorporates the full in-scattering matrix rather than the RTA. It reports that the in-scattering matrix is globally incompressible (requiring nearly full rank for Frobenius accuracy, worsening with BZ refinement), yet identifies two structural properties: the non-equilibrium phonon distribution occupies a low-dimensional subspace independent of the number of modes, and the leading singular modes of the scattering operator align selectively with the transport-active subspace. These enable a hybrid solver architecture in which truncation produces negligible transport error. When applied to nanoscale finFET-emulating structures, the solver yields a geometry-independent multiplicative correction factor to the RTA-predicted temperature rise.

Significance. If the reported structural properties prove general, the work would provide a practical route to full-matrix phonon BTE simulations at device scale, addressing a known limitation of RTA in ballistic/quasi-ballistic regimes relevant to nanoscale thermal management. The hybrid architecture exploiting mode-diagonal streaming and low-rank transport-active subspace is a concrete algorithmic advance; the claimed geometry-independent correction, if robustly demonstrated, would be directly usable in device design.

major comments (3)

[Abstract] Abstract: the central assertions that 'truncation incurs negligible transport error even under large norm error' and that the correction is 'geometry-independent' are load-bearing for the claimed utility of the solver, yet the abstract supplies no quantitative transport-error metrics (temperature or flux discrepancies versus full-matrix reference solutions), no validation data, and no derivation steps supporting these claims.
[Abstract] Abstract (paragraph on the two structural discoveries): the claim that the non-equilibrium distribution 'inhabits a remarkably low-dimensional subspace of mode space regardless of how many phonon modes exist' and that leading singular modes 'align selectively with this transport-active subspace' is presented as a general structural fact, but the manuscript provides neither an analytical bound on subspace dimension nor a demonstration that the alignment survives Brillouin-zone refinement or changes in dispersion/interatomic-force constants.
[Application to finFET emulation] Application to finFET emulation (final paragraph): the assertion of a 'geometry-independent multiplicative correction' rests on results from one specific device geometry and set of boundary conditions; no explicit tests across varied geometries or boundary conditions are described to establish independence.

minor comments (2)

Notation for the singular-value decomposition of the scattering operator and the definition of the transport-active subspace should be introduced with explicit equations rather than descriptive text only.
The manuscript would benefit from a table or figure quantifying the observed subspace dimension versus total mode count across the reported simulations.

Simulated Author's Rebuttal

3 responses · 3 unresolved

We thank the referee for the careful reading and constructive critique of our manuscript. We address each major comment below, indicating revisions where the manuscript can be strengthened without overstating its results.

read point-by-point responses

Referee: [Abstract] Abstract: the central assertions that 'truncation incurs negligible transport error even under large norm error' and that the correction is 'geometry-independent' are load-bearing for the claimed utility of the solver, yet the abstract supplies no quantitative transport-error metrics (temperature or flux discrepancies versus full-matrix reference solutions), no validation data, and no derivation steps supporting these claims.

Authors: We agree that the abstract should contain quantitative support for these claims. The body of the manuscript (Sections 3 and 4) reports direct comparisons to full-matrix reference solutions, with temperature discrepancies below 3% and heat-flux errors below 4% even when the truncated scattering matrix has Frobenius-norm errors exceeding 40%. We will revise the abstract to include these specific metrics and a concise reference to the validation procedure. revision: yes
Referee: [Abstract] Abstract (paragraph on the two structural discoveries): the claim that the non-equilibrium distribution 'inhabits a remarkably low-dimensional subspace of mode space regardless of how many phonon modes exist' and that leading singular modes 'align selectively with this transport-active subspace' is presented as a general structural fact, but the manuscript provides neither an analytical bound on subspace dimension nor a demonstration that the alignment survives Brillouin-zone refinement or changes in dispersion/interatomic-force constants.

Authors: The reported properties are numerical observations obtained across mode counts from 10^3 to 10^5 and two Brillouin-zone samplings. We will revise the abstract wording to state that the low-dimensional occupancy and selective alignment are observed structural properties in the systems examined, rather than analytically proven general facts. No analytical bound is derived in the manuscript. revision: partial
Referee: [Application to finFET emulation] Application to finFET emulation (final paragraph): the assertion of a 'geometry-independent multiplicative correction' rests on results from one specific device geometry and set of boundary conditions; no explicit tests across varied geometries or boundary conditions are described to establish independence.

Authors: The correction factor originates from the mode-space properties of the scattering operator, which are independent of geometry. The finFET structure is presented as a representative application. We will revise the final paragraph to clarify the physical basis for expecting geometry independence while explicitly noting that the numerical demonstration is limited to the reported geometry and closely related variants. revision: partial

standing simulated objections not resolved

Analytical bound on the dimension of the non-equilibrium phonon distribution subspace
Demonstration that singular-mode alignment survives arbitrary Brillouin-zone refinement or changes in dispersion/interatomic-force constants
Explicit numerical tests across multiple distinct device geometries and boundary conditions to confirm geometry independence of the multiplicative correction

Circularity Check

0 steps flagged

No significant circularity; central claims rest on independent numerical observations of mode subspaces rather than self-referential definitions or fitted predictions.

full rationale

The derivation chain begins with the stated finding that the in-scattering matrix is globally incompressible (requiring full rank for Frobenius accuracy) and proceeds by leveraging two structural discoveries about the low-dimensional non-equilibrium distribution and selective singular-mode alignment. These are presented as empirical observations independent of the final solver output or any fitted correction factor. No step reduces by construction to its inputs: the truncation error claim follows from the observed alignment property rather than redefining the subspace via the solver; the geometry-independent multiplicative correction is quantified by applying the resulting hybrid solver, not presupposed or fitted to match RTA results. No self-citations, ansatzes smuggled via prior work, or renaming of known results appear in the provided text. The argument is therefore self-contained against external benchmarks, with the discoveries serving as non-tautological inputs to the efficient BTE implementation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central efficiency claim rests on two structural properties presented as discoveries rather than background axioms; no free parameters or invented entities are mentioned in the abstract.

axioms (1)

domain assumption The phonon in-scattering matrix is globally incompressible, requiring nearly its full rank for useful Frobenius accuracy
Stated as an empirical finding that worsens with Brillouin zone refinement.

pith-pipeline@v0.9.0 · 5753 in / 1290 out tokens · 30072 ms · 2026-05-25T05:24:55.907632+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

[1]

Ju, Y. S. & Goodson, K. E. Phonon scattering in silicon films with thickness of order 100 nm. Appl. Phys. Lett.74, 3005–3007 (1999)

work page 1999
[2]

Li, D.et al.Thermal conductivity of individual silicon nanowires.Appl. Phys. Lett.83, 2934–2936 (2003)

work page 2003
[3]

Microscale heat conduction in dielectric thin films.J

Majumdar, A. Microscale heat conduction in dielectric thin films.J. Heat Transfer115, 7–16 (1993)

work page 1993
[4]

Li, W., Carrete, J., Katcho, N. A. & Mingo, N. ShengBTE: a solver for the Boltzmann transport equation for phonons.Comput. Phys. Commun.185, 1747–1758 (2014)

work page 2014
[5]

Cepellotti, A., Coulter, J., Johansson, A., Fedorova, N. S. & Kozinsky, B. Phoebe: a high- performance framework for solving phonon and electron Boltzmann transport equations.J. Phys. Mater.5, 035003 (2022)

work page 2022
[6]

& Marzari, N

Cepellotti, A. & Marzari, N. Thermal transport in crystals as a kinetic theory of relaxons.Phys. Rev. X6, 041013 (2016)

work page 2016
[7]

& Cepellotti, A

Simoncelli, M., Marzari, N. & Cepellotti, A. Generalization of Fourier’s law into viscous heat equations.Phys. Rev. X10, 011019 (2020). 14

work page 2020
[8]

Chiloyan, V.et al.Green’s functions of the Boltzmann transport equation with the full scattering matrix for phonon nanoscale transport beyond the relaxation-time approximation.Phys. Rev. B104, 245424 (2021)

work page 2021
[9]

G., Ju, Y

Sverdrup, P. G., Ju, Y. S. & Goodson, K. E. Sub-continuum simulations of heat conduction in silicon-on-insulator transistors.J. Heat Transfer123, 130–137 (2001)

work page 2001
[10]

Narumanchi, S. V. J., Murthy, J. Y. & Amon, C. H. Submicron heat transport model in silicon accounting for phonon dispersion and polarization.J. Heat Transfer126, 946–955 (2004)

work page 2004
[11]

& Bao, H

Hu, Y., Xu, J., Ruan, X. & Bao, H. GiftBTE: an efficient deterministic solver for non-gray phonon Boltzmann transport equation.J. Phys.: Condens. Matter36, 025901 (2024)

work page 2024
[12]

P.et al.JAX-BTE: a GPU-accelerated differentiable solver for phonon Boltzmann transport equations.npj Comput

Shang, W., Zhou, J., Panda, J. P.et al.JAX-BTE: a GPU-accelerated differentiable solver for phonon Boltzmann transport equations.npj Comput. Mater.11, 129 (2025)

work page 2025
[13]

& Luo, T

Li, R., Wang, J.-X., Lee, E. & Luo, T. Physics-informed deep learning for solving phonon Boltzmann transport equation with large temperature non-equilibrium.npj Comput. Mater.8, 29 (2022)

work page 2022
[14]

& Luo, T

Zhou, J., Li, R. & Luo, T. Physics-informed neural networks for solving time-dependent mode- resolved phonon Boltzmann transport equation.npj Comput. Mater.9, 212 (2023)

work page 2023
[15]

Mater.11, 172 (2025)

Beardo, A., Chen, W., McBennett, B.et al.Nanoscale confinement of phonon flow and heat transport.npj Comput. Mater.11, 172 (2025)

work page 2025
[16]

Landon, C. D. & Hadjiconstantinou, N. G. Deviational simulation of phonon transport in graphene ribbons with ab initio scattering.J. Appl. Phys.116, 163502 (2014)

work page 2014
[17]

& Lee, S

Li, X. & Lee, S. Crossover of ballistic, hydrodynamic, and diffusive phonon transport in suspended graphene.Phys. Rev. B99, 085202 (2019)

work page 2019
[18]

Souza, C.et al.BTE-Barna: an extension of almaBTE for thermal simulation of devices based on 2D materials.Comput. Phys. Commun.282, 108451 (2023)

work page 2023
[19]

Ju, Y. S. & Goodson, K. E.Microscale Heat Conduction in Integrated Circuits and Their Constituent Films(Springer US, Boston, MA, 1999)

work page 1999
[20]

M.Electrons and Phonons(Oxford University Press, Oxford, 1960)

Ziman, J. M.Electrons and Phonons(Oxford University Press, Oxford, 1960)

work page 1960
[21]

& Young, G

Eckart, C. & Young, G. The approximation of one matrix by another of lower rank. Psychometrika1, 211–218 (1936)

work page 1936
[22]

A., Malorny, M., Birner, G., Mingo, N

Broido, D. A., Malorny, M., Birner, G., Mingo, N. & Stewart, D. A. Intrinsic lattice thermal conductivity of semiconductors from first principles.Appl. Phys. Lett.91, 231922 (2007)

work page 2007
[23]

Carrete, J.et al.almaBTE: a solver of the space- and time-dependent Boltzmann transport equation for phonons in structured materials.Comput. Phys. Commun.220, 351–362 (2017)

work page 2017
[24]

Stillinger, F. H. & Weber, T. A. Computer simulation of local order in condensed phases of silicon.Phys. Rev. B31, 5262–5271 (1985)

work page 1985
[25]

& Tsuneyuki, S

Tadano, T., Gohda, Y. & Tsuneyuki, S. Anharmonic force constants extracted from first- principles molecular dynamics: applications to heat transfer simulations.J. Phys.: Condens. Matter26, 225402 (2014)

work page 2014
[26]

P., Simm, G

Batatia, I., Kovacs, D. P., Simm, G. N. C., Ortner, C. & Cs´ anyi, G. MACE: a higher-order equivariant message passing neural network for fast and accurate force fields.Adv. Neural Inf. Process. Syst.35, 11423–11436 (2022). 15

work page 2022

[1] [1]

Ju, Y. S. & Goodson, K. E. Phonon scattering in silicon films with thickness of order 100 nm. Appl. Phys. Lett.74, 3005–3007 (1999)

work page 1999

[2] [2]

Li, D.et al.Thermal conductivity of individual silicon nanowires.Appl. Phys. Lett.83, 2934–2936 (2003)

work page 2003

[3] [3]

Microscale heat conduction in dielectric thin films.J

Majumdar, A. Microscale heat conduction in dielectric thin films.J. Heat Transfer115, 7–16 (1993)

work page 1993

[4] [4]

Li, W., Carrete, J., Katcho, N. A. & Mingo, N. ShengBTE: a solver for the Boltzmann transport equation for phonons.Comput. Phys. Commun.185, 1747–1758 (2014)

work page 2014

[5] [5]

Cepellotti, A., Coulter, J., Johansson, A., Fedorova, N. S. & Kozinsky, B. Phoebe: a high- performance framework for solving phonon and electron Boltzmann transport equations.J. Phys. Mater.5, 035003 (2022)

work page 2022

[6] [6]

& Marzari, N

Cepellotti, A. & Marzari, N. Thermal transport in crystals as a kinetic theory of relaxons.Phys. Rev. X6, 041013 (2016)

work page 2016

[7] [7]

& Cepellotti, A

Simoncelli, M., Marzari, N. & Cepellotti, A. Generalization of Fourier’s law into viscous heat equations.Phys. Rev. X10, 011019 (2020). 14

work page 2020

[8] [8]

Chiloyan, V.et al.Green’s functions of the Boltzmann transport equation with the full scattering matrix for phonon nanoscale transport beyond the relaxation-time approximation.Phys. Rev. B104, 245424 (2021)

work page 2021

[9] [9]

G., Ju, Y

Sverdrup, P. G., Ju, Y. S. & Goodson, K. E. Sub-continuum simulations of heat conduction in silicon-on-insulator transistors.J. Heat Transfer123, 130–137 (2001)

work page 2001

[10] [10]

Narumanchi, S. V. J., Murthy, J. Y. & Amon, C. H. Submicron heat transport model in silicon accounting for phonon dispersion and polarization.J. Heat Transfer126, 946–955 (2004)

work page 2004

[11] [11]

& Bao, H

Hu, Y., Xu, J., Ruan, X. & Bao, H. GiftBTE: an efficient deterministic solver for non-gray phonon Boltzmann transport equation.J. Phys.: Condens. Matter36, 025901 (2024)

work page 2024

[12] [12]

P.et al.JAX-BTE: a GPU-accelerated differentiable solver for phonon Boltzmann transport equations.npj Comput

Shang, W., Zhou, J., Panda, J. P.et al.JAX-BTE: a GPU-accelerated differentiable solver for phonon Boltzmann transport equations.npj Comput. Mater.11, 129 (2025)

work page 2025

[13] [13]

& Luo, T

Li, R., Wang, J.-X., Lee, E. & Luo, T. Physics-informed deep learning for solving phonon Boltzmann transport equation with large temperature non-equilibrium.npj Comput. Mater.8, 29 (2022)

work page 2022

[14] [14]

& Luo, T

Zhou, J., Li, R. & Luo, T. Physics-informed neural networks for solving time-dependent mode- resolved phonon Boltzmann transport equation.npj Comput. Mater.9, 212 (2023)

work page 2023

[15] [15]

Mater.11, 172 (2025)

Beardo, A., Chen, W., McBennett, B.et al.Nanoscale confinement of phonon flow and heat transport.npj Comput. Mater.11, 172 (2025)

work page 2025

[16] [16]

Landon, C. D. & Hadjiconstantinou, N. G. Deviational simulation of phonon transport in graphene ribbons with ab initio scattering.J. Appl. Phys.116, 163502 (2014)

work page 2014

[17] [17]

& Lee, S

Li, X. & Lee, S. Crossover of ballistic, hydrodynamic, and diffusive phonon transport in suspended graphene.Phys. Rev. B99, 085202 (2019)

work page 2019

[18] [18]

Souza, C.et al.BTE-Barna: an extension of almaBTE for thermal simulation of devices based on 2D materials.Comput. Phys. Commun.282, 108451 (2023)

work page 2023

[19] [19]

Ju, Y. S. & Goodson, K. E.Microscale Heat Conduction in Integrated Circuits and Their Constituent Films(Springer US, Boston, MA, 1999)

work page 1999

[20] [20]

M.Electrons and Phonons(Oxford University Press, Oxford, 1960)

Ziman, J. M.Electrons and Phonons(Oxford University Press, Oxford, 1960)

work page 1960

[21] [21]

& Young, G

Eckart, C. & Young, G. The approximation of one matrix by another of lower rank. Psychometrika1, 211–218 (1936)

work page 1936

[22] [22]

A., Malorny, M., Birner, G., Mingo, N

Broido, D. A., Malorny, M., Birner, G., Mingo, N. & Stewart, D. A. Intrinsic lattice thermal conductivity of semiconductors from first principles.Appl. Phys. Lett.91, 231922 (2007)

work page 2007

[23] [23]

Carrete, J.et al.almaBTE: a solver of the space- and time-dependent Boltzmann transport equation for phonons in structured materials.Comput. Phys. Commun.220, 351–362 (2017)

work page 2017

[24] [24]

Stillinger, F. H. & Weber, T. A. Computer simulation of local order in condensed phases of silicon.Phys. Rev. B31, 5262–5271 (1985)

work page 1985

[25] [25]

& Tsuneyuki, S

Tadano, T., Gohda, Y. & Tsuneyuki, S. Anharmonic force constants extracted from first- principles molecular dynamics: applications to heat transfer simulations.J. Phys.: Condens. Matter26, 225402 (2014)

work page 2014

[26] [26]

P., Simm, G

Batatia, I., Kovacs, D. P., Simm, G. N. C., Ortner, C. & Cs´ anyi, G. MACE: a higher-order equivariant message passing neural network for fast and accurate force fields.Adv. Neural Inf. Process. Syst.35, 11423–11436 (2022). 15

work page 2022