pith. sign in

arxiv: 2509.20809 · v2 · submitted 2025-09-25 · ⚛️ physics.optics · cs.NA· math.NA· physics.comp-ph

Fast 3D Nanophotonic Inverse Design using Volume Integral Equations

Amirhossein Fallah , Constantine Sideris This is my paper

Pith reviewed 2026-05-18 14:31 UTC · model grok-4.3

classification ⚛️ physics.optics cs.NAmath.NAphysics.comp-ph
keywords nanophotonicsinverse designvolume integral equationadjoint methodcomputational efficiency3D photonicselectromagnetic simulationoptimization
0
0 comments X

The pith

Volume integral equations replace finite-difference solvers to accelerate 3D nanophotonic inverse design by multiple orders of magnitude.

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

The paper establishes that a volume integral equation formulation can act as a fast forward model inside inverse-design optimization loops for nanophotonic devices. If correct, this substitution would let designers run far more iterations or handle larger structures within practical run times. The authors derive a matching adjoint method to compute gradients efficiently and add a unidirectional mode excitation technique suited to the integral-equation setting. Benchmarks then show the new approach runs multiple orders of magnitude faster than conventional finite-difference solvers in both time and frequency domains. The method is demonstrated by designing a 3 dB power splitter, a dual-wavelength Bragg grating, and a selective mode reflector.

Core claim

The central claim is that the volume integral equation formulation, supplied with a derived adjoint method for gradient computation and a unidirectional mode excitation strategy, delivers multiple orders of magnitude improvement in computational efficiency over conventional finite-difference methods for both time- and frequency-domain simulations used in nanophotonic inverse design, as confirmed by direct benchmarks and by the successful creation of three representative devices.

What carries the argument

The volume integral equation formulation together with its tailored adjoint method for optimization gradients and unidirectional mode excitation strategy.

If this is right

  • Optimization loops for nanophotonic devices become feasible at larger scales and higher iteration counts.
  • Both time-domain and frequency-domain analyses inside the same workflow gain the reported efficiency.
  • Concrete devices such as power splitters, Bragg gratings, and mode reflectors can be designed end-to-end with the new solver.
  • Overall runtime advantages shorten the full inverse-design cycle for next-generation optical components.

Where Pith is reading between the lines

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

  • The speed-up could open inverse design to structures whose electrical size currently makes finite-difference runs prohibitive.
  • The same integral-equation machinery might be reused for related inverse problems in acoustics or larger-scale electromagnetics.
  • Pairing the solver with gradient-based or learning-assisted optimizers could reduce total design time even further.

Load-bearing premise

The volume integral equation discretization and adjoint derivation remain accurate and stable for the subwavelength feature sizes and material contrasts typical of the target nanophotonic devices.

What would settle it

A side-by-side run of the identical inverse-design task using both the volume integral equation solver and a converged finite-difference reference that produces substantially different optimized device performance.

Figures

Figures reproduced from arXiv: 2509.20809 by Amirhossein Fallah, Constantine Sideris.

Figure 1
Figure 1. Figure 1: Straight silicon waveguide under fundamental mode excitation. (a) Simulation [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Runtime comparison of the JVIE, commercial FDTD, and FDFD solvers vs. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Fabry-Parot resonator with selective mode reflectors on each end. (b) Selective [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Power splitter simulation setup: The mode source is enforced on a plane per [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
read the original abstract

Designing nanophotonic devices with minimal human intervention has gained substantial attention due to the complexity and precision required in modern optical technologies. While inverse design techniques typically rely on conventional electromagnetic solvers as forward models within optimization routines, the substantial electrical size and subwavelength characteristics of nanophotonic structures necessitate significantly accelerated simulation methods. In this work, we introduce a forward modeling approach based on the volume integral equation (VIE) formulation as an efficient alternative to traditional finite-difference (FD)-based methods. We derive the adjoint method tailored specifically for the VIE framework to efficiently compute optimization gradients and present a novel unidirectional mode excitation strategy compatible with VIE solvers. Comparative benchmarks demonstrate that our VIE-based approach provides multiple orders of magnitude improvement in computational efficiency over conventional FD methods in both time and frequency domains. To validate the practical utility of our approach, we successfully designed three representative nanophotonic components: a 3 dB power splitter, a dual-wavelength Bragg grating, and a selective mode reflector. Our results underscore the significant runtime advantages offered by the VIE-based framework, highlighting its promising role in accelerating inverse design workflows for next-generation nanophotonic devices.

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.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces a volume integral equation (VIE) formulation as an efficient forward model for 3D nanophotonic inverse design. It derives an adjoint method for gradient computation within the VIE framework, presents a unidirectional mode excitation strategy, reports comparative benchmarks claiming multiple orders of magnitude speedup over finite-difference (FD) solvers in time and frequency domains, and demonstrates the method by designing a 3 dB power splitter, a dual-wavelength Bragg grating, and a selective mode reflector.

Significance. If the VIE discretization and adjoint remain accurate for subwavelength high-contrast features, the reported efficiency gains would enable substantially faster inverse-design loops for complex nanophotonic devices, reducing reliance on computationally expensive FD solvers and accelerating exploration of 3D structures.

major comments (2)
  1. [Abstract] Abstract: the central claim that 'comparative benchmarks demonstrate... multiple orders of magnitude improvement in computational efficiency' is not supported by any reported quantitative error metrics (e.g., L2 field errors, transmission discrepancies), mesh-convergence data, or explicit accuracy tolerances relative to the FD reference solvers. Without these, the runtime advantage cannot be assessed at equivalent fidelity for the target subwavelength, high-contrast devices.
  2. [Device design results] Device-design results (3 dB splitter, Bragg grating, selective mode reflector): the successful designs constitute existence proofs but supply no forward-model fidelity metrics (e.g., comparison of VIE-computed transmission or reflection spectra against an independent FD reference at the final optimized geometries), leaving open whether the VIE operator and its adjoint preserve the accuracy needed to support the efficiency claim.
minor comments (2)
  1. [Method] Notation for the VIE operator and its discretization should be introduced with explicit reference to the underlying integral kernel and material contrast handling to aid reproducibility.
  2. [Figures] Figure captions for the benchmark timing plots should state the mesh resolution, material indices, and error tolerance used for both VIE and FD runs.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the opportunity to respond to the referee's report on our manuscript. We address each of the major comments below and indicate the revisions we will make.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that 'comparative benchmarks demonstrate... multiple orders of magnitude improvement in computational efficiency' is not supported by any reported quantitative error metrics (e.g., L2 field errors, transmission discrepancies), mesh-convergence data, or explicit accuracy tolerances relative to the FD reference solvers. Without these, the runtime advantage cannot be assessed at equivalent fidelity for the target subwavelength, high-contrast devices.

    Authors: We appreciate the referee's emphasis on the need for quantitative accuracy metrics to support the efficiency claims. The original manuscript presents runtime benchmarks but does not include explicit error comparisons. In the revised manuscript, we will add L2 norm errors between VIE and FD field solutions, discrepancies in key performance metrics such as transmission, and mesh-convergence studies with specified accuracy tolerances. These additions will enable evaluation of the speedup at equivalent fidelity levels. revision: yes

  2. Referee: [Device design results] Device-design results (3 dB splitter, Bragg grating, selective mode reflector): the successful designs constitute existence proofs but supply no forward-model fidelity metrics (e.g., comparison of VIE-computed transmission or reflection spectra against an independent FD reference at the final optimized geometries), leaving open whether the VIE operator and its adjoint preserve the accuracy needed to support the efficiency claim.

    Authors: We agree that providing fidelity metrics for the optimized devices is important to validate the method. We will revise the manuscript to include comparisons of the transmission and reflection spectra computed using the VIE forward model against those from an independent FD solver for each of the three designed components. This will confirm the accuracy of the VIE-based optimization results. revision: yes

Circularity Check

0 steps flagged

No significant circularity in VIE adjoint derivation or benchmarks

full rationale

The paper presents a direct derivation of the adjoint method from the volume integral equation (VIE) formulation using standard adjoint calculus, as described in the abstract. No self-definitional steps, fitted parameters renamed as predictions, or load-bearing self-citations are evident in the provided text. The comparative efficiency benchmarks and successful designs of the 3 dB splitter, Bragg grating, and mode reflector serve as independent validation rather than reducing to inputs by construction. The derivation chain remains self-contained against external electromagnetic solvers, consistent with a normal non-circular finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the standard VIE integral operator for Maxwell's equations and the assumption that the adjoint of that operator can be formed without additional approximations that degrade accuracy for nanophotonic scales.

axioms (1)
  • domain assumption The volume integral equation operator accurately represents electromagnetic scattering inside subwavelength dielectric and metallic nanostructures.
    Invoked when the paper replaces FD solvers with VIE as the forward model.

pith-pipeline@v0.9.0 · 5739 in / 1153 out tokens · 46806 ms · 2026-05-18T14:31:33.360582+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    M.; Cheung, K

    Luan, E.; Shoman, H.; Ratner, D. M.; Cheung, K. C.; Chrostowski, L. Silicon photonic biosensors using label-free detection. Sensors 2018, 18, 3519

    work page 2018

  2. [2]

    2D Material-Based Optical Biosensor: Status and Prospect

    Lei, Z.; Guo, B. 2D Material-Based Optical Biosensor: Status and Prospect. Advanced Science 2021, 9, 2102924

    work page 2021

  3. [3]

    Integrated photonic quantum technologies

    Wang, J.; Sciarrino, F.; Laing, A.; Thompson, M. Integrated photonic quantum technologies. Nature Photonics 2021, 14, 273--284

    work page 2021

  4. [4]

    J.; Chen, H.-W.; Fang, A

    Heck, M. J.; Chen, H.-W.; Fang, A. W.; Koch, B. R.; Liang, D.; Park, H.; Sysak, M. N.; Bowers, J. E. Hybrid Silicon Photonics for Optical Interconnects. IEEE J. Sel. Top. Quantum Electron 2011, 17, 333--346

    work page 2011

  5. [5]

    P., N.and Ho; Xue, J.; Lim, L

    Li, C. P., N.and Ho; Xue, J.; Lim, L. W.; Chen, G.; Fu, Y. H.; Lee, L. Y. T. A progress review on solid‐state LiDAR and nanophotonics‐based LiDAR sensors. Laser and Photonics Reviews 2022, 16, 2100511

    work page 2022

  6. [6]

    Large-scale optical neural networks based on photoelectric multiplication

    Hamerly, R.; Bernstein, L.; Sludds, A.; Soljačić, M.; Englund, D. Large-scale optical neural networks based on photoelectric multiplication. Physical Review X 2019, 9, 021032

    work page 2019

  7. [7]

    J.; Tait, A

    Shastri, B. J.; Tait, A. N.; Ferreira de Lima, T.; Pernice, W. H.; Bhaskaran, H.; Wright; D., C.; Prucnal, P. R. Photonics for artificial intelligence and neuromorphic computing. Nature Photonics 2021, 15, 102--114

    work page 2021

  8. [8]

    Y.; Jin, W.; Vuckovic, J.; Rodriguez, A

    Molesky, S.; Lin, Z.; Piggott, A. Y.; Jin, W.; Vuckovic, J.; Rodriguez, A. W. Inverse design in nanophotonics. Nature Photonics 2018, 12, 659--670

    work page 2018

  9. [9]

    Niederberger, A. C. R.; Fattal, D. A.; Gauger, N. R.; Fan, S.; Beausoleil, R. G. Sensitivity analysis and optimization of sub-wavelength optical gratings using adjoints. Optics Express 2014, 22, 12971

    work page 2014

  10. [10]

    Y.; Lu, J.; Lagoudakis, K

    Piggott, A. Y.; Lu, J.; Lagoudakis, K. G.; Petykiewicz, J.; Babinec, T.; Vucǩovic, J. Inversedesignanddemonstrationofacompact and broadband on-chip wavelength demultiplexer. Nature Photonics 2015, 9, 374--377

    work page 2015

  11. [11]

    Fabrication- constrained nanophotonic inverse design

    Piggott, A.; Petykiewicz, J.; Su, L.; Vucǩovic, J. Fabrication- constrained nanophotonic inverse design. Scientific Reports 2017, 7, 1--7

    work page 2017

  12. [12]

    Sideris, C.; Garza, E.; Bruno, O. P. Ultrafast simulation and optimization of nanophotonic devices with integral equation methods. ACS Photonics 2019, 6, 3233--3240

    work page 2019

  13. [13]

    D.; Bruno, O

    Sideris, C.; Khachaturian, A.; White, A. D.; Bruno, O. P.; Hajimiri, A. Foundry-fabricated grating coupler demultiplexer inverse- designed via fast integral methods. Communications Physics 2022, 5, 1--8

    work page 2022

  14. [14]

    Fast Inverse Design of 3D Nanophotonic Devices Using Boundary Integral Methods

    Garza, E.; Sideris, C. Fast Inverse Design of 3D Nanophotonic Devices Using Boundary Integral Methods. ACS Photonics 2022, 10, 824--835

    work page 2022

  15. [15]

    P.; Ba, J

    Kingma, D. P.; Ba, J. Adam: A method for stochastic optimization. arXiv 2014,

    work page 2014

  16. [16]

    Particle swarm optimization (PSO)

    Marini, F.; Walczak, B. Particle swarm optimization (PSO). A tutorial. Chemometrics and Intelligent Laboratory Systems 2015, 149, 153--165

    work page 2015

  17. [17]

    Accelerated distributed Nesterov gradient descent

    Qu, G.; Li, N. Accelerated distributed Nesterov gradient descent. IEEE Transactions on Automatic Control 2019, 65, 2566--2581

    work page 2019

  18. [18]

    H.; Lu, P.; Nocedal, J

    Zhu, C.; Byrd, R. H.; Lu, P.; Nocedal, J. Algorithm 778: L-BFGS-B: Fortran subroutines for large-scale bound-constrained optimization. ACM Transactions on mathematical software (TOMS) 1997, 23, 550--560

    work page 1997

  19. [19]

    I.; Sertel, K.; Volakis, J

    Sancer, M. I.; Sertel, K.; Volakis, J. L.; Van Alstine, P. On volume integral equations. IEEE transactions on antennas and propagation 2006, 54, 1488--1495

    work page 2006

  20. [20]

    S.; Meincke, P.; Breinbjerg, O

    Kim, O. S.; Meincke, P.; Breinbjerg, O. Method of moments solution of volume integral equations using higher-order hierarchical Legendre basis functions. Radio Science 2004, 39, 1--7

    work page 2004

  21. [21]

    Discretization of volume integral equation formulations for extremely anisotropic materials

    Markkanen, J.; Yla-Oijala, P.; Sihvola, A. Discretization of volume integral equation formulations for extremely anisotropic materials. IEEE Transactions on Antennas and Propagation 2012, 60, 5195--5202

    work page 2012

  22. [22]

    C.; Cao, X.; Yla-Oijala, P

    Markkanen, J.; Lu, C. C.; Cao, X.; Yla-Oijala, P. Analysis of volume integral equation formulations for scattering by high-contrast penetrable objects. IEEE Transactions on Antennas and Propagation 2012, 60, 2367--2374

    work page 2012

  23. [23]

    Stable FFT-JVIE solvers for fast analysis of highly inhomogeneous dielectric objects

    Polimeridis, A.; Villena, J.; Daniel, L.; White, J. Stable FFT-JVIE solvers for fast analysis of highly inhomogeneous dielectric objects. Journal of Computational Physics 2014, 269, 280--296

    work page 2014

  24. [24]

    W.; Nachtigal, N

    Freund, R. W.; Nachtigal, N. M. QMR: a quasi-minimal residual method for non-Hermitian linear systems. Numerische mathematik 1991, 60, 315--339

    work page 1991

  25. [25]

    Nazareth, J. L. Conjugate gradient method. Wiley Interdisciplinary Reviews: Computational Statistics 2009, 1, 348--353

    work page 2009

  26. [26]

    Saad, Y.; Schultz, M. H. GMRES: A generalized minimal residual algorithm for solving nonsymmetric linear systems. SIAM Journal on scientific and statistical computing 1986, 7, 856--869

    work page 1986

  27. [27]

    P.; Polimeridis, A

    Groth, S. P.; Polimeridis, A. G.; Tambova, A.; White, J. K. Circulant preconditioning in the volume integral equation method for silicon photonics. Journal of the Optical Society of America A 2019, 36, 1079--1088

    work page 2019

  28. [28]

    P.; Giannakopoulos, I

    Georgakis, I. P.; Giannakopoulos, I. I.; Litsarev, M. S.; Polimeridis, A. G. A Fast Volume Integral Equation Solver With Linear Basis Functions for the Accurate Computation of EM Fields in MRI. IEEE Transactions on Antennas and Propagation 2020, 69, 4020--4032

    work page 2020

  29. [29]

    P.; Polimeridis, A

    Groth, S. P.; Polimeridis, A. G.; Tambova, A.; White, J. K. Adiabatic Absorbers in Photonics Simulations with the Volume Integral Equation Method. Journal of Lightwave Technology 2018, 36, 3765 -- 3777

    work page 2018

  30. [30]

    V.; A., P

    Su, L.; Vercruysse, D.; Skarda, J.; Sapra, N. V.; A., P. J.; J., V. SPINS: Nanophotonic Inverse Design Software. 2022; https://github.com/stanfordnqp/spins-b, Accessed: 2024-02-14

    work page 2022

  31. [31]

    2024; https://www.ansys.com/products/optics/fdtd, Accessed: 2024-02-14

    Ansys Lumerical: Photonic Design Software. 2024; https://www.ansys.com/products/optics/fdtd, Accessed: 2024-02-14

    work page 2024

  32. [32]

    Constraining continuous topology optimizations to discrete solutions for photonic applications

    Ballew, C.; Roberts, G.; Zheng, T.; Faraon, A. Constraining continuous topology optimizations to discrete solutions for photonic applications. ACS photonics 2023, 10, 836--844

    work page 2023

  33. [33]

    J.; Timurdogan, E.; Wright, J

    Biberman, A.; Shaw, M. J.; Timurdogan, E.; Wright, J. B.; Watts, M. R. Ultralow-loss silicon ring resonators. Optics Letters 2012, 37, 4236--4238

    work page 2012

  34. [34]

    A gradient-oriented binary search method for photonic device design

    Chen, H.; Jia, H.; Wang, T.; Yang, J. A gradient-oriented binary search method for photonic device design. Journal of Lightwave Technology 2021, 39, 2407--2412

    work page 2021

  35. [35]

    continuous phase

    Paul, J.; Sideris, C. Accelerated 3D Maxwell Integral Equation Solver Using the Interpolated Factored Green Function Method. IEEE Transactions on Antennas and Propagation 2025, 73, 3814 -- 3826 mcitethebibliography main.tex0000664000000000000000000015115715065134344011243 0ustar rootroot [journal=jacsat,manuscript=article] achemso chemformula [T1] fontenc...

    work page 2025

  36. [36]

    G.; Villena, J

    Polimeridis, A. G.; Villena, J. F.; Daniel, L.; White, J. K. Stable FFT-JVIE Solvers for Fast Analysis of Highly Inhomogeneous Dielectric Objects. J. Comput. Phys. 2014, 269, 280--296

    work page 2014

  37. [37]

    P.; Polimeridis, A

    Groth, S. P.; Polimeridis, A. G.; Tambova, A.; White, J. K. Adiabatic Absorbers in Photonics Simulations with the Volume Integral Equation Method. J. Lightwave Technol. 2018, 36, 3765--3777

    work page 2018

  38. [38]

    Theory of Complex Functions; Springer-Verlag: Berlin, 1991

    Remmert, R. Theory of Complex Functions; Springer-Verlag: Berlin, 1991

    work page 1991