pith. sign in

arxiv: 2605.04197 · v2 · pith:JYJKTQM4new · submitted 2026-05-05 · 🧮 math.DS · cs.NA· math.NA

Calculating Domain of Attraction Boundary of Power Systems Based on the Gentlest Ascent Dynamics

Sixu Wu , Chenmin Zhang , Aiqing Zhu , Yang Liu , Jianxi Lin , Yifa Tang This is my paper

Pith reviewed 2026-05-20 23:17 UTC · model grok-4.3

classification 🧮 math.DS cs.NAmath.NA
keywords domain of attractionpower systemstransient stabilitygentlest ascent dynamicsstable manifoldssaddle pointsperiodic orbits
0
0 comments X

The pith

The domain of attraction boundary in power systems equals the closure of stable manifolds of index-1 critical elements.

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

This paper develops methods to calculate the boundary of the domain of attraction for post-disturbance stable equilibria in power systems. It applies gentlest ascent dynamics to locate 1-saddle points, adjoint operators for periodic orbits, and stable manifold algorithms to trace the boundary. The work transforms the stability identification problem into constructing these unstable features and their attracting sets. Under certain assumptions the authors prove that the boundary is precisely the closure of the union of those stable manifolds and they establish stability results for a perturbed version of the gentlest ascent system. Experiments on two-machine and three-machine models confirm that the computed structures accurately reproduce the geometric boundary in cases with saddles only or with periodic orbits present.

Core claim

The authors prove that the DOA boundary is the closure of the union of stable manifolds of index-1 critical elements and establish a stability theory for a perturbed GAD system. They employ GAD for 1-saddle points, adjoint operators for periodic orbits, and stable manifold algorithms to compute the boundary in benchmark power system models.

What carries the argument

Gentlest ascent dynamics for locating index-1 saddle points together with stable manifold construction algorithms and adjoint methods for periodic orbits.

If this is right

  • DOA boundaries become computable by locating index-1 critical elements and building their stable manifolds.
  • The method handles both saddle-point-only cases and cases containing periodic orbits.
  • Transient stability after faults can be assessed by examining whether post-disturbance states lie inside the constructed region.
  • Numerical tests on small generator models show the algorithms reproduce the expected boundary geometry.

Where Pith is reading between the lines

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

  • The same manifold-tracing approach could apply to stability questions in other classes of nonlinear oscillators and networks.
  • Stability results for the perturbed GAD system may improve understanding of numerical continuation methods in related dynamical problems.
  • Scaling the algorithms to higher-dimensional models could support stability screening in grids that include inverter-based resources.

Load-bearing premise

Certain assumptions under which the domain of attraction boundary equals the closure of the union of stable manifolds of index-1 critical elements.

What would settle it

A concrete power system trajectory that starts inside the computed manifold boundary yet diverges from the stable equilibrium, or one that starts outside yet converges.

Figures

Figures reproduced from arXiv: 2605.04197 by Aiqing Zhu, Chenmin Zhang, Jianxi Lin, Sixu Wu, Yang Liu, Yifa Tang.

Figure 1
Figure 1. Figure 1: Two-machine system. -4 -3 -2 -1 0 1 2 3 4 1 (rad) -3 -2 -1 0 1 2 3 2 (rad) Asymptotically Stable Equilibrium 1-Saddles Adjoint Unstable Directions (a) 1-saddles and their associated unstable eigen-directions. -6 -4 -2 0 2 4 6 1 (rad) -5 -4 -3 -2 -1 0 1 2 3 4 2 (rad) (b) Boundary of the region of attraction. associated unstable eigen-directions, and then compute the boundary of the region of attraction. The… view at source ↗
Figure 1
Figure 1. Figure 1: Two-machine system. -4 -3 -2 -1 0 1 2 3 4 1 (rad) -3 -2 -1 0 1 2 3 2 (rad) Asymptotically Stable Equilibrium 1-Saddles Adjoint Unstable Directions (a) 1-saddles and their associated unstable eigen-directions. -6 -4 -2 0 2 4 6 1 (rad) -5 -4 -3 -2 -1 0 1 2 3 4 2 (rad) (b) Boundary of the region of attraction. 5.2 Three-Machine System First, we give the parameters for n = 3: H = (6.5, 6.5, 6.5), D = (0.1, 0.1… view at source ↗
Figure 2
Figure 2. Figure 2: Three-machine system. 5 -2 0 3 (rad) 5 2 2 (rad) 0 1 (rad) 0 -5 -5 Asymptotically Stable Equilibrium 1-Saddles Adjoint Unstable Directions (a) 1-saddles and their associated unstable eigen-directions. -6 -4 5 -2 0 5 3 (rad) 2 2 (rad) 0 4 1 (rad) 6 0 -5 -5 (b) Boundary of the region of attraction. 24 view at source ↗
Figure 2
Figure 2. Figure 2: Three-machine system. 5 -2 0 3 (rad) 5 2 2 (rad) 0 1 (rad) 0 -5 -5 Asymptotically Stable Equilibrium 1-Saddles Adjoint Unstable Directions (a) 1-saddles and their associated unstable eigen-directions. -6 -4 5 -2 0 5 3 (rad) 2 2 (rad) 0 4 1 (rad) 6 0 -5 -5 (b) Boundary of the region of attraction. 5.3 Three-Machine System with a Periodic Orbit First, we give the parameters for n = 3: H = (6.5, 6.5, 6.175), … view at source ↗
Figure 3
Figure 3. Figure 3: 1-saddle and its stable manifold for the three-machine system with a periodic view at source ↗
Figure 3
Figure 3. Figure 3: 1-saddle and its stable manifold for the three-machine system with a periodic [PITH_FULL_IMAGE:figures/full_fig_p028_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Periodic orbit and its associated unstable eigen-direction for the three view at source ↗
Figure 4
Figure 4. Figure 4: Periodic orbit and its associated unstable eigen-direction for the three [PITH_FULL_IMAGE:figures/full_fig_p028_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Stable manifolds of the periodic orbits for the three-machine system with a view at source ↗
Figure 5
Figure 5. Figure 5: Stable manifolds of the periodic orbits for the three-machine system with a [PITH_FULL_IMAGE:figures/full_fig_p029_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Boundary of the region of attraction for the three-machine system with a view at source ↗
Figure 6
Figure 6. Figure 6: Boundary of the region of attraction for the three-machine system with a [PITH_FULL_IMAGE:figures/full_fig_p029_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Level set of the boundary of the region of attraction for the two-machine view at source ↗
Figure 7
Figure 7. Figure 7: Level set of the boundary of the region of attraction for the two-machine [PITH_FULL_IMAGE:figures/full_fig_p033_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Level set of the boundary of the region of attraction for the three-machine view at source ↗
Figure 8
Figure 8. Figure 8: Level set of the boundary of the region of attraction for the three-machine [PITH_FULL_IMAGE:figures/full_fig_p033_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Level set of the boundary of the region of attraction for the three-machine view at source ↗
Figure 9
Figure 9. Figure 9: Level set of the boundary of the region of attraction for the three-machine [PITH_FULL_IMAGE:figures/full_fig_p034_9.png] view at source ↗
read the original abstract

The power system, a fundamental public utility, is increasingly important due to growing global electricity demand. Recent large-scale blackouts (e.g., Iberian Peninsula, UK) have raised concerns about transient stability under impact faults. Transient stability is determined by post-disturbance synchronizing capability of synchronous generators, formulated as identifying the domain of attraction (DOA) boundary of the asymptotically stable equilibrium. Using a benchmark model of synchronous-generator-dominated power systems, this report employs a gentlest ascent dynamics (GAD) method for 1-saddle points, an adjoint operator method for periodic orbits, and stable manifold algorithms to compute the DOA boundary. These algorithms transform DOA boundary determination into constructing unstable critical elements (saddle points and periodic orbits) and their stable manifolds. Theoretically, under certain assumptions we prove that the DOA boundary is the closure of the union of stable manifolds of index-1 critical elements, and establish a stability theory for a perturbed GAD system. Numerical experiments on two-machine and three-machine systems (with only saddle points or with periodic orbits) validate the effectiveness and accuracy. Results show the algorithms accurately capture the geometric structure of the DOA boundary, providing a new numerical tool for transient stability analysis.

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 / 1 minor

Summary. The paper proposes using the gentlest ascent dynamics (GAD) method to locate index-1 saddle points, an adjoint operator approach for periodic orbits, and stable manifold algorithms to compute the domain of attraction (DOA) boundary for the post-fault stable equilibrium in synchronous-generator-dominated power system models. It claims a theoretical result that, under certain assumptions, the DOA boundary equals the closure of the union of stable manifolds of index-1 critical elements, along with a stability theory for a perturbed GAD system, and reports numerical validation on two- and three-machine benchmark systems (with and without periodic orbits).

Significance. If the unspecified assumptions hold and the numerical accuracy is confirmed, the work supplies a new computational framework for transient stability analysis in power systems by reducing DOA boundary determination to the construction of unstable critical elements and their stable manifolds. This could be useful for large-scale blackout prevention, especially if the GAD-based search scales better than existing methods for locating saddles and periodic orbits in high-dimensional swing equations.

major comments (2)
  1. [Abstract and theoretical development] Abstract and theoretical development: The central claim that the DOA boundary is the closure of the union of stable manifolds of index-1 critical elements (saddles and periodic orbits) is proved only under 'certain assumptions.' These assumptions are not enumerated, so it is impossible to verify whether they hold for the two-machine and three-machine models, particularly the case containing periodic orbits. This is load-bearing because the justification for using GAD to locate the critical elements and then compute their stable manifolds rests entirely on this reduction.
  2. [Numerical experiments section] Numerical experiments section: The reported validation on two- and three-machine systems asserts that the algorithms 'accurately capture the geometric structure' of the DOA boundary, yet no quantitative error metrics, baseline comparisons against other DOA estimation techniques, or details on post-disturbance trajectory generation are provided. Without these, the empirical support for accuracy and practical utility remains incomplete.
minor comments (1)
  1. [Abstract] The abstract refers to 'a benchmark model of synchronous-generator-dominated power systems' without giving the explicit differential equations or parameter values; these should be stated early so readers can reproduce the critical-element computations.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the insightful comments on our manuscript. We address the major comments point by point below, indicating where revisions will be made to strengthen the paper.

read point-by-point responses

  1. Referee: [Abstract and theoretical development] The central claim that the DOA boundary is the closure of the union of stable manifolds of index-1 critical elements (saddles and periodic orbits) is proved only under 'certain assumptions.' These assumptions are not enumerated, so it is impossible to verify whether they hold for the two-machine and three-machine models, particularly the case containing periodic orbits. This is load-bearing because the justification for using GAD to locate the critical elements and then compute their stable manifolds rests entirely on this reduction.

    Authors: We agree that explicitly listing the assumptions would improve clarity and allow readers to verify their applicability. The proof in the manuscript is based on assumptions including the hyperbolicity of the index-1 critical elements (saddles and periodic orbits) and the transversality of their stable manifolds to the DOA boundary. We will revise the manuscript by adding a subsection that enumerates these assumptions clearly and discusses their validity for the benchmark systems, including the three-machine system with periodic orbits. revision: yes

  2. Referee: [Numerical experiments section] The reported validation on two- and three-machine systems asserts that the algorithms 'accurately capture the geometric structure' of the DOA boundary, yet no quantitative error metrics, baseline comparisons against other DOA estimation techniques, or details on post-disturbance trajectory generation are provided. Without these, the empirical support for accuracy and practical utility remains incomplete.

    Authors: We acknowledge the need for more rigorous quantitative validation. In the revised manuscript, we will include quantitative error metrics, such as the maximum deviation of the computed boundary from reference trajectories obtained via numerical integration of the system dynamics. We will also add comparisons with baseline methods like the energy function approach or closest UEP method where applicable, and provide details on the post-disturbance trajectory generation process used for validation. This will better demonstrate the accuracy and utility of the proposed framework. revision: yes

Circularity Check

0 steps flagged

No significant circularity; theoretical claim applies standard dynamical systems results to power-system model

full rationale

The paper's strongest claim is a proof, under unspecified assumptions, that the DOA boundary equals the closure of the union of stable manifolds of index-1 critical elements, together with a stability theory for a perturbed GAD system. No equation or derivation in the abstract or described content reduces this result to a fitted quantity, self-citation chain, or renamed input; the GAD and adjoint-operator constructions are presented as independent numerical tools that locate the critical elements whose manifolds are then computed. The work therefore remains self-contained against external dynamical-systems benchmarks, with the 'certain assumptions' serving as a standard caveat rather than a hidden self-definition.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based solely on the abstract; full text would be required to enumerate all modeling assumptions and numerical parameters. The central theoretical statement rests on unspecified 'certain assumptions' about the power system vector field and critical elements.

axioms (1)
  • domain assumption Under certain assumptions the DOA boundary is the closure of the union of stable manifolds of index-1 critical elements
    Stated directly in the abstract as the theoretical foundation.

pith-pipeline@v0.9.0 · 5765 in / 1136 out tokens · 43947 ms · 2026-05-20T23:17:25.270001+00:00 · methodology

Review history (2 revisions) →

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

51 extracted references · 51 canonical work pages

  1. [1]

    Perko.Differential Equations and Dynamical Systems, volume 7 ofTexts in Applied Mathematics

    L. Perko.Differential Equations and Dynamical Systems, volume 7 ofTexts in Applied Mathematics. Springer, third edition, 2000. doi:10.1007/978-1-4613-0003-8

    work page doi:10.1007/978-1-4613-0003-8 2000

  2. [2]

    Kundur.Power System Stability and Control

    P. Kundur.Power System Stability and Control. McGraw-hill, New York, 1994. ISBN 0-07-035958-X

    work page 1994

  3. [3]

    Genesio, M

    R. Genesio, M. Tartaglia, and A. Vicino. On the estimation of asymptotic stability regions: State of the art and new proposals.IEEE Transactions on Automatic Control, 30(8):747–755, 1985

    work page 1985

  4. [4]

    H. D. Chiang, M. W. Hirsch, and F. F. Wu. Stability regions of nonlinear au- tonomous dynamical systems.IEEE Transactions on Automatic Control, 33(1): 16–27, January 1988. ISSN 1558-2523. doi:10.1109/9.357

    work page doi:10.1109/9.357 1988

  5. [5]

    H. D. Chiang.Direct Methods for Stability Analysis of Electric Power Systems. Wiley, 2011. ISBN 978-0-470-48440-1. doi:10.1002/9780470872130

    work page doi:10.1002/9780470872130 2011

  6. [6]

    B. Stott. Power system dynamic response calculations.Proceedings of the IEEE, 67(2):219–241, 1979

    work page 1979

  7. [7]

    M. K. Bakhshizadeh, S. Ghosh, G. Y. Yang, and L. Kocewika. Transient stability analysis of grid-connected converters in wind turbine systems based on linear lya- punov function and reverse-time trajectory.Journal of Modern Power Systems and Clean Energy, 12(3):782–790, 2024

    work page 2024

  8. [8]

    Hurayb, D

    K. Hurayb, D. Niebur, and M. Kabalan. Large-signal stability analysis of grid- connected droop-controlled inverter with local rl loads. In2021 North American Power Symposium (NAPS), pages 1–6, 2021

    work page 2021

  9. [9]

    Kabalan, P

    M. Kabalan, P. Singh, and D. Niebur. Nonlinear lyapunov stability analysis of seven models of a dc/ac droop controlled inverter connected to an infinite bus. IEEE Transactions on Smart Grid, 10(1):772–781, 2019

    work page 2019

  10. [10]

    A. A. Fouad and V. Vittal.Power System Transient Stability Analysis Using the Transient Energy Function Method. Prentice Hall, 1991

    work page 1991

  11. [11]

    M. A. Pai.Energy Function Analysis for Power System Stability. Springer, 1989. 35

    work page 1989

  12. [12]

    C. C. Chu and H. D. Chiang. Constructing analytical energy functions for loss- less network-reduction power system models: Framework and new developments. Circuits, Systems and Signal Processing, 18(1):1–16, 1999

    work page 1999

  13. [13]

    C. C. Chu and H. D. Chiang. Constructing analytical energy functions for network- preserving power system models.Circuits, Systems and Signal Processing, 24(4): 363–383, 2005

    work page 2005

  14. [14]

    Athay, R

    T. Athay, R. Podmore, and S. Virmani. A practical method for the direct analysis of transient stability.IEEE Transactions on Power Apparatus and Systems, PAS-98 (2):573–584, 1979

    work page 1979

  15. [15]

    M. A. Pai and S. Varwandkar. On the inclusion of transfer conductances in lya- punov functions for multimachine power systems.IEEE Transactions on Automatic Control, 22(6):983–985, 1977

    work page 1977

  16. [16]

    Uemura, J

    K. Uemura, J. Matsuki, I. Yamada, and T. Tsuji. Approximation of an energy function in transient stability analysis of power systems.Electrical Engineering in Japan, 92(6):96–100, 1972

    work page 1972

  17. [17]

    A. H. El-Abiad and K. Nagappan. Transient stability regions of multimachine power systems.IEEE Transactions on Power Apparatus and Systems, PAS-85(2):169–179, 1966

    work page 1966

  18. [18]

    H. D. Chiang and J. S. Thorp. The closest unstable equilibrium point method for power system dynamic security assessment.IEEE Transactions on Circuits and Systems, 36(9):1187–1200, 1989

    work page 1989

  19. [19]

    H. D. Chiang, F. Wu, and P. Varaiya. Foundations of direct methods for power system transient stability analysis.IEEE Transactions on Circuits and Systems, 34 (2):160–173, 1987

    work page 1987

  20. [20]

    H. D. Chiang, F. F. Wu, and P. P. Varaiya. A bcu method for direct analysis of power system transient stability.IEEE Transactions on Power Systems, 9(3): 1194–1208, 1994

    work page 1994

  21. [21]

    Kakimoto, Y

    N. Kakimoto, Y. Ohnogi, H. Matsuda, and H. Shibuya. Transient stability analysis of large-scale power system by lyapunov’s direct method.IEEE Transactions on Power Apparatus and Systems, PAS-103(1):160–167, 1984

    work page 1984

  22. [22]

    H. D. Chiang, F. F. Wu, and P. P. Varaiya. Foundations of the potential en- ergy boundary surface method for power system transient stability analysis.IEEE Transactions on Circuits and Systems, 35(6):712–728, 1988

    work page 1988

  23. [23]

    P. A. Parrilo.Structured Semidefinite Programs and Semialgebraic Geometry Meth- ods in Robustness and Optimization. PhD thesis, California Institute of Technology, 2000. 36

    work page 2000

  24. [24]

    Papachristodoulou and S

    A. Papachristodoulou and S. Prajna. A tutorial on sum of squares techniques for systems analysis. InProceedings of the 2005 American Control Conference, pages 2686–2700, 2005

    work page 2005

  25. [25]

    Anghel, F

    M. Anghel, F. Milano, and A. Papachristodoulou. Algorithmic construction of lya- punov functions for power system stability analysis.IEEE Transactions on Circuits and Systems I: Regular Papers, 60(9):2533–2546, September 2013. ISSN 1558-0806. doi:10.1109/tcsi.2013.2246233

    work page doi:10.1109/tcsi.2013.2246233 2013

  26. [26]

    Izumi, H

    S. Izumi, H. Somekawa, X. Xin, and T. Yamasaki. Estimation of regions of attrac- tion of power systems by using sum of squares programming.Electrical Engineering, 100(4):2205–2216, May 2018. ISSN 1432-0487. doi:10.1007/s00202-018-0690-z

    work page doi:10.1007/s00202-018-0690-z 2018

  27. [27]

    Y. Liu, Z. Chen, H. Yao, L. Yi, and Q. H. Wu. Estimating critical clearing time of grid faults using da of state-reduction model of power sys- tems.CSEE Journal of Power and Energy Systems, 10(2):807–820, April 2024. doi:10.17775/CSEEJPES.2022.07170

    work page doi:10.17775/cseejpes.2022.07170 2024

  28. [28]

    Chesi.Domain of Attraction: Analysis and Control via SOS Programming

    G. Chesi.Domain of Attraction: Analysis and Control via SOS Programming. Springer, 2011

    work page 2011

  29. [29]

    Takagi and M

    T. Takagi and M. Sugeno. Fuzzy identification of systems and its applications to modeling and control.IEEE Transactions on Systems, Man, and Cybernetics, SMC-15(1):116–132, 1985

    work page 1985

  30. [30]

    T. M. Guerra and L. Vermeiren. Lmi-based relaxed nonquadratic stabilization conditions for nonlinear systems in the takagi-sugeno’s form.Automatica, 40(5): 823–829, 2004

    work page 2004

  31. [31]

    X. L. Li, Z. Wang, Y. X. Liu, Z. G. Wang, L. Zhu, L. Guo, C. Zhang, J. B. Zhu, and C. S. Wang. The largest estimated domain of attraction and its applications for transient stability analysis of pll synchronization in weak-grid-connected vscs. IEEE Transactions on Power Systems, 38(5):4107–4121, 2023

    work page 2023

  32. [32]

    W. Z. Tang, B. Li, S. C. Hou, X. Q. Shao, and H. J. Yu. Transient stability analysis for grid-connected renewable power generation systems based on lmi optimization modelling.Electronics, 13(24):5052, 2024

    work page 2024

  33. [33]

    Y. W. Ding, F. Gao, and M. M. Khan. Transient stability analysis of microgrid con- sidering impact of grid-following converter’s current controller.IEEE Transactions on Power Electronics, 39(8):9100–9105, 2024

    work page 2024

  34. [34]

    Y. C. Chang, N. Roohi, and S. C. Gao. Neural lyapunov control. InProceedings of the 33rd International Conference on Neural Information Processing Systems, pages 3245–3254, 2019. 37

    work page 2019

  35. [35]

    Huang, S

    T. Huang, S. C. Gao, and L. Xie. A neural lyapunov approach to transient stability assessment of power electronics-interfaced networked microgrids.IEEE Transac- tions on Smart Grid, 13(1):106–118, 2022

    work page 2022

  36. [36]

    T. Q. Zhao, J. H. Wang, X. N. Lu, and Y. H. Du. Neural lyapunov control for power system transient stability: a deep learning-based approach.IEEE Transactions on Power Systems, 37(2):955–966, 2022

    work page 2022

  37. [37]

    Y. F. Liu, L. J. Zhou, S. Chen, J. R. Zhang, and Y. Wang. An improved neural lyapunov method for transient stability assessment of networked microgrids.IEEE Transactions on Smart Grid, 15(2):1410–1422, 2024

    work page 2024

  38. [38]

    S. T. Zhang, Z. X. Zhu, and Y. Li. A critical review of data-driven transient stability assessment of power systems: principles, prospects and challenges.Energies, 14(21): 7238, 2021

    work page 2021

  39. [39]

    stability regions of nonlinear autonomous dynamical systems

    M. W. Fisher and I. A. Hiskens. Comments on “stability regions of nonlinear autonomous dynamical systems”.IEEE Transactions on Automatic Control, 66 (12):6194–6196, December 2021. ISSN 2334-3303. doi:10.1109/tac.2021.3061674

    work page doi:10.1109/tac.2021.3061674 2021

  40. [40]

    M. W. Fisher and I. A. Hiskens. Hausdorff continuity of region of attraction bound- ary under parameter variation with application to disturbance recovery.SIAM Jour- nal on Applied Dynamical Systems, 21(1):327–365, 2022. doi:10.1137/20M1371944

    work page doi:10.1137/20m1371944 2022

  41. [41]

    W. N. E and X. Zhou. The gentlest ascent dynamics.Nonlinearity, 24(6):1831, May 2011. doi:10.1088/0951-7715/24/6/008

    work page doi:10.1088/0951-7715/24/6/008 2011

  42. [42]

    H. D. Chiang and L. F. C. Alberto.Stability Regions of Nonlinear Dynamical Systems: Theory, Estimation, and Applications. Cambridge, 2015. ISBN 978-1- 107-03540-9. doi:10.1017/CBO9781139548861

    work page doi:10.1017/cbo9781139548861 2015

  43. [43]

    Azimi, O

    S. Azimi, O. Ashtari, and T. M. Schneider. Constructing periodic orbits of high- dimensional chaotic systems by an adjoint-based variational method.Phys. Rev. E, 105:014217, Jan 2022. doi:10.1103/PhysRevE.105.014217

    work page doi:10.1103/physreve.105.014217 2022

  44. [44]

    Y. F. Tang, R. L. Zhang, B. B. Zhu, A. Q. Zhu, and S. X. Wu.Geometric Al- gorithms for Dynamical Systems with Applications (in Chinese), volume 215 of Modern Mathematical Basic Series. Science Press, 2026. ISBN 9787030846334

    work page 2026

  45. [45]

    J. M. Lee.Introduction to Smooth Manifolds, volume 218 ofGraduate Texts in Math- ematics. Springer, second edition, 2013. ISBN 978-1-4419-9981-8. doi:10.1007/978- 1-4419-9982-5

    work page doi:10.1007/978- 2013

  46. [46]

    S. T. Gu and X. P. Zhou. Simplified gentlest ascent dynamics for saddle points in non-gradient systems.Chaos: An Interdisciplinary Journal of Nonlinear Science, 28(12):123106, December 2018. doi:10.1063/1.5046819. 38

    work page doi:10.1063/1.5046819 2018

  47. [47]

    doi: 10.1109/TPAMI.2013.50

    Yoshua Bengio, Aaron Courville, and Pascal Vincent. Representation learning: a review and new perspectives. 35(8):1798–1828. doi:10.1109/TPAMI.2013.50

    work page doi:10.1109/tpami.2013.50 2013

  48. [48]

    Gradient-based learning applied to document recognition,

    Yann LeCun, L´ eon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient-based learning applied to document recognition. 86(11):2278–2324. doi:10.1109/5.726791

    work page doi:10.1109/5.726791

  49. [49]

    Dahl, Abdel-rahman Mohamed, Navdeep Jaitly, Andrew Senior, Vincent Vanhoucke, Patrick Nguyen, Tara N

    Geoffrey Hinton, Li Deng, Dong Yu, George E. Dahl, Abdel-rahman Mohamed, Navdeep Jaitly, Andrew Senior, Vincent Vanhoucke, Patrick Nguyen, Tara N. Sainath, and Brian Kingsbury. Deep neural networks for acoustic modeling in speech recognition. 29(6):82–97. doi:10.1109/MSP.2012.2205597

    work page doi:10.1109/msp.2012.2205597 2012

  50. [50]

    Dropout: a simple way to prevent neural networks from overfitting

    Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdinov. Dropout: a simple way to prevent neural networks from overfitting. 15(1):1929–1958

    work page 1929

  51. [51]

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural networks. 25:1090–1098. 39

    work page