Pith. sign in

REVIEW 3 major objections 4 minor 48 references

Learning with Mandelbrot and Julia

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Using only the first four iterates of the quadratic orbit, supervised classifiers label Mandelbrot and Julia set membership at roughly 95% and 88% accuracy, outperforming the classical |z_n|>2 threshold rule; the paper conjectures a measura

desk verdict An honestly executed ML benchmark on fractal orbits whose central interpretation is undermined by using a 100-iteration escape proxy as ground truth. read the letter →

arxiv 2509.00903 v1 pith:4ECC5NUY submitted 2025-08-31 nlin.CD cs.LG

V.R. Tjahjono , S.F. Feng , E.R.M. Putri , H. Susanto This is my paper
classification nlin.CDcs.LG MSC 37F3537F4568T0568T0765C60 PACS 05.45.Df07.05.Mh02.10.Ox02.30.Sa02.60.-x
keywords MandelbrotsetsJuliaMachinelearningFractalsSupervisedclassificationOrbititeratesBoundednessDynamicalsystems
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Four early iterates of the quadratic orbit, not the full asymptotic trajectory, are enough for standard supervised classifiers to label points in the Mandelbrot and Julia sets as bounded or unbounded, reaching about 95% and 88% accuracy and beating the classical |z_n|>2 escape test. The paper frames this as evidence that the first few iterates carry geometric information about an asymptotic property, and states a conjecture: near the boundary, a measurable function of finitely many early orbit magnitudes can decide membership with error tending to zero as the number of iterates grows. If true, fractal classification can be made dramatically cheaper, and machine learning becomes a tool for generating conjectures in pure mathematics rather than only an approximation scheme.

What carries the argument

The load-bearing object is the orbit of the quadratic map z_{n+1}=z_n^2+c, truncated to its first four nonzero terms. Each training point is the vector of real and imaginary parts of those terms, and the label is 'unbounded' if any of the first 100 iterates exceeds the escape radius r_w=2, otherwise 'bounded'; this same 100-step rule is the threshold baseline the ML models must beat. The classifiers—especially KNN and RF—carry the argument: their strong performance with short inputs, plus feature-importance analysis showing later iterates add signal, supports the conjecture that a measurable early-iterate function exists.

What would settle it

Take the same training setup but evaluate on a fresh test set labeled by iterating until |z_n|>2 or n=10^6, or by a certified boundary oracle. If KNN/RF accuracy against these long-horizon labels falls to the ~82.5% majority baseline in the boundary region while staying near 95% on the 100-iteration labels, the reported result is an artifact of the finite-time proxy; conversely, sustained high accuracy would support Hypothesis 1.

Watch

Extended reading notes

Core claim

The authors' central claim is that asymptotic boundedness of quadratic orbits—the property defining the Mandelbrot and Julia sets—can be approximated well from the first few iterates of the orbit. Across seven models (CART, KNN, MLP, LSTM, BiLSTM, RF, CNN), training on the first four iterates gives roughly 95% accuracy on the Mandelbrot domain, 88–90% on the Julia set, and 91–94% near the Mandelbrot boundary, while the classical thresholding rule |z_n|>2 stays at 79–80% on the full domains and near 17% at the boundary. The paper states this as Hypothesis 1: for any epsilon greater than zero there is a compact neighborhood of the boundary, an integer I, and a measurable function of the first

Load-bearing premise

The load-bearing premise is that labels from the 100-iteration rule 'escape if any first-100 iterate exceeds magnitude 2' match true asymptotic boundedness; near the boundary this proxy can mislabel slow-escaping points, and if it is unfaithful the accuracies measure agreement with a finite-time approximation, not with the mathematical sets.

Editorial extensions

If this is right

  • Mandelbrot and Julia membership is, on these sampled domains, learnable from four complex iterate values; thresholding is not the best available classifier.
  • Numerical fractal visualization and escape-time computations can be accelerated by training a cheap classifier on a small number of iterates instead of iterating until escape.
  • If Hypothesis 1 holds, a measurable function of finitely many early magnitudes can approximate membership in a neighborhood of the boundary with error tending to zero, so the boundary's classification complexity is concentrated in early-orbit geometry.
  • Model comparisons give concrete diagnostics: CART over-relies on the first iterate, recurrent networks contribute little at four iterates, and KNN/RF are the most robust; these empirical regularities are targets for further mathematical analysis.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if Hypothesis 1 holds, boundary membership is, up to small probability, a function of finitely many early magnitudes; this suggests a quantitative finite-escape criterion beyond the classical |z_n|>2 sufficient condition.
  • A testable extension the paper does not run: relabel the test points by iterating far beyond 100 steps and re-measure accuracy; this would separate learning of the true asymptotic set from learning the 100-iteration proxy.
  • One could estimate the minimal number of iterates needed for a target accuracy on each fractal family and check whether the rapid convergence of the classifiers predicted in Hypothesis 1 holds numerically.
  • The same training design could be applied to other families, such as higher-degree polynomials or rational maps, to see whether finite-early-iterate regularity is generic or special to quadratic orbits.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper trains seven supervised classifiers (CART, KNN, MLP, LSTM, BiLSTM, RF, CNN) on the first I=1,...,4 complex iterates of quadratic orbits and evaluates them as classifiers of membership in the Mandelbrot set, a Julia set, and a box around the Mandelbrot boundary. Labels are generated by the classical escape rule with radius 2 and a 100-iteration cutoff. The authors report that all ML methods outperform a THRESH baseline matched on the same number of iterates, with the best models reaching about 95% accuracy on the Mandelbrot set, about 88% on the Julia set, and 91-93% near the boundary. The paper also proposes Hypothesis 1, asserting that a measurable function of the first I iterate magnitudes can classify boundary membership with error tending to zero. Code is released.

Significance. If the central claim were established, the paper would offer a striking example of short-time dynamical information encoding asymptotic boundedness, and the model comparison could motivate new conjectures about fractal geometry. The experimental protocol has genuine strengths: ten independent training runs, one-million-point test sets, standard deviations, and a baseline matched on the number of iterates. However, the evidence currently supports classification of a finite-time escape rule, not classification of the mathematical Mandelbrot/Julia sets. The mismatch between the labels, the stated asymptotic target, and the features used in Hypothesis 1 means that the headline conclusions go beyond what the experiments demonstrate. With a corrected label oracle or a carefully reframed finite-time claim, the systematic benchmark would be a useful contribution.

major comments (3)
  1. [Section 4.2; Section 6] The labels used in every experiment are not asymptotic membership: a point is labeled bounded iff none of the first 100 iterates exceeds r_w=2, so points whose escape time exceeds 100 are mislabeled as bounded. Section 6's statement that the ground truth is 'analytically determined' is therefore inaccurate. Tables 2-4 measure agreement with this 100-iteration rule, not with the mathematical sets M and J. In particular, the boundary experiment (§5.3) samples around |Q^100_c(0)|=r_w, exactly where the proxy is least reliable, and the reported 82.5% unbounded rate is an artifact of the cutoff. Thus the accuracy values do not support Hypothesis 1's claim about true boundary membership.
  2. [Hypothesis 1; Section 4.2] Hypothesis 1 concerns f(|z_1|,...,|z_I|), i.e., magnitudes only, but the experiments use the full complex coordinates Re(z_i), Im(z_i) (and Re(c), Im(c)) as inputs. The reported accuracies therefore do not instantiate the hypothesis; a classifier restricted to magnitudes could have substantially different performance. The conjecture should be restated to use the actual inputs, or supported by additional magnitude-only experiments.
  3. [Section 4.1-4.2; Tables 2-4] The comparison with THRESH is ambiguous. The THRESH scores in Tables 2-4 use only the first I iterates, whereas the 'classical threshold algorithm' described in Section 4.2 is run for 100 iterations and is used to generate labels. Run for 100 iterations, THRESH is the label oracle and would achieve essentially perfect accuracy on the surrogate task. The abstract's claim that ML methods outperform 'traditional numerical approaches' therefore conflates two different thresholds. The paper should state explicitly that THRESH is restricted to the same short decision horizon, and the conclusion should be phrased as 'with only I iterates at decision time.'
minor comments (4)
  1. [Section 4.3; Table 5] Hyperparameter reporting is inconsistent. The text says 50 hidden neurons for LSTM/BiLSTM, while Table 5 lists LSTMunits/biLSTMUnits as 7. The text says total iterations 500 and MLP max_iter 150, while Table 5 gives MaxEpochs 200. These discrepancies should be reconciled for reproducibility.
  2. [Section 5.4; Tables 2-4] The claim of 'rapid convergence' of classification error as I grows is not visible in the tables: KNN/RF on the Mandelbrot set change from 95.6/95.7 (I=1) to 95.8/95.6 (I=4), and the boundary accuracies are essentially flat. The convergence part of Hypothesis 1 should be presented as pure speculation, not as a pattern visible in the data.
  3. [Section 5.3] Even on the finite-time proxy, the near-boundary results should be read against the trivial classifier: 82.5% of points are labeled unbounded, so an always-unbounded classifier achieves 82.5%. The 91-93% figures for CART/KNN/RF are only modestly above that baseline, and the class imbalance makes accuracy a less informative metric than, say, balanced accuracy or ROC-AUC.
  4. [Section 2.1] The notation 'zk := Q^k_c(z)' followed by 'so that c = z1 is the first nonzero term' is confusing because the Mandelbrot orbit starts at z0=0; consider defining z0 explicitly and avoiding the reuse of z for both the map variable and the orbit term.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline 'asymptotic' results measure a 100-iteration escape proxy, so the central claim reduces to fitting that proxy.

  1. self definitional [Section 4.2 (label generation) and Section 6 (claim of asymptotic target)]
    "The labels (bounded vs. unbounded) were determined using the classical threshold algorithm, which was run for 100 iterations. A point was labeled as unbounded if any of the first 100 iterations of its orbit had a magnitude greater than r_w = 2. ... high classification accuracy was achieved using only four orbit iterates, despite the target variable representing an asymptotic property of the dynamical system."

    Every training and test label is produced by the finite-time rule of Section 4.2, not by the asymptotic definition of M in Section 2.1. The same threshold family is also the baseline (THRESH, Section 4.1), so the reported accuracies and the 'ML beats THRESH' comparisons measure agreement with a 100-iteration escape oracle. Calling this target 'an asymptotic property' and saying the ground truth is 'analytically determined' (Section 6) renames the proxy as the mathematical target. Thus the central claim that early iterates encode asymptotic boundedness is a fit to the finite-time labels, not an independent derivation; Hypothesis 1 about true boundary membership is not tested by these experiments.

full rationale

The pipeline is internally honest only if the task is predicting the 100-iteration threshold rule: classifiers are trained on 10k fresh samples and tested on 1M new points, and model outputs are not used to set labels. For that finite-time task, the accuracies are real and would warrant a score of 0-2. The circularity arises from the Section 6 interpretation: the paper equates the finite-time target with the asymptotic property and claims 'ground truth is analytically determined.' Since the labels come from the same threshold family as the baseline THRESH, the headline numbers (approx. 95%, 88%, and 91-93% near the boundary) and the claimed superiority over THRESH all validate agreement with a 100-iteration escape oracle, not membership in the true Mandelbrot or Julia sets. Additionally, Hypothesis 1 is stated for magnitudes |z_1|,...,|z_I|, while all experiments use full complex coordinates Re(z_i), Im(z_i), and the boundary experiment has a 82.5% majority class which several models collapse to. These are supporting weaknesses, but the load-bearing circular step is the substitution of the finite-time threshold rule for the asymptotic target. No problematic self-citation or uniqueness import is present.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The benchmark rests on a finite-time escape oracle (100 iterations, radius 2) whose outputs are treated as ground truth; the boundary sampling adds a hand-chosen box size and a numerically solved level set; the conjecture postulates a magnitude-only classifier without testing it. The I sweep (I=1..4) is a design choice, not a fitted quantity, but it shapes the headline results.

free parameters (5)
  • Label iteration budget = 100
    Hand-chosen in Section 4.2; defines the ground truth. Points with escape time above 100 are labeled bounded, biasing the boundary experiment.
  • Escape threshold r_w = 2
    Fixed in Section 4.1 although the text notes r_w depends on c; used both for label generation and the THRESH baseline.
  • Boundary box side length = 0.07
    Hand-chosen in Section 4.2 for sampling 'near the boundary'; controls class balance (82.5% majority in Section 5.3) and the measured accuracies.
  • LSTM/BiLSTM hidden units = 7 (Table 5) or 50 (Section 4.3)
    The paper reports two contradictory values; the architecture used is ambiguous and reproducibility is compromised.
  • Per-model hyperparameters = KNN k=3; RF 100 trees; MLP (7,7,7); CNN 32 filters; dropout 0.2-0.3
    Hand-chosen from Table 5 and Appendix A; not tuned systematically, but they affect every reported accuracy.
assumptions (5)
  • standard math Escape criterion: if |z_n| > 2 and |z_n| > |c| then the quadratic orbit diverges; used to label all data.
    Proved in Section 4.1 via the triangle inequality; a sufficient condition for divergence.
  • ad hoc to paper Membership in the fractal set is approximated by escape within 100 iterations at threshold 2.
    Section 4.2 label rule; an approximation of the asymptotic definition, not the definition itself, and most inaccurate near the boundary.
  • ad hoc to paper The curve {|Q^100_c(0)| = 2}, solved numerically, is an adequate stand-in for the Mandelbrot boundary with 0.07 boxes.
    Sections 4.2 and 5.3 boundary sampling; this is a finite-time lemniscate-like level set, not the fractal boundary, and the observed 82.5% majority class shows heavy asymmetry.
  • domain assumption Uniform sampling in each domain is representative for accuracy evaluation.
    Section 4.2; the Mandelbrot domain is dominated by exterior points (roughly 25% in-set area), so accuracy is driven by the majority class.
  • ad hoc to paper Hypothesis 1: a measurable function of the first I iterate magnitudes classifies membership near the boundary with error less than epsilon and rapid convergence as I grows.
    Section 6; this is the paper's conjecture. It is untested in the magnitude-only form, and the empirical error for KNN and CART is flat in I over Tables 2 to 4, so the rapid-convergence clause lacks support.
invented entities (1)
  • Classification function f over iterate magnitudes (Hypothesis 1)
    purpose: Postulated function from (|z1|,...,|zI|) to {0,1} that approximates membership near the fractal boundary with arbitrarily small error.
    No falsifiable handle is provided: the experiments use full complex features, not magnitude sequences, and the error trend across I shows no convergence to zero in the tested range (Tables 2 to 4). It is the content of the conjecture, not an established object.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning with Mandelbrot and Julia." pith.science (2026). https://pith.science/paper/4ECC5NUY

@misc{pith2026250900903,
  author       = {Pith},
  title        = {Pith review of: Learning with Mandelbrot and Julia},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4ECC5NUY}},
  note         = {Machine review of arXiv:2509.00903}
}
read the original abstract

Recent developments in applied mathematics increasingly employ machine learning (ML)-particularly supervised learning-to accelerate numerical computations, such as solving nonlinear partial differential equations. In this work, we extend such techniques to objects of a more theoretical nature: the classification and structural analysis of fractal sets. Focusing on the Mandelbrot and Julia sets as principal examples, we demonstrate that supervised learning methods-including Classification and Regression Trees (CART), K-Nearest Neighbors (KNN), Multilayer Perceptrons (MLP), and Recurrent Neural Networks using both Long Short-Term Memory (LSTM) and Bidirectional LSTM (BiLSTM), Random Forests (RF), and Convolutional Neural Networks (CNN)-can classify fractal points with significantly higher predictive accuracy and substantially lower computational cost than traditional numerical approaches, such as the thresholding technique. These improvements are consistent across a range of models and evaluation metrics. Notably, KNN and RF exhibit the best overall performance, and comparative analyses between models (e.g., KNN vs. LSTM) suggest the presence of novel regularity properties in these mathematical structures. Collectively, our findings indicate that ML not only enhances classification efficiency but also offers promising avenues for generating new insights, intuitions, and conjectures within pure mathematics.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 46 canonical work pages

  1. [1]

    Physics and Fractal Structures

    Jean-Fran¸ cois Gouyet. Physics and Fractal Structures . Masson Springer, 1996

    work page 1996

  2. [2]

    The Fractal Geometry of Nature , volume 1

    Benoit B Mandelbrot. The Fractal Geometry of Nature , volume 1. WH Freeman New York, 1982

    work page 1982

  3. [3]

    Fractal-based methods in analysis

    Herb Kunze, Davide La Torre, Franklin Mendivil, and Edward R Vrscay. Fractal-based methods in analysis . Springer Science & Business Media, 2011

    work page 2011

  4. [4]

    Introduction to the Modern Theory of Dynamical Systems , volume 54 of Encyclopedia of Mathematics and its Applications

    Anatole Katok and Boris Hasselblatt. Introduction to the Modern Theory of Dynamical Systems , volume 54 of Encyclopedia of Mathematics and its Applications . Cam- bridge University Press, Cambridge, 1995

    work page 1995

  5. [5]

    A first course in chaotic dynamical systems: Theory and experiment

    Robert L Devaney, Peter B Siegel, A John Mallinckrodt, and Susan McKay. A first course in chaotic dynamical systems: Theory and experiment. Computers in Physics , 7(4):416–417, 1993

    work page 1993

  6. [6]

    Mandelbrot set and Julia sets of fractional order

    Marius-F Danca and Michal Feˇ ckan. Mandelbrot set and Julia sets of fractional order. Nonlinear Dynamics , 111(10):9555–9570, 2023

    work page 2023

  7. [7]

    Zalcman functions and similarity be- tween the Mandelbrot set, Julia sets, and the tricorn

    Tomoki Kawahira. Zalcman functions and similarity be- tween the Mandelbrot set, Julia sets, and the tricorn. Analysis and Mathematical Physics , 10(2):16, 2020

    work page 2020

  8. [8]

    Similarity between the Mandelbrot set and Julia sets

    Tan Lei. Similarity between the Mandelbrot set and Julia sets. Communications in mathematical physics , 134:587– 617, 1990

    work page 1990

Show all 48 references
  1. [9]

    Haus- dorff dimension of Julia sets in the logistic family

    Neil Dobbs, Jacek Graczyk, and Nicolae Mihalache. Haus- dorff dimension of Julia sets in the logistic family. Com- munications in Mathematical Physics , 399(2):673–716, 2023

    2023

  2. [10]

    On the directional derivative of the Hausdorff dimension of quadratic polynomial Julia sets at-2

    Ludwik Jaksztas. On the directional derivative of the Hausdorff dimension of quadratic polynomial Julia sets at-2. Advances in Mathematics , 433:109297, 2023

    2023

  3. [11]

    Automatic prediction of tumour malignancy in breast cancer with fractal di- mension

    Alan Chan and Jack A Tuszynski. Automatic prediction of tumour malignancy in breast cancer with fractal di- mension. Royal Society open science , 3(12):160558, 2016

    2016

  4. [12]

    A machine learning approach to auto- matic detection of irregularity in skin lesion border using dermoscopic images

    Abder-Rahman Ali, Jingpeng Li, Guang Yang, and Sally Jane O’Shea. A machine learning approach to auto- matic detection of irregularity in skin lesion border using dermoscopic images. PeerJ Computer Science , 6:e268, 2020

    2020

  5. [13]

    Ma- chine learning and fractal theory models for landslide sus- ceptibility mapping: Case study from the Jinsha River Basin

    Qiao Hu, Yi Zhou, Shixing Wang, and Futao Wang. Ma- chine learning and fractal theory models for landslide sus- ceptibility mapping: Case study from the Jinsha River Basin. Geomorphology, 351:106975, 2020

    2020

  6. [14]

    Predicting the future of discrete sequences from fractal representations of the past

    Peter Tino and Georg Dorffner. Predicting the future of discrete sequences from fractal representations of the past. Machine Learning, 45(2):187–217, 2001

    2001

  7. [15]

    Schema genetic algorithm for fractal image compres- sion

    Ming-Sheng Wu, Jyh-Horng Jeng, and Jer-Guang Hsieh. Schema genetic algorithm for fractal image compres- sion. Engineering Applications of Artificial Intelligence , 20(4):531–538, 2007

    2007

  8. [16]

    Pre-training without natural images

    Hirokatsu Kataoka, Kazushige Okayasu, Asato Mat- sumoto, Eisuke Yamagata, Ryosuke Yamada, Nakamasa Inoue, Akio Nakamura, and Yutaka Satoh. Pre-training without natural images. In Proceedings of the Asian Con- ference on Computer Vision , 2020

    2020

  9. [17]

    Improving fractal pre-training

    Connor Anderson and Ryan Farrell. Improving fractal pre-training. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages 1300–1309, 2022

    2022

  10. [18]

    Learning fractals by gradient descent

    Cheng-Hao Tu, Hong-You Chen, David Carlyn, and Wei- Lun Chao. Learning fractals by gradient descent. In Proceedings of the AAAI Conference on Artificial Intel- ligence, volume 37, pages 2456–2464, 2023

    2023

  11. [19]

    Machine learning and fractal geometry

    Peter Bloem. Machine learning and fractal geometry. Master’s thesis, University of Amsterdam, Amsterdam, 2010

    2010

  12. [20]

    What do deep neural networks under- stand of fractals?, 2017

    Gr´ egory Chˆ atel. What do deep neural networks under- stand of fractals?, 2017

    2017

  13. [21]

    On the quadratic mapping z → z2 − µ for complex µ and z: the fractal structure of its M set, and scaling

    Benoit B Mandelbrot. On the quadratic mapping z → z2 − µ for complex µ and z: the fractal structure of its M set, and scaling. Physica D: Nonlinear Phenomena , 7(1-3):224–239, 1983

    1983

  14. [22]

    Disconnected Julia sets

    Paul Blanchard. Disconnected Julia sets. In Chaotic Dynamics and Fractals , pages 181–201. Elsevier, 1986

    1986

  15. [23]

    Clinical time series pre - diction: Toward a hierarchical dynamical system frame- work

    Zitao Liu and Milos Hauskrecht. Clinical time series pre - diction: Toward a hierarchical dynamical system frame- work. Artificial intelligence in medicine , 65(1):5–18, 2015

    2015

  16. [24]

    An intelligent syste m for financial time series prediction combining dynamical systems theory, fractal theory, and statistical methods

    Oscar Castillo and Patricia Melin. An intelligent syste m for financial time series prediction combining dynamical systems theory, fractal theory, and statistical methods. In Proceedings of 1995 Conference on Computational In- telligence for Financial Engineering (CIFEr) , pages 151–

    1995

  17. [25]

    Classification of chaotic time series with deep learning

    Nicolas Boull´ e, Vassilios Dallas, Yuji Nakatsukasa, a nd D Samaddar. Classification of chaotic time series with deep learning. Physica D: Nonlinear Phenomena , 403:132261, 2020

    2020

  18. [26]

    Deep learning for time series classification: a review

    Hassan Ismail Fawaz, Germain Forestier, Jonathan We- ber, Lhassane Idoumghar, and Pierre-Alain Muller. Deep learning for time series classification: a review. Data Min- ing and Knowledge Discovery , 33(4):917–963, 2019

    2019

  19. [27]

    Inception- time: Finding alexnet for time series classification

    Hassan Ismail Fawaz, Benjamin Lucas, Germain Forestier, Charlotte Pelletier, Daniel F Schmidt, Jonathan Weber, Geoffrey I Webb, Lhassane Idoumghar, Learning with Mandelbrot and Julia 11 Pierre-Alain Muller, and Fran¸ cois Petitjean. Inception- time: Finding alexnet for time ser...

    1936

  20. [28]

    Philipp Grohs, Fabian Hornung, Arnulf Jentzen, and Philippe von Wurstemberger. A Proof that Artificial Neu- ral Networks Overcome the Curse of Dimensionality in the Numerical Approximation of Black–Scholes Partial Differential Equations , volume 284 of Memoirs of the American Ma...

    2023

  21. [29]

    Solving high-dimensional partial differential equations using dee p learning

    Jiequn Han, Arnulf Jentzen, and E Weinan. Solving high-dimensional partial differential equations using dee p learning. Proceedings of the National Academy of Sci- ences, 115(34):8505–8510, 2018

    2018

  22. [30]

    Solving differential equations using deep neural networks

    Craig Michoski, Miloˇ s Milosavljevi´ c, Todd Oliver, an d David R Hatch. Solving differential equations using deep neural networks. Neurocomputing, 399:193–212, 2020

    2020

  23. [31]

    A deep-genetic algorithm (deep-GA) approach for high-dimensional nonlinear parabolic par- tial differential equations

    Endah RM Putri, Muhammad L Shahab, Mohammad Iqbal, Imam Mukhlash, Amirul Hakam, Lutfi Mardianto, and Hadi Susanto. A deep-genetic algorithm (deep-GA) approach for high-dimensional nonlinear parabolic par- tial differential equations. Computers & Mathematics with Applications , 1...

    2024

  24. [32]

    Deep hidden physics models: Deep learn- ing of nonlinear partial differential equations

    Maziar Raissi. Deep hidden physics models: Deep learn- ing of nonlinear partial differential equations. The Jour- nal of Machine Learning Research , 19(1):932–955, 2018

    2018

  25. [33]

    Hidden physics models: Machine learning of nonlinear partial dif- ferential equations

    Maziar Raissi and George Em Karniadakis. Hidden physics models: Machine learning of nonlinear partial dif- ferential equations. Journal of Computational Physics , 357:125–141, 2018

    2018

  26. [34]

    Physics-informed neural networks: A deep learn- ing framework for solving forward and inverse problems involving nonlinear partial differential equations

    Maziar Raissi, Paris Perdikaris, and George E Karni- adakis. Physics-informed neural networks: A deep learn- ing framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics , 378:686–707, 2019

    2019

  27. [35]

    Neural networks for bifurcation and linear stability analysis of steady states in partial differential equations

    Muhammad Luthfi Shahab and Hadi Susanto. Neural networks for bifurcation and linear stability analysis of steady states in partial differential equations. Applied Mathematics and Computation , 483:128985, 2024

    2024

  28. [36]

    Neural networks for bifurcation and linear sta- bility analysis of steady states in partial differential equ a- tions

    Muhammad Luthfi Shahab and Hadi Susanto. Corrigen- dum to “Neural networks for bifurcation and linear sta- bility analysis of steady states in partial differential equ a- tions”[Appl. Math. Comput. 483 (2024) 128985]. Applied Mathematics and Computation , 495:129319, 2025

    2024

  29. [37]

    Neural networks for high-dimensional solutions and snaking bifurcations in nonlinear lattices

    Muhammad Luthfi Shahab, Fidya Almira Suheri, Rudy Kusdiantara, and Hadi Susanto. Neural networks for high-dimensional solutions and snaking bifurcations in nonlinear lattices. Manuscript under review, 2025

    2025

  30. [38]

    Un- derstanding and mitigating gradient flow pathologies in physics-informed neural networks

    Sifan Wang, Yujun Teng, and Paris Perdikaris. Un- derstanding and mitigating gradient flow pathologies in physics-informed neural networks. SIAM Journal on Sci- entific Computing , 43(5):A3055–A3081, 2021

    2021

  31. [39]

    Extended physics-informed neural networks (XPINNs): A general- ized space-time domain decomposition based deep learn- ing framework for nonlinear partial differential equations

    Ameya D Jagtap and George Em Karniadakis. Extended physics-informed neural networks (XPINNs): A general- ized space-time domain decomposition based deep learn- ing framework for nonlinear partial differential equations . Communications in Computational Physics , 28(5), 2020

    2020

  32. [40]

    A composite neural network that learns from multi-fidelity data: Appli- cation to function approximation and inverse PDE prob- lems

    Xuhui Meng and George Em Karniadakis. A composite neural network that learns from multi-fidelity data: Appli- cation to function approximation and inverse PDE prob- lems. Journal of Computational Physics , 401:109020, 2020

    2020

  33. [41]

    Transfer learning on physics-informed neural networks for tracking the hemo- dynamics in the evolving false lumen of dissected aorta

    Mitchell Daneker, Shengze Cai, Ying Qian, Eric Myzelev, Arsh Kumbhat, He Li, and Lu Lu. Transfer learning on physics-informed neural networks for tracking the hemo- dynamics in the evolving false lumen of dissected aorta. Nexus, 1(2), 2024

    2024

  34. [42]

    Physics guided neural networks for modelling of non- linear dynamics

    Haakon Robinson, Suraj Pawar, Adil Rasheed, and Omer San. Physics guided neural networks for modelling of non- linear dynamics. Neural Networks , 154:333–345, 2022

    2022

  35. [43]

    Zero-shot forecast- ing of chaotic systems

    Yuanzhao Zhang and William Gilpin. Zero-shot forecast- ing of chaotic systems. arXiv preprint arXiv:2409.15771 , 2024

    2024 arXiv

  36. [44]

    Discovering governing equations from data by sparse iden- tification of nonlinear dynamical systems

    Steven L Brunton, Joshua L Proctor, and J Nathan Kutz. Discovering governing equations from data by sparse iden- tification of nonlinear dynamical systems. Proceedings of the national academy of sciences , 113(15):3932–3937, 2016

    2016

  37. [45]

    Data-driven discovery of coordinates and governing equations

    Kathleen Champion, Bethany Lusch, J Nathan Kutz, and Steven L Brunton. Data-driven discovery of coordinates and governing equations. Proceedings of the National Academy of Sciences , 116(45):22445–22451, 2019

    2019

  38. [46]

    Chaos as an interpretable benchmark for forecasting and data-driven modelling

    William Gilpin. Chaos as an interpretable benchmark for forecasting and data-driven modelling. arXiv preprint arXiv:2110.05266, 2021

    2021 arXiv

  39. [47]

    Tjahjono

    V.R. Tjahjono. Scripts for Mandelbrot and Julia paper. https://github. com/venansiusrt/fractal-paper, 2025. Ac- cessed: 2025-06-13

    2025

  40. [48]

    Discovering physical concepts with neural networks

    Raban Iten, Tony Metger, Henrik Wilming, L ´ ıdia Del Rio, and Renato Renner. Discovering physical concepts with neural networks. Physical Review Letters , 124(1):010508, 2020. A T raining Details A GitHub repository containing all implementation details necessary to reproduce...

    2020

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.