From Pen to Palette: Mathematical Analysis of van Gogh's Psychological Trajectory

Anna Singley; Eli Goldwyn; Jessica Logue; Mark Pitzer

REVIEW 4 major objections 7 minor 19 references

From Pen to Palette: Mathematical Analysis of van Gogh's Psychological Trajectory

T0 review · 4 major / 7 minor · reviewed 2026-07-30 · grok-4.5

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

Pith's one-line read Van Gogh’s letters and paintings show coordinated mathematical early-warning signs before the 1888 ear crisis.

desk verdict Solid dual-stream archival case study with a real matched n=524 series, but the CSD “early warning” claim is retrospective window-tuning around known crises without nulls, and the modern-SVM→1880s clinical labels are unvalidated. read the letter →

arxiv 2607.23457 v1 pith:O5B25YUH submitted 2026-07-26 physics.soc-ph

Anna Singley , Eli Goldwyn , Mark Pitzer , Jessica Logue This is my paper · ORCID

classification physics.soc-ph

keywords criticalslowingdownvanGoghsentimentanalysisfractaldimensionpsychologicaltippingpointsdynamicalsystemsinterpersonaltheoryofsuicidecomputationalarthistory

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

This paper treats van Gogh’s final decade as a dynamical system and asks whether his psychology left measurable early-warning signs before major breakdowns. It scores emotional content in hundreds of letter segments and measures the fractal complexity of hundreds of paintings, then looks for the statistical signatures that dynamical-systems theory calls critical slowing down—rising variance and autocorrelation as a system nears a tipping point. The clearest case is the months before the December 1888 ear incident, when desire-for-death language, acquired-capability language, and brushwork complexity all became more volatile together. A modest but significant link also appears between death-related sentiment and visual complexity across more than five hundred matched days. The point is that archival words and pictures can be read as multi-modal sensors of a single unstable psychological system, giving quantitative backing to biographical crisis narratives.

What carries the argument

Critical slowing down (CSD): the rise in variance, lag-1 autocorrelation, and higher moments that dynamical systems exhibit when they approach a tipping point; here computed on rolling monthly windows of letter-sentiment scores and Minkowski–Bouligand fractal dimension of preprocessed brushwork.

What would settle it

Re-score the same letter paragraphs with an independently trained period-appropriate or human-coded sentiment scheme; if the pre-ear coordinated variance and autocorrelation spikes in desire-for-death and acquired-capability disappear, the CSD claim fails.

Watch

Extended reading notes

Core claim

Van Gogh’s psychological system showed critical-slowing-down signatures—coordinated rises in variance and lag-1 autocorrelation across desire for death, acquired capability, and painting fractal dimension—in the roughly four-to-six months before the 23 December 1888 ear incident, together with a statistically significant positive correlation (r = 0.193, n = 524) between visual complexity and death sentiment.

Load-bearing premise

That a sentiment model trained on modern social-media posts labeled with today’s clinical suicide categories produces valid continuous scores for the same categories when applied to segmented nineteenth-century personal letters.

Editorial extensions

If this is right

Psychological tipping points can leave detectable multi-modal statistical traces in historical archives, not only in modern clinical time series.
Visual complexity of artistic output can couple to conscious death-related cognition even when the linear correlation is modest.
The ear crisis and the final suicide show different volatility signatures—chaotic pre-ear oscillations versus locked-in high variance before death—suggesting distinct dynamical routes to crisis.
CSD-style rolling variance and autocorrelation offer an objective complement to traditional biographical periodization of an artist’s life.
The same pipeline can be reapplied to other figures who left dense dated correspondence and dated creative work.

Reading between the lines

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

If domain transfer of the sentiment model is the main risk, a small expert-annotated sample of van Gogh letters could calibrate or replace the social-media SVM and immediately strengthen or refute the time-series results.
The dissociation noted before the ear incident—high visual complexity and acquired capability with relatively low explicit death sentiment—suggests impulsive or non-ideational pathways that modern risk models might also need to track via non-linguistic channels.
Geographic and genre differences in fractal dimension (Paris vs Arles vs Saint-Rémy; self-portraits vs still lifes) could be tested as covariates to separate stylistic fashion from psychological volatility.

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.

Desk Editor's Note

read the letter

The new piece here is not fractal van Gogh (Abry, Aragón already did versions of that) or letter scholarship. It is the joint pipeline: 719 letter segments scored on suicide-theory channels, 569 dated paintings with Minkowski dimension after a DoG–LIC–Otsu pipeline, temporally matched to n=524, then rolling variance/autocorrelation read against a dated crisis table. That dual series and the concrete numbers (death-sentiment vs fractal r=0.193, plus the other ~0.10 rs) are real empirical output this paper owns.

What it does well: clear event table, explicit matching rule (30-day window, mean separation 2.1 days), honest reporting of small effects, and the multi-modal framing is useful for digital humanities and computational psychiatry people who want a worked historical example. The image pipeline is reproducible in principle; the Fligner–Killeen variance-of-variance tests are a nice extra layer. Citation pattern is fine—CSD classics, Joiner IPTS, the prior van Gogh math papers.

Soft spots, in proportion. The load-bearing one is the CSD detection claim. Phases are defined relative to the already-known ear date (baseline 4–6 mo before, CSD 2–4, stabilization 0–2), windows were swept 5–30 points to an “optimal” month on the same data, and there is no surrogate or random-date null anywhere. On a nonstationary decade-long series that is exactly the setup that produces apparent pre-event rises. Short-window lag-1 AC is also noisy. Even perfect sentiment scores would not fix that. Second, and almost as serious: the SVM is trained on modern social-media posts labeled with contemporary clinical constructs, then applied to segmented 19th-century letters with no domain validation, no period lexicon check, no human inter-rater on a van Gogh sample. If that transfer fails, the named dimensions (“acquired capability,” etc.) are not what the time series measure. Effect sizes are small and the daily-interpolation n=524 inflates precision a bit. No code or locked data.

Who it is for: people building multi-modal historical psych pipelines, and CSD-in-psychiatry method readers who can stomach a single-case illustration. Not for someone wanting a clean prospective early-warning result.

I would send it to referees. It is important enough as a methods-plus-case paper and formally grounded enough to deserve the fight over null models and classifier validation rather than a desk reject. I would not cite the CSD claim as settled; I might cite the matched series and the correlation table after revision.

Referee Report

4 major / 7 minor

Summary. The authors apply computational methods to van Gogh's final decade: an SVM sentiment classifier (trained on >100k modern social-media posts, Platt-scaled to continuous scores) is applied to 719 letter segments spanning 1872–1890, and box-counting fractal dimension is computed for 569 dated paintings after a DoG/LIC/Otsu preprocessing pipeline. After matching letters and paintings within a 30-day window and resampling to a daily grid by linear interpolation (n=524 pairs), they report a significant correlation between fractal dimension and "desire for death" sentiment (r=0.193, p<0.001), plus weaker correlations with stress, substance use, anger, and acquired capability. Framing the trajectory with critical slowing down (CSD) theory, they argue that rolling variance and lag-1 autocorrelation rose in a coordinated way across multiple dimensions in the 4–6 months before the 23 Dec 1888 ear incident, with a different (persistently elevated, less erratic) signature before the 1890 suicide, and interpret this as evidence of detectable psychological tipping-point precursors in a historical figure.

Significance. If the results hold, the paper demonstrates a replicable, quantitative, multi-modal methodology for studying psychological dynamics in archival figures — a genuinely interesting bridge between dynamical-systems early-warning theory, computational art history, and digital humanities. Credit is due for several real strengths: a large digitized corpus (700+ letter segments, 569 paintings), a fully specified and reproducible image-processing pipeline, explicit test statistics, honest reporting of modest effect sizes (the headline r=0.193 is not oversold as strong), and a framework that is in principle falsifiable via the CSD predictions it invokes. However, the central claims — that CSD signatures are "detectable" prospectively, and that the cross-modal correlation is statistically meaningful — currently rest on analysis choices that are not adequately defended: event-anchored windows with tuned parameters and no null model, an unvalidated domain transfer of the sentiment classifier, and significance tests that ignore the dependence structure created by interpolation. The paper's conclusions should be regarded as provisional until these are addressed.

major comments (4)

[§2.4 and §3.2] The claim of 'prospective early warning signals' (§3.2) is undermined by the analysis design: the baseline/CSD/stabilization phases are defined relative to the already-known crisis date (§2.4: 4–6, 2–4, 0–2 months before), and the rolling window size was 'systematically varied' over 5–30 points on the same data from which the signature is then reported. No surrogate or null analysis is presented anywhere: how often would an arbitrarily chosen date in 1881–1890, analyzed with the same tuned windows, show a comparable coordinated pre-event rise? Given that the series are nonstationary (the paper itself reports a significant decadal trend in fractal dimension, Fig. 3A), elevated local variance in some window before a hand-picked event is expected by chance at some rate. A minimal fix is a surrogate test: repeat the full pipeline for many pseudo-event dates drawn from the record and report w
[§2.2] The entire sentiment time series — and hence both the cross-modal correlations and the CSD indicators — depends on an SVM trained on modern social-media posts labeled with contemporary clinical suicide-theory categories, applied to segmented 19th-century personal correspondence (mostly to Theo). No validation on the target domain is reported: no hand-labeled subset of van Gogh letters, no inter-rater agreement, no comparison of score distributions between training and target domains. If the domain transfer fails, 'desire for death' and 'acquired capability' scores are not measuring the claimed constructs. This is load-bearing for every downstream claim. The paper should include at minimum a validation set of van Gogh letter segments coded by the psychology/history co-authors against the same rubric, with classifier agreement reported.
[§3.1] The reported significance of the headline correlation (r=0.193, p<0.001, n=524) is computed under an independence assumption that the pipeline violates. Both series are resampled to a common daily grid by linear interpolation (§3.1, para 2), so adjacent daily values are mechanically correlated and the effective sample size is far below 524; the t-test p-values are therefore anti-conservative, possibly dramatically so given that the underlying sampling is letters and paintings separated by days. The same concern applies to the other four reported correlations (r≈0.10–0.12). The analysis should be redone with tests robust to autocorrelation (e.g., block bootstrap, effective-n adjustment, or prewhitening) on the unmatched original series. Additionally, seven sentiment dimensions were tested against fractal dimension with no multiple-comparison control; at Bonferroni α≈0.007, the acquired-ca
[§3.2 and Fig. 5] The 'coordinated multi-dimensional CSD' argument (§3.2) is weaker than presented on two grounds. First, the sentiment dimensions are not independent channels: desire for death, acquired capability, social isolation, etc. are outputs of a single classifier applied to the same text segments, so correlated volatility among them can reflect shared measurement structure rather than coupled psychological dynamics; only the text-versus-painting comparison (Fig. 5) is genuinely cross-modal, and it is supported only by 'qualitative alignment' of autocorrelation traces. Second, fractal dimension is confounded with period, location, and genre: the paper's own Figs. 6.1–6.2 show systematic differences (Paris vs. Arles vs. Saint-Rémy; self-portraits vs. still lifes), and Fig. 3A shows a significant decadal trend. The mid-1888 rise in fractal-dimension autocorrelation coincides with the Arles period's

minor comments (7)

[§3.2, variance paragraph] Several reported summary statistics appear internally inconsistent: desire for death 'Pre-ear' is given as σ²=0.0088 with IQR=0.00020–0.0026 (the headline value lies outside the stated IQR); the 'Between events' IQR is printed as 0.0064–0.0011 (lower bound exceeds upper bound); social isolation Pre-suicide has σ²=0.0015 with IQR=0.00010–0.00021, and acquired capability Pre-ear σ²=0.0015 with IQR=0.00035–0.0040 is also outside its IQR. Please audit these numbers.
[Fig. 6 caption vs. §3.2] Fig. 6 caption states points are sorted by whether they are 'within 90 days of the event listed,' but the text describes 90-day rolling variance aggregated over three multi-year periods (Pre-ear spanning 1874–1888). Caption and text should be reconciled.
[Fig. 4] Fig. 4's vertical axis is labeled 'change in autocorrelation,' but §3.2 discusses the figure primarily in terms of volatility/variance; please clarify what quantity is plotted.
[Table 2] Table 2 is not in chronological order (event #6 dated 1883-09-01 follows #5 dated 1885-03-26; #2 1881 follows #1 1880). Since Fig. 4 references these event numbers, ordering errors risk confusing readers.
[§2.2, Abstract, §2.3] Grammar/typography: 'Sentiments are selected based on both psychological theory and relevance to van Gogh's documented experiences are:' (§2.2); the abstract's 'followed by stabilization before the crisis' is confusing (stabilization precedes the crisis in the authors' own phase scheme); §2.3 references 'Figure 6.1' and 'Figure 6.2' before the appendix is introduced.
[End matter] No code- or data-availability statement is included. Given the emphasis on a 'replicable methodology' (§1, §5), releasing the classifier scores, painting metadata, and analysis scripts would substantially strengthen the paper.
[Table 1] Table 1 uses literary excerpts (Sexton, Shakespeare, Plath) to illustrate the classifier; since the van Gogh corpus is 'too extensive' for tabulation, consider instead showing a few short anonymized letter segments with scores — this would be more informative about on-domain behavior.

Circularity Check

1 steps flagged · score 1.0 of 10

No derivation-by-construction circularity; empirical multi-modal correlations and standard CSD statistics stand independently, with only a minor non-load-bearing self-citation supporting the psychology application of CSD.

self citation load bearing [§1 Introduction, citation [14]]
"CSD theory has recently shown promise in predicting psychological transitions, including mood disorder episodes and suicide risk [14]."

Reference [14] is a concurrent arXiv preprint by the same lead author (Singley et al.). It supplies only soft precedent that CSD applies to psychological risk, not a uniqueness result or any numerical input to the van Gogh analysis. The actual CSD indicators and cross-modal correlations are computed from the letter/painting data and rest on external CSD citations (Scheffer, Dakos). Non-load-bearing; raises score only to 1.

full rationale

The paper’s central results are empirical computations on external data, not algebraic identities of fitted constants. Fractal dimensions are obtained from an independent image-processing pipeline (DoG → LIC → Otsu → Minkowski box-counting) on 569 paintings; sentiment scores come from an SVM trained on a separate modern social-media corpus and applied to letter paragraphs; Pearson r and rolling variance/lag-1 autocorrelation are then computed on the aligned series. None of these steps defines the output in terms of the claimed input. CSD theory is imported from Scheffer, Dakos and related external literature; the single overlapping-author citation ([14], Singley et al. on veteran suicide risk) is used only as supporting precedent that CSD has been applied to psychological transitions, not as a uniqueness theorem or as the source of the van Gogh statistics. Window-size selection (“systematically varied… optimal window of one month”) and event-anchored phase definitions (baseline / CSD / stabilization relative to known biographical dates) are researcher degrees of freedom that weaken prospective-detection claims, but they do not make the reported variance or autocorrelation equal to the inputs by construction. No parameter fitted to one quantity is renamed as a prediction of a closely related quantity. Therefore the derivation chain is self-contained against the circularity criteria; residual concerns belong to correctness/validity (domain transfer of the sentiment model, lack of surrogate nulls, post-hoc windowing), not circularity.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The load-bearing claim rests on transferring modern NLP clinical labels and ecological CSD indicators onto sparse historical letters and paintings, plus several analysis knobs (match window, rolling window, phase bins around known events). No new physical entity is postulated; the fragile parts are measurement validity and analysis degrees of freedom rather than novel ontology.

free parameters (5)

rolling_window_size = one month (chosen from 5–30 data-point sweep)
Windows swept over 5–30 points; authors select an ‘optimal’ one-month window for CSD detection, which directly shapes variance/autocorrelation spikes.
letter_painting_match_window = 30 days (mean separation 2.1 days after match)
30-day proximity used to build n=524 pairs; changes which sentiment values align to which fractal values and thus the reported r and p.
CSD_phase_bin_edges = 4–6 / 2–4 / 0–2 months before event
Baseline 4–6 mo, CSD 2–4 mo, stabilization 0–2 mo before events are hand-specified relative to known crises and control the ‘coordinated spike then stabilize’ narrative.
variance_comparison_windows = 90-day rolling variance; three biographical super-periods
90-day rolling variance and Pre-ear / Between / Pre-suicide period cuts determine Fligner-Killeen and box-plot contrasts.
image_processing_hyperparameters = not numerically fixed in text
DoG scales, LIC, Otsu threshold, and box-counting grid choices affect fractal dimension levels (e.g., Starry Night 1.93) and all downstream correlations.

assumptions (6)

domain assumption Critical slowing down indicators (rising variance, lag-1 autocorrelation, variance-of-variance) that warn of tipping points in ecological/climate systems also index approaching psychological crises in an individual.
Invoked from Scheffer/Dakos-style CSD literature and recent psych applications (§1, §2.4) as the interpretive frame for rolling statistics.
domain assumption Interpersonal theory of suicide constructs (desire for death, acquired capability, thwarted belongingness/isolation, etc.) are appropriately operationalized as SVM output scores on letter paragraphs.
Category list and Joiner/Van Orden framing in §2.2; scores treated as continuous psychological state variables.
ad hoc to paper A classifier trained on >100k modern social-media posts transfers to van Gogh’s 1872–1890 correspondence without destructive domain shift.
Stated training setup in §2.2; no period-matched validation is reported, yet all sentiment CSD claims depend on it.
domain assumption Minkowski–Bouligand dimension of DoG/LIC/Otsu-processed brushwork is a monotone-enough proxy for visual/emotional/cognitive arousal complexity.
§2.3 links higher fractal dimension to denser irregular structure and elevated arousal; used as the visual CSD channel.
ad hoc to paper Linear interpolation onto a common daily grid preserves correlation and autocorrelation structure enough for Pearson tests and lag-1 estimates.
Explicit resampling step under §3.1 correlation methods; can induce spurious smoothness and cross-series coupling.
standard math Standard SVM, Platt scaling, box-counting, and Fligner-Killeen procedures behave as in the methodological literature.
Routine statistical/ML tools cited via Platt, Falconer, SVM references.

how reviews work

0 comments

Cite this review

Pith. "Pith review of From Pen to Palette: Mathematical Analysis of van Gogh's Psychological Trajectory." pith.science (2026). https://pith.science/paper/O5B25YUH

@misc{pith2026260723457,
  author       = {Pith},
  title        = {Pith review of: From Pen to Palette: Mathematical Analysis of van Gogh's Psychological Trajectory},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O5B25YUH}},
  note         = {Machine review of arXiv:2607.23457}
}

abstract

Vincent van Gogh's prolific artistic output and personal correspondence offer a unique window into the psychological trajectory of a 19th-century artist. This study applies computational methods to analyze emotional patterns across van Gogh's final decade through sentiment analysis of over 700 letter segments and fractal dimension analysis of 569 paintings. Using critical slowing down (CSD) theory from dynamical systems, we examine whether van Gogh's psychological system exhibited mathematical signatures of approaching tipping points. Our analysis reveals a statistically significant correlation between visual complexity and death sentiment ($r = 0.193, p < 0.001, n = 524$), alongside evidence for CSD patterns in multiple psychological dimensions before major crises. The 1888 ear incident shows coordinated variance spikes across acquired capability, desire for death, and visual complexity preceding the event, followed by stabilization before the crisis. These findings demonstrate that mathematical approaches can detect psychological transition signatures in historical figures while revealing the complex multi-modal nature of psychological systems.

Figures

Figures reproduced from arXiv: 2607.23457 by the authors.

**Figure 2.** An example of the image processing pipeline employed to obtain the fractal dimension of brushwork in van [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** A series of temporal analysis plots of van Gogh’s psychological trajectory. (A) Visual complexity (fractal [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Historical analysis of psychological indicators aligned with major biographical events, demonstrating the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 5.** Figure 5: Autocorrelation (lag-1) plot of "Desire for Death" sentiment and the fractal dimension of paintings with a [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

**Figure 6.** Figure 6: Box plots showing the variance in Social Isolation, Desire for Death, and Acquired Capability sentiments [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

**Figure 6.1.** Figure 6.1: Violin plot of the fractal dimension of van Gogh paintings by location, showing variation in visual complexity [PITH_FULL_IMAGE:figures/full_fig_p011_6_1.png]

**Figure 6.2.** Figure 6.2: Violin plot of the fractal dimension of van Gogh works by type. [PITH_FULL_IMAGE:figures/full_fig_p011_6_2.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 2 linked inside Pith

[1]

When van gogh meets mandelbrot: Multifractal classification of painting’s texture.Signal Processing, 93(3):554–572, 2013

Patrice Abry, Herwig Wendt, and Stéphane Jaffard. When van gogh meets mandelbrot: Multifractal classification of painting’s texture.Signal Processing, 93(3):554–572, 2013

2013
[2]

Turbulent luminance in impassioned van gogh paintings.Journal of Mathematical Imaging and Vision, 30(3):275–283, 2008

José Luis Aragón, Gerardo G Naumis, Ming Bai, Manuel Torres, and Philip K Maini. Turbulent luminance in impassioned van gogh paintings.Journal of Mathematical Imaging and Vision, 30(3):275–283, 2008

2008
[3]

Color and emotion-a psychophysical analysis of van gogh’s work.PsyArt, 2009

Karen G Bekker and AY Bekker. Color and emotion-a psychophysical analysis of van gogh’s work.PsyArt, 2009

2009
[4]

On the appropriateness of platt scaling in classifier calibration.Information Systems, 95:101641, 2021

Björn Böken. On the appropriateness of platt scaling in classifier calibration.Information Systems, 95:101641, 2021

2021
[5]

Critical slowing down as early warning for the onset of collapse in mutualistic communities.Proceedings of the National Academy of Sciences, 111(49):17546–17551, 2014

Vasilis Dakos and Jordi Bascompte. Critical slowing down as early warning for the onset of collapse in mutualistic communities.Proceedings of the National Academy of Sciences, 111(49):17546–17551, 2014

2014
[6]

Robustness of variance and autocorrela- tion as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

Vasilis Dakos, Egbert H Van Nes, Paolo d’Odorico, and Marten Scheffer. Robustness of variance and autocorrela- tion as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

2012
[7]

John Wiley & Sons, 2013

Kenneth Falconer.Fractal geometry: mathematical foundations and applications. John Wiley & Sons, 2013

2013
[8]

The yellow house: Van gogh, gauguin, and nine turbulent weeks in arles.American Journal of Psychiatry, 163(12):2202–2202, 2006

Robert Freedman. The yellow house: Van gogh, gauguin, and nine turbulent weeks in arles.American Journal of Psychiatry, 163(12):2202–2202, 2006

2006

Show all 19 references

[9]

Vincent van gogh’s illness: acute intermittent porphyria?BMJ: British Medical Journal, 303(6817):1589, 1991

Loretta S Loftus and Wilfred Niels Arnold. Vincent van gogh’s illness: acute intermittent porphyria?BMJ: British Medical Journal, 303(6817):1589, 1991

1991
[10]

Fractal contours: Order, chaos, and art.Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(6), 2024

John McDonough and Andrzej Herczy´nski. Fractal contours: Order, chaos, and art.Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(6), 2024

2024
[11]

Critical slowing down indicators.Europhysics Letters, 132(1):18001, 2020

Fahimeh Nazarimehr, Sajad Jafari, Matjaž Perc, and Julien C Sprott. Critical slowing down indicators.Europhysics Letters, 132(1):18001, 2020

2020
[12]

Real-time optical image processing using difference-of-gaussians wavelets.Optical Engineering, 33(7):2296–2302, 1994

Ting-Chung Poon and Kwongchoi Caisy Ho. Real-time optical image processing using difference-of-gaussians wavelets.Optical Engineering, 33(7):2296–2302, 1994

1994
[13]

Early-warning signals for critical transitions.Nature, 461(7260):53–59, 2009

Marten Scheffer, Jordi Bascompte, William A Brock, Victor Brovkin, Stephen R Carpenter, Vasilis Dakos, Hermann Held, Egbert H Van Nes, Max Rietkerk, and George Sugihara. Early-warning signals for critical transitions.Nature, 461(7260):53–59, 2009

2009
[14]

A stochastic compartmental model of suicide risk dynamics in us veterans.arXiv preprint arXiv:2508.18553, 2025

Anna Singley, Carrie Manore, Hannah Highlander, and Ben McMahon. A stochastic compartmental model of suicide risk dynamics in us veterans.arXiv preprint arXiv:2508.18553, 2025

2025 arXiv
[15]

Theory of support vector machines.University of London, 117(827):188–191, 1996

MO Stitson, JAE Weston, A Gammerman, V V ovk, and V Vapnik. Theory of support vector machines.University of London, 117(827):188–191, 1996

1996
[16]

Radial basis function kernel optimization for support vector machine classifiers.arXiv preprint arXiv:2007.08233, 2020

Karl Thurnhofer-Hemsi, Ezequiel López-Rubio, Miguel A Molina-Cabello, and Kayvan Najarian. Radial basis function kernel optimization for support vector machine classifiers.arXiv preprint arXiv:2007.08233, 2020

2007 arXiv
[17]

Atheneum, 1963

Vincent Van Gogh.The letters of Vincent van Gogh. Atheneum, 1963

1963
[18]

Van gogh museum collection

Van Gogh Museum. Van gogh museum collection. Digital image database, 2024. Accessed: [2024]

2024
[19]

The interpersonal theory of suicide.Psychological review, 117(2):575, 2010

Kimberly A Van Orden, Tracy K Witte, Kelly C Cukrowicz, Scott R Braithwaite, Edward A Selby, and Thomas E Joiner Jr. The interpersonal theory of suicide.Psychological review, 117(2):575, 2010. 12

2010

Pith tools

Reviewed July 30, 2026 · model on record in the stance chip above.

[1] [1]

When van gogh meets mandelbrot: Multifractal classification of painting’s texture.Signal Processing, 93(3):554–572, 2013

Patrice Abry, Herwig Wendt, and Stéphane Jaffard. When van gogh meets mandelbrot: Multifractal classification of painting’s texture.Signal Processing, 93(3):554–572, 2013

2013

[2] [2]

Turbulent luminance in impassioned van gogh paintings.Journal of Mathematical Imaging and Vision, 30(3):275–283, 2008

José Luis Aragón, Gerardo G Naumis, Ming Bai, Manuel Torres, and Philip K Maini. Turbulent luminance in impassioned van gogh paintings.Journal of Mathematical Imaging and Vision, 30(3):275–283, 2008

2008

[3] [3]

Color and emotion-a psychophysical analysis of van gogh’s work.PsyArt, 2009

Karen G Bekker and AY Bekker. Color and emotion-a psychophysical analysis of van gogh’s work.PsyArt, 2009

2009

[4] [4]

On the appropriateness of platt scaling in classifier calibration.Information Systems, 95:101641, 2021

Björn Böken. On the appropriateness of platt scaling in classifier calibration.Information Systems, 95:101641, 2021

2021

[5] [5]

Critical slowing down as early warning for the onset of collapse in mutualistic communities.Proceedings of the National Academy of Sciences, 111(49):17546–17551, 2014

Vasilis Dakos and Jordi Bascompte. Critical slowing down as early warning for the onset of collapse in mutualistic communities.Proceedings of the National Academy of Sciences, 111(49):17546–17551, 2014

2014

[6] [6]

Robustness of variance and autocorrela- tion as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

Vasilis Dakos, Egbert H Van Nes, Paolo d’Odorico, and Marten Scheffer. Robustness of variance and autocorrela- tion as indicators of critical slowing down.Ecology, 93(2):264–271, 2012

2012

[7] [7]

John Wiley & Sons, 2013

Kenneth Falconer.Fractal geometry: mathematical foundations and applications. John Wiley & Sons, 2013

2013

[8] [8]

The yellow house: Van gogh, gauguin, and nine turbulent weeks in arles.American Journal of Psychiatry, 163(12):2202–2202, 2006

Robert Freedman. The yellow house: Van gogh, gauguin, and nine turbulent weeks in arles.American Journal of Psychiatry, 163(12):2202–2202, 2006

2006