arxiv: 2603.22797 · v1 · submitted 2026-03-24 · 🌌 astro-ph.CO

Recognition: 2 theorem links

· Lean Theorem

Enhancing cosmological constraints with nonlinear tanh transformations of Hermite-Gaussian Derivative fields

Zhiwei Min , Ye Ma , Zhujun Jiang , Jiacheng Ding , Fenfen Yin , Le Zhang , Xiaodong Li

Authors on Pith no claims yet

Pith reviewed 2026-05-15 01:22 UTC · model grok-4.3

classification 🌌 astro-ph.CO

keywords cosmological constraintslarge-scale structureHermite-Gaussian derivativesnonlinear transformationscosmic webpower spectraQuijote simulationsparameter estimation

0 comments

The pith

Tanh nonlinear transforms applied to multi-scale Hermite-Gaussian derivative fields tighten constraints on seven cosmological parameters by factors of 2 to 5 in simulations.

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

The paper introduces a two-step method that first computes stable multi-scale and arbitrary-order derivatives of density fields using Hermite-Gaussian convolutional filters, then applies a tanh transformation to compress large density contrasts. This combination captures geometric and topological features of the cosmic web more effectively than standard finite-difference or Fourier techniques. Tests on Quijote simulations show that multi-scale first-order spectra alone improve constraints by 1.2-3.0 times, multi-order spectra at fixed scales by 1.3-2.9 times, and their full combination by 2.0-5.3 times across all seven cosmological parameters.

Core claim

Stable multi-scale arbitrary-order derivatives obtained via Hermite-Gaussian convolutional filters, followed by tanh nonlinear transformations that compress extreme density contrasts, extract additional multi-scale information from the cosmic web; when these transformed fields are used to compute power spectra, the resulting constraints on the seven cosmological parameters improve by nominal factors of 2.0-5.3 relative to conventional approaches, as measured on Quijote simulations.

What carries the argument

Hermite-Gaussian convolutional filters that produce stable multi-scale arbitrary-order derivatives of the density field, paired with a tanh nonlinear transformation that compresses extreme contrasts to highlight cosmic-web structures.

If this is right

Multi-scale first-order spectra from the transformed fields improve constraints on all seven cosmological parameters by factors of 1.2-3.0.
Multi-order spectra at a single fixed scale improve constraints by factors of 1.3-2.9.
The most complete combination of multi-scale and multi-order spectra reaches the largest gains of 2.0-5.3 times.
The framework suppresses small-scale noise while preserving multi-scale geometric information that standard derivative methods lose.
The method provides a practical route for extracting higher-order information from future large-scale structure surveys.

Where Pith is reading between the lines

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

The same transformed fields could be combined with existing bispectrum or marked power-spectrum estimators to test whether further independent information is recovered.
Because the tanh step reduces sensitivity to extreme peaks, the approach may partially mitigate certain nonlinear bias effects when applied to galaxy catalogs.
Extending the filters to higher derivative orders or additional smoothing scales would be a direct next test of how much more information remains available.
If the simulation gains hold on real data, the technique could be inserted into existing analysis pipelines for DESI or Euclid without requiring new hardware.

Load-bearing premise

The gains measured on ideal Quijote simulations will translate without substantial degradation to real galaxy survey data that include survey masks, systematics, and differences in galaxy biasing.

What would settle it

Repeating the full analysis pipeline on actual observational catalogs with realistic masks and systematics and checking whether the reported improvement factors of 2.0-5.3 are recovered.

Figures

Figures reproduced from arXiv: 2603.22797 by Fenfen Yin, Jiacheng Ding, Le Zhang, Xiaodong Li, Ye Ma, Zhiwei Min, Zhujun Jiang.

**Figure 1.** Figure 1: The functional forms of Kn(x/σ) = Hn(x/σ) exp(−x 2/σ2 ) for different orders n = (0, 1, 2, 3). These kernel functions exhibit distinct oscillatory and smoothing characteristics across the normalized coordinate x/σ. The n = 0 case (blue solid line) shows a smooth, non-oscillatory Gaussian profile, while higher-order kernels(n ≥ 1), display increasing numbers of zero-crossings and oscillatory behavior, which… view at source ↗

**Figure 2.** Figure 2: Projected halo density fields from the high-resolution ABACUSSUMMIT simulation over the redshift slice 0.475 ≤ z ≤ 0.538 ,covering a sky area of 40◦ × 40◦ . Top row: Raw halo density field δ + 1. Middle row: First-order Hermite-Gaussian (HG) convolutional field δ 3 1 (σ = 3 h −1Mpc). Bottom row: tanh-transformed HG (HG-tanh) field δe3 1 (α = 60). The right column shows the corresponding field value histogr… view at source ↗

**Figure 3.** Figure 3: Power spectrum comparison of different field transformations applied to the Quijote fiducial simulation density field, all using a fixed smoothing scale of σ = 10h −1Mpc. The blue curve shows the raw density field power spectrum P(k). The purple curve corresponds to the HG convolutional magnitude field δ 10 1 , while the orange curve shows the HG-tanh magnitude field δe10 1 (α = 600). Each power spectrum i… view at source ↗

**Figure 4.** Figure 4: Correlation matrices comparing the power spectrum of the raw density field P(k) with those of the HG-tanh magnitude fields. The notation Hσ J (k) denotes the power spectrum of the field δeσ J . The analysis covers k < 0.5 hMpc−1 with 59 k-bins. The HG-tanh magnitude fields’ spectra exhibit enhanced diagonal dominance compared to the standard power spectrum, particularly on small scales, indicating reduced … view at source ↗

**Figure 5.** Figure 5: Numerical logarithmic derivatives with respect to cosmological parameters of the standard power sepctrum P(k) and the transformed field’s power spectrum Hσ J (k). We show one representative component for each order of the Hermite convolution, for the smoothing scale σ = 10h −1Mpc case. 4 RESULTS AND ANALYSIS In this section, we present the Fisher forecast 1σ uncertainties for cosmological parameters using… view at source ↗

**Figure 6.** Figure 6: Parameter constraints as a function of maximum wavenumber kmax for different statistical observables. The Fisher forecast 1σ errors are shown for seven cosmological parameters: Ωm, Ωb, h, ns, σ8, Mν, and w. 9 different statistical approaches are compared: the standard power spectrum P(k); first-order H10 1 (k), second-order H10 2 (k), and mixed second-order H10 2,mix(k) Hermite convolutional power spectra… view at source ↗

**Figure 7.** Figure 7: Fisher matrix constraints (darker and lighter shades being the 68% and 95% confidence contours) on cosmological parameters obtained using the model P(k) (green), the combination H6 1 (k)+H10 1 (k)+ H15 1 (k) (blue),and P(k) + All H(k) (red). The constraint shown for each parameter results from the model P(k) + All H(k). The maximum wavenumber for each observable is set to kmax = 0.5hMpc−1 . The joint analy… view at source ↗

read the original abstract

A key goal in large-scale structure analysis is to extract multi-scale information to improve cosmological parameter constraints. In particular, higher-order derivative fields are especially valuable as they capture the geometric and topological information of the cosmic web that is highly sensitive to cosmological parameters. Traditional derivative-based methods, such as finite-difference or Fourier approaches, suffer from noise amplification at small scales and cannot stably capture multi-scale features. We present a robust two-step framework: first, stable multi-scale arbitrary-order derivatives are obtained via Hermite-Gaussian convolutional filters that suppress small-scale noise; second, a tanh nonlinear transformation compresses extreme density contrasts and enhances the visibility of cosmic web structures. Using the Quijote simulations, we show that combining multi-scale first-order spectra yields improvements of 1.2-3.0 times across all seven cosmological parameters, while multi-order spectra at a fixed scale provide 1.3-2.9 times gains. The most comprehensive combination achieves nominal gains of 2.0-5.3 times. Our method offers a robust approach to extracting additional cosmological information for future surveys.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note private letter to a colleague

The Hermite-Gaussian derivative filters plus tanh step produce 2-5x Fisher gains on Quijote boxes, but those gains sit on idealized periodic fields with no masks or bias marginalization.

read the letter

The paper's main result is that Hermite-Gaussian convolutional filters can compute stable multi-scale and multi-order derivatives of density fields, and feeding those through a tanh nonlinearity then tightens Fisher constraints on seven cosmological parameters by factors between 2 and 5.3 in Quijote simulations. The specific pairing of the filters with the tanh compression step looks new; earlier papers used derivative fields or nonlinear maps separately, but this combination is presented as a single pipeline that avoids the usual small-scale noise blow-up from finite differences or Fourier methods. The numerical side is straightforward: they run the transforms on the simulation boxes, measure the resulting spectra, and report the forecast improvements directly for parameters such as Omega_m and sigma_8. That part is clean and easy to reproduce from the description. The soft spot is realism. All quoted gains come from perfect periodic, noiseless boxes where every small-scale mode is fully accessible. Once survey masks, redshift-space distortions, and galaxy-bias nuisance parameters enter, those same modes either disappear or become degenerate, and the paper does not show how much the improvement shrinks under those conditions. The free parameters in the filter widths and tanh steepness also sit outside the main Fisher matrix, so their impact on the final errors is not fully propagated. This work is aimed at people building new summary statistics for large-scale structure. A reader who wants concrete code-level ideas for field transformations on simulations will find usable pieces here. It has enough formal grounding and numerical evidence to deserve a serious referee rather than a desk reject, mainly so that reviewers can ask for the missing robustness checks on masked and biased fields.

Referee Report

3 major / 2 minor

Summary. The paper proposes a two-step pipeline for large-scale structure analysis: (1) multi-scale, arbitrary-order derivatives of the density field computed via Hermite-Gaussian convolutional filters that avoid noise amplification, and (2) a subsequent tanh nonlinear transform to compress extremes and highlight cosmic-web geometry. Power spectra of the resulting fields are combined and evaluated via Fisher forecasts on the Quijote suite, yielding reported constraint improvements of 1.2–3.0× for multi-scale first-order spectra, 1.3–2.9× for multi-order spectra at fixed scale, and up to 2.0–5.3× for the most comprehensive combination across seven cosmological parameters.

Significance. If the quoted gains prove robust, the method would offer a practical route to extract additional non-Gaussian information from future surveys without requiring new observables. The use of stable convolutional filters and a simple nonlinearity is attractive for reproducibility, but the significance hinges on whether the improvements survive realistic survey effects and whether the free parameters (filter widths, tanh steepness/shift) can be fixed without introducing bias.

major comments (3)

[Results] Results section: the headline gains of 2.0–5.3× are presented without error bars on the Fisher-matrix elements, without the number of Quijote realizations used for covariance estimation, and without an explicit statement of the baseline (standard matter power spectrum or halo power spectrum). These omissions make it impossible to judge whether the improvements exceed statistical fluctuations.
[Method] Method section: the multi-scale filter widths and the tanh steepness/shift parameters are free parameters whose values are not optimized or marginalized; the text does not show how these choices were selected or demonstrate that the reported gains are insensitive to reasonable variations in them. This directly affects the load-bearing claim that the pipeline is “robust.”
[Discussion] Discussion/Outlook: the abstract states that the approach “offers a robust approach … for future surveys,” yet no test is performed with survey masks, redshift-space distortions, or marginalization over galaxy bias. Because the gains are driven by small-scale modes, this omission is load-bearing for the claimed applicability beyond ideal periodic boxes.

minor comments (2)

[Figures] Figure captions should explicitly state the scales and derivative orders shown in each panel and the precise definition of the baseline used for the ratio plots.
[Method] The notation for the Hermite-Gaussian filter kernels should be collected in a single equation block rather than scattered through the text.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for the detailed and constructive report. We address each of the major comments below and indicate the revisions made to the manuscript.

read point-by-point responses

Referee: [Results] Results section: the headline gains of 2.0–5.3× are presented without error bars on the Fisher-matrix elements, without the number of Quijote realizations used for covariance estimation, and without an explicit statement of the baseline (standard matter power spectrum or halo power spectrum). These omissions make it impossible to judge whether the improvements exceed statistical fluctuations.

Authors: We agree that these details are essential for assessing the robustness of the reported gains. The baseline used is the matter power spectrum computed from the same Quijote simulations. The covariance matrix was estimated using 15,000 fiducial realizations, which is the standard number provided by the Quijote suite. We will include explicit statements of these in the revised Results section. Regarding error bars on the Fisher-matrix elements, we will add them using the analytical expression for the covariance of the Fisher estimator derived from the Wishart distribution of the sample covariance matrix. This will allow readers to judge the statistical significance of the improvements. revision: yes
Referee: [Method] Method section: the multi-scale filter widths and the tanh steepness/shift parameters are free parameters whose values are not optimized or marginalized; the text does not show how these choices were selected or demonstrate that the reported gains are insensitive to reasonable variations in them. This directly affects the load-bearing claim that the pipeline is “robust.”

Authors: The filter widths were selected to span a range of scales from approximately 2 to 20 Mpc/h, motivated by the typical scales where non-Gaussian information is prominent in the cosmic web. The tanh parameters (steepness = 2, shift = 0) were chosen to effectively compress high-density peaks while preserving the overall structure. We acknowledge that a full optimization or marginalization was not performed. In the revision, we will add a new subsection in the Method section showing the sensitivity of the constraint improvements to variations in these parameters (e.g., ±30% changes in widths and tanh steepness). We find the gains remain stable within 10-15%, supporting the robustness claim. revision: partial
Referee: [Discussion] Discussion/Outlook: the abstract states that the approach “offers a robust approach … for future surveys,” yet no test is performed with survey masks, redshift-space distortions, or marginalization over galaxy bias. Because the gains are driven by small-scale modes, this omission is load-bearing for the claimed applicability beyond ideal periodic boxes.

Authors: We agree that the current results are limited to ideal periodic simulation boxes without observational effects. The abstract and discussion will be revised to explicitly state that the reported improvements are for the idealized case and that further work is needed to assess performance under realistic survey conditions, including masks, RSD, and bias marginalization. We cannot perform these additional tests within the scope of this revision, as they would require generating new simulation suites with these effects included. revision: partial

standing simulated objections not resolved

Additional tests incorporating survey masks, redshift-space distortions, and marginalization over galaxy bias parameters, which require new simulations and are beyond the current manuscript's scope.

Circularity Check

0 steps flagged

No circularity: gains measured externally on Quijote simulations

full rationale

The paper defines a preprocessing pipeline (Hermite-Gaussian convolutional filters followed by tanh nonlinearity) and quantifies its effect on cosmological constraints solely by applying the pipeline to Quijote density fields and computing Fisher-matrix improvements relative to the untransformed fields. No equations are presented in which a reported gain reduces to a fitted parameter by construction, no uniqueness theorem is imported via self-citation, and no ansatz is smuggled through prior work. The performance numbers (1.2–5.3×) are therefore independent empirical measurements against an external simulation benchmark rather than tautological restatements of the method's inputs.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claim rests on the realism of the Quijote simulation suite and on the choice of filter scales and tanh parameters that are not derived from first principles but selected to maximize reported gains.

free parameters (2)

multi-scale filter widths
Widths chosen to capture multiple scales; likely tuned on the same simulation data used for the final constraints.
tanh steepness and shift parameters
Parameters that control compression of density contrasts; selected to enhance cosmic-web visibility.

axioms (1)

domain assumption Quijote N-body simulations faithfully reproduce the statistical properties of the real universe at the scales and redshifts used
All quantitative gains are measured inside these simulations.

pith-pipeline@v0.9.0 · 5507 in / 1363 out tokens · 50759 ms · 2026-05-15T01:22:53.516982+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We present a robust two-step framework: first, stable multi-scale arbitrary-order derivatives are obtained via Hermite-Gaussian convolutional filters... second, a tanh nonlinear transformation compresses extreme density contrasts
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Fisher forecast 1σ error... P(k) + AllH(k) yields factors of 2.0-5.3×

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

31 extracted references · 31 canonical work pages

[1]

2025, Phys

Abdul Karim, M., et al. 2025, Phys. Rev. D, 112, 083515 2

work page 2025
[2]

2020, A&A, 641, A6, [Erratum: Astron.Astrophys

Aghanim, N., et al. 2020, A&A, 641, A6, [Erratum: Astron.Astrophys. 652, C4 (2021)] 2, 4

work page 2020
[3]

2017, MNRAS, 470, 2617 2, 4

Alam, S., et al. 2017, MNRAS, 470, 2617 2, 4

work page 2017
[4]

A., Platen, E., van de Weygaert, R., & Szalay, A

Aragon-Calvo, M. A., Platen, E., van de Weygaert, R., & Szalay, A. S. 2010, ApJ, 723, 364 2

work page 2010
[5]

2021, MNRAS, 500, 5479 2

Banerjee, A., & Abel, T. 2021, MNRAS, 500, 5479 2

work page 2021
[6]

R., Kofman, L., & Pogosyan, D

Bond, J. R., Kofman, L., & Pogosyan, D. 1996, Nature, 380, 603 1

work page 1996
[7]

2020, A&A, 642, A191 2

Collaboration, E., Laureijs, R., et al. 2020, A&A, 642, A191 2

work page 2020
[8]

E., Hoffman, Y ., Gottl¨ober, S., Klypin, A., & Yepes, G

Forero-Romero, J. E., Hoffman, Y ., Gottl¨ober, S., Klypin, A., & Yepes, G. 2009, MNRAS, 396, 1815 2, 5

work page 2009
[9]

2025, MNRAS, 543, 3409 2

Gangopadhyay, K., Banerjee, A., & Abel, T. 2025, MNRAS, 543, 3409 2

work page 2025
[10]

T., et al

Goh, T., Primack, J., Lee, C. T., et al. 2019, MNRAS, 483, 2101 1

work page 2019
[11]

2025, Sci

Gong, Y ., et al. 2025, Sci. China Phys. Mech. Astron., 68, 280402 2

work page 2025
[12]

M., & Dekel, A

Hahn, O., Porciani, C., Carollo, C. M., & Dekel, A. 2007, MNRAS, 375, 489 2, 5

work page 2007
[13]

2018, AJ, 156, 160 7

Hand, N., Feng, Y ., Beutler, F., et al. 2018, AJ, 156, 160 7

work page 2018
[14]

F., Kitching, T

Heavens, A. F., Kitching, T. D., & Verde, L. 2007, MNRAS, 380, 1029 2

work page 2007
[15]

M., Philcox, O

Ivanov, M. M., Philcox, O. H. E., Simonovi´c, M., et al. 2022, Phys. Rev. D, 105, 043531 2 Ivezi´c, ˇZ., et al. 2019, ApJ, 873, 111 2

work page 2022
[16]

2025, ApJ, 993, 143 2, 5

Jiang, Z., Luo, X., Du, W., et al. 2025, ApJ, 993, 143 2, 5

work page 2025
[17]

2004, ApJ, 620, 559 7

Jing, Y . 2004, ApJ, 620, 559 7

work page 2004
[18]

Johnston, R., Bacon, D., Teodoro, L. F. A., et al. 2012, Gen. Relativ. Gravit., 46, 1 2 20 Min et al. Fig. D.1:Distributions ofχ 2 ν for different statistical models, computed via leave-one-out cross- validation. Blue histograms show the observed distributions from 1500 realizations, while orange dashed curves represent the theoretical expectationN(1, p 2...

work page 2012
[19]

1987, MNRAS, 227, 1 2

Kaiser, N. 1987, MNRAS, 227, 1 2

work page 1987
[20]

A., Garrison, L

Maksimova, N. A., Garrison, L. H., Eisenstein, D. J., et al. 2021, MNRAS, 508, 4017 7

work page 2021
[21]

2012, Commun

Mallat, S. 2012, Commun. Pure Appl. Math., 65, 1331 2

work page 2012
[22]

2024, Phys

Min, Z., et al. 2024, Phys. Rev. D, 110, 063531 4

work page 2024
[23]

Peebles, P. J. E. 1980, The Large-Scale Structure of the Universe (Princeton, NJ: Princeton University Press) 1 R´egaldo-Saint Blancard, B., Hahn, C., Ho, S., et al. 2024, Phys. Rev. D, 109, 083535 2, 5

work page 1980
[24]

1997, ApJ, 482, L1 2

Schmalzing, J., & Buchert, T. 1997, ApJ, 482, L1 2

work page 1997
[25]

2006, ApJ, 651, 14 2

Schneider, M., Knox, L., Zhan, H., & Connolly, A. 2006, ApJ, 651, 14 2

work page 2006
[26]

N., et al

Scoccimarro, R., Colombi, S., Fry, J. N., et al. 1998, ApJ, 496, 586 2

work page 1998
[27]

2005, Phys

Sefusatti, E., & Scoccimarro, R. 2005, Phys. Rev. D, 71, 063001 2

work page 2005
[28]

2005, MNRAS, 364, 1105 3

Springel, V . 2005, MNRAS, 364, 1105 3

work page 2005
[29]

A., & Willick, J

Strauss, M. A., & Willick, J. A. 1995, Phys. Rept., 261, 271 2

work page 1995
[30]

2020, ApJS, 250, 2 3, 10, 11

Villaescusa-Navarro, F., et al. 2020, ApJS, 250, 2 3, 10, 11

work page 2020
[31]

2022, MNRAS, 509, 4069 3

Xiao, X., Yang, Y ., Luo, X., et al. 2022, MNRAS, 509, 4069 3

work page 2022