{"id":"07992039-31e9-4356-be4f-e36f6eb732db","arxiv_id":"2411.16500","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The matter density PDF can be predicted in extended cosmologies with large deviations theory, and wave-based forward models capture phase-space dynamics beyond the fluid approximation.","lead":"This PhD thesis applies large deviations theory to predict the probability distribution of cosmic matter density in modified gravity and dark energy cosmologies, and uses wave-based forward models to study dark matter dynamics. A generalist might read it to see how non-Gaussian statistics could sharpen cosmological constraints from upcoming surveys like Euclid.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The LDT PDF claim rests on a calibrated non-linear variance input: Eq (5.23) is not first-principles, so 'accurate prediction' is partly inherited from simulation calibration rather than demonstrated.","rationale":"The reader's weakest_assumption correctly identifies the calibrated non-linear variance as the soft spot in the LDT PDF construction. I agree with that assessment. The thesis's central claim is that the matter PDF can be accurately predicted in extended cosmologies, but the non-linear variance in Eq (5.23) is either measured from the very simulations being compared or rescaled from a fiducial simulation using an empirical ansatz. This does not invalidate the work, because the calibrated model may still be a useful effective tool, but it does mean the word 'predicted' is too strong, and the complementarity statement is conditional on that calibration. I do not see a more load-bearing concern: the spherical collapse approximation is explicitly tested, and the wave-model claims in abstract item 3 are supported by detailed stationary-phase and catastrophe-theory analyses, though they are weaker than the PDF claim because they are more exploratory. The reader's CONDITIONAL verdict remains appropriate, and my stress-test does not move it.","tokens_in":56108,"tokens_out":4432,"duration_ms":49449,"concrete_test":"Hold out one of the modified-gravity simulation suites (e.g., the F5/f(R) suite) entirely. Compute sigma^2_lnrho for that cosmology using Eq (5.23) with the fiducial LambdaCDM variance taken from a genuinely independent source (e.g., Quijote or a separate LambdaCDM run), and compare the resulting pyLDT PDF to the F5 simulation measurements. If the residuals exceed the 2% target within |lnrho - <lnrho>| < 2sigma, then the prediction is calibration-dependent and the claimed accuracy is not self-contained.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5.3.3 (Eq 5.23) supplies the non-linear log-density variance sigma^2_lnrho by rescaling the fiducial cosmology's measured variance with the ratio of linear variances. The thesis explicitly acknowledges (Section 4.3.4) that this ingredient is either a free parameter or measured from simulations. Since the PDF width and normalization are controlled by this variance, the excellent agreement in Figures 5.2-5.3 does not establish a fully predictive theory: the target simulations are used, at minimum, to fix the fiducial non-linear variance, and then Eq (5.23) is asserted to transfer it to modified gravity and dark-energy models. The quoted 0.2-1% accuracy of Eq (5.23) is an a posteriori calibration on the same simulation suite being validated. The central claim of abstract item 1 ('accurately predicted') therefore holds only conditionally on this calibration. The spherical collapse approximation (5.20) is also a posteriori, but it is a smaller effect than the variance input for the PDF width. This is the most load-bearing weakness because it affects both the headline prediction claim and the Fisher forecasts in Sections 5.6.2-5.6.3: if the variance is biased, the forecasted complementarity to two-point statistics is not reliable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This thesis (arXiv:2411.16500) develops and tests non-Gaussian one-point statistics of the cosmic matter field, with large deviations theory (LDT) as the central tool. Chapter 5 extends the LDT matter PDF to extended cosmologies — Hu-Sawicki f(R) gravity, nDGP, and w0waCDM — using modified linear growth, an Einstein-de Sitter spherical-collapse mapping rescaled by linear variances, and a lognormal rescaling of the fiducial non-linear log-density variance; the resulting PDF is compared with dedicated N-body simulations and used in Fisher forecasts for a Euclid-like survey. Chapter 6 derives covariance matrices of the one-point PDF from a two-cell joint PDF, including super-sample covariance and analytic bias functions in a minimal tree model. Chapters 7 and 8 develop a Schrödinger-Poisson forward model with an effective Planck constant as an alternative closure of the dark-matter Vlasov hierarchy, analysing interference, caustics, and density statistics. The thesis is compiled from four published papers and one preprint and is explicit about the author's contributions.","tokens_in":56424,"tokens_out":8312,"duration_ms":81122,"significance":"If the central claims are borne out, the LDT matter PDF is a survey-ready statistic that is sensitive to modified gravity and evolving dark energy in ways complementary to the power spectrum, and the wave-based model provides a useful tool for multi-streaming dynamics and fuzzy dark matter statistics. The thesis has concrete strengths: the pyLDT code is released, comparisons are made to dedicated simulations (including multiple realisations for f(R)), the Fisher forecasts are transparent about scale cuts and covariance assumptions, and several technical derivations are placed in appendices, making the work reproducible. The main limitation is that the headline PDF prediction depends on calibrated ingredients, so the 'accurately predicted' claim is conditional until those ingredients are independently validated or the claim is appropriately qualified.","major_comments":[{"comment":"The width and normalisation of the LDT matter PDF are controlled by the non-linear log-density variance sigma^2_NL,ln(rho). In this chapter that variance is either measured from the very simulation suites used for validation, or obtained from Eq. (5.23), which rescales the fiducial cosmology's measured variance by the ratio of linear variances. The quoted 0.2-1% accuracy of Eq. (5.23) is an a posteriori calibration on the same DGP/f(R)/w0waCDM runs used to produce Figures 5.2-5.3. Therefore the agreement in those figures, and the Fisher forecasts in Sections 5.6.2-5.6.3, do not by themselves establish a first-principles prediction. I ask the authors to validate Eq. (5.23) on an independent simulation suite or with an independently calibrated model for sigma^2_NL,ln(rho), and to qualify the abstract claim accordingly, or to show that the Fisher results are insensitive to plausible errors in this input.","section":"Section 5.3.3, Eq. (5.23)"},{"comment":"The spherical-collapse ingredient is also asserted rather than derived. The approximation delta_ext_L(rho) approximately (sigma_L^Lambda / sigma_L^ext) delta_EdS_L(rho) is described in the text as justified a posteriori by comparison with simulations. For f(R) gravity specifically, mass conservation and shell crossing are violated, so no exact mapping is available; the validation is indirect, via the reduced cumulants S3 and S4 in Eq. (5.22), rather than a direct test of the mapping. Because delta_L(rho) enters the rate function in Eq. (4.19), an error in this mapping propagates directly into the PDF shape. A sensitivity test (for example, using the exact DGP mapping for DGP, or an explicitly screened f(R) mapping) would show whether the claimed agreement is robust; if the mapping is only valid in the mildly non-linear regime, that limitation should be stated in the abstract.","section":"Section 5.3.2, Eq. (5.20)"},{"comment":"The claim that the wave-based forward model 'can encode the full phase-space dynamics beyond a perfect fluid' is stronger than what is demonstrated. The Schrödinger-Poisson model with an effective Planck constant is an alternative closure ansatz; the thesis shows that it reduces to Zel'dovich trajectories in the stationary-phase limit, produces interference patterns, and yields universal scaling near caustics, but it does not establish equivalence to the full Vlasov hierarchy. The Wigner representation in Appendix D.5 is a formal rewriting of the Schrödinger equation, not a proof that the closure captures all multi-stream moments. I recommend either adding a comparison against a full Vlasov-Poisson solution in a multi-streaming configuration beyond the simple examples used, or softening the 'full phase-space dynamics' wording to 'a wave-mechanical closure reproducing several multi-streaming effects'.","section":"Abstract item 3 and Chapter 7"},{"comment":"The Fisher forecasts assume that the joint covariance of the PDF and power spectrum is independent of cosmology and gravity theory and is equal to the Quijote LambdaCDM covariance. The text acknowledges this and argues that the induced error is small because sigma8 changes by only 1.6% (F6) and 3.8% (DGPw). This is a reasonable first approximation, but the headline improvements (up to a factor of six in detection significance) are conditional on it. I would like to see a robustness check in which the covariance is rescaled by the sigma8-induced variance change, or a simple two-parameter covariance model is varied, demonstrating that the predicted complementarity is not an artifact of using the LambdaCDM covariance.","section":"Section 5.6.1"}],"minor_comments":[{"comment":"The heading 'Forecasting constrating power with the Fisher formalism' contains a typo; it should read 'constraining power'.","section":"Section 5.6 heading"},{"comment":"The statement that dark energy refers to 'any component with equation of state w < 1/3' should read w < -1/3; the current inequality would include ordinary matter, which does not accelerate the expansion.","section":"Section 2.4"},{"comment":"The square-root symbols in these equations render as '/radicaltp/radicalvertex' artifacts in the present version; the final typeset version must use proper radical notation.","section":"Equations (4.29) and (5.18)"},{"comment":"The sentence '...potentially complemented with predictions for effects induced by variations in the local mean density (Jamie' is incomplete; the accompanying citation appears to be cut off and should be restored.","section":"Section 5.6.1"}],"recommendation":"major_revision","confidential_remarks":"The thesis is an honest and mostly well-organised compilation of the author's papers, and the statement of contributions is exemplary. The main risk for publication is not novelty but the conditional nature of the headline PDF prediction, which is already disclosed in Section 4.3.4 and Section 5.3.3. I would advise the editor to insist on the additional validation described in Major Comments 1 and 2, or a corresponding qualification of the abstract claims, before considering the paper as a research article."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my read. This is essentially a thesis by publication. Chapters 5-8 are versions of Cataneo et al. 2022, Gough & Uhlemann 2022a/2022b/2024, and Uhlemann et al. 2023. If you have followed that literature, you will not find new equations here. The value is in the packaging and the pedagogy: the background chapters and the LDT formalism chapter are unusually clear, and the appendices contain real derivations, such as matching the spherical collapse index to skewness, the minimal tree model, and the Wigner equation. The pyLDT code is public, and the claims are checked against N-body simulations, including f(R) and DGP suites.\n\nThe central scientific claim—that the matter PDF can be accurately predicted in modified gravity and dynamical dark energy and adds constraining power beyond the power spectrum—is defensible, but it is not fully first-principles. The non-linear variance of the log-density, which controls the PDF width, is either measured from the same simulations or approximated by Eq (5.23), a lognormal rescaling calibrated to a fiducial cosmology. The thesis is honest about this: Section 4.3.4 calls it a free parameter or measured input. That means the 'prediction' inherits calibration. The claim is still useful for survey forecasts, but it should be stated as a calibrated model, not a derivation. The same applies to the spherical collapse rescaling in Eq (5.20); the thesis says it is justified a posteriori.\n\nI also have a moderate concern about the Fisher forecasts: the covariance matrix is assumed cosmology-independent and taken from Quijote. That is a conventional choice, but for modified gravity it is an assumption, and Section 5.6 acknowledges the resulting uncertainty only in passing.\n\nWhat is genuinely good: the LDT derivation in Chapter 4 recovers the tree-level skewness and is presented more carefully than in the original papers; the covariance work in Chapter 6 is a nice extension; and the wave-based forward model chapters are honest about their limitations. The thesis is a useful compendium for a graduate student entering this area.\n\nMy verdict: as a journal submission, I would send it to review, because the underlying work is peer-reviewed and clearly presented, but I would make the authors specify what this version adds. I would not cite the thesis itself; I would cite the underlying papers. Bring it to reading group if someone wants to learn LDT.","headline":"A solid, well-written PhD thesis that packages four published papers and one preprint; no new results beyond those papers, but a clear and honest entry point to LDT-based matter PDFs.","tokens_in":56907,"tokens_out":2476,"would_cite":false,"duration_ms":25520,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This thesis argues that the one-point probability distribution of the cosmic matter density, predicted by large deviations theory from Gaussian initial conditions and spherical collapse, stays accurate in modified gravity and dynamical…","keywords":["large deviations theory","matter density PDF","modified gravity","dark energy","cosmic structure formation","wave dark matter","one-point statistics","phase-space dynamics"],"falsifier":"Take an f(R) or w0waCDM cosmology, compute the LDT PDF using only the lognormal-rescaled non-linear variance, and compare it against N-body measurements on $10\\,h^{-1}\\,\\mathrm{Mpc}$ spheres at $z=0$ and $z=1$; if residuals exceed a few percent within two log-density standard deviations, the variance input is inadequate.","tokens_in":42,"feed_emoji":"🌌","tokens_out":8349,"duration_ms":138882,"temperature":0.7,"pith_summary":"The thesis advances a practical claim: the smoothed matter density PDF is a survey-ready statistic, not just a theoretical curiosity. Using large deviations theory, it predicts the density PDF in modified gravity and evolving dark energy models to within a few percent, and Fisher forecasts show the PDF halves dark-energy parameter uncertainties and boosts modified-gravity detection significance by up to six times relative to the power spectrum alone. A second thread derives one-point PDF covariances from the joint two-cell PDF, recovering super-sample covariance that simulation boxes miss. A third thread uses a wavefunction forward model of dark matter to encode full phase-space dynamics beyond a perfect fluid, showing that interference patterns decompose into classical multi-stream trajectories and that caustics obey universal scaling. Together these results position non-Gaussian one-point statistics as a complement to two-point analyses for current and upcoming surveys.","feed_headline":"Matter density PDF halves dark energy errors in survey forecasts","feed_subtitle":"Adding the density PDF to power-spectrum forecasts cuts dark-energy errors in half and lifts modified-gravity detection by up to six times.","key_machinery":"The load-bearing object is the large-deviations decay-rate function $\\psi_\\rho(\\rho) = \\tfrac{1}{2}\\,\\delta_L(\\rho)^2\\,\\sigma_L^2(R,z)/\\sigma_L^2(R\\rho^{1/3},z)$, built from the Gaussian rate function for linear densities and contracted through the spherical-collapse mapping $F:\\delta_L\\mapsto\\rho$; the log-density $\\mu=\\ln\\rho$ is used as the transformed variable to extend convexity, and the PDF is recovered through a saddle-point approximation to the inverse Laplace transform. Its three ingredients are the linear variance $\\sigma_L^2$, the spherical-collapse mapping (parametrised as $(1-\\delta_L/\\nu_{\\mathrm{SC}})^{-\\nu_{\\mathrm{SC}}}$ in Einstein-de Sitter), and the non-linear log-density variance $\\sigma^2_{\\mathrm{NL},\\ln\\rho}$. For the dynamics chapters, the central mechanism is the Schrödinger wavefunction forward model whose stationary-phase 'unweaving' separates the wavefunction into stream components, with catastrophe theory providing universal scaling near caustics.","core_discovery":"The central claim is that non-Gaussian one-point statistics of the cosmic density field can be predicted accurately enough to serve as cosmological probes. For the matter PDF, the large-deviations construction—Gaussian linear density, spherical-collapse mapping, and linear variance—extends to modified gravity and dynamical dark energy with two substitutions: the Einstein-de Sitter spherical collapse mapping rescaled by the ratio of linear variances, and a lognormal rescaling for the non-linear log-density variance. The thesis validates this against N-body simulations for f(R) gravity, DGP gravity, and w0waCDM on $10\\,h^{-1}\\,\\mathrm{Mpc}$ spheres, and shows that the PDF adds information to the matter power spectrum in Fisher forecasts. For dynamics, it claims a wavefunction forward model encodes the full Vlasov phase-space behaviour beyond the perfect-fluid closure, with interference patterns unwoven into streams and universal scaling near caustics. For covariances, it claims the joint two-cell PDF predicts the one-point PDF covariance, including density-dependent clustering and super-sample covariance.","pith_inferences":["Editorial inference: the same two-cell covariance machinery that predicts PDF covariances could be extended to build covariance matrices for counts-in-cells and weak-lensing PDFs in real surveys, where simulation-only covariances are prohibitively expensive.","Editorial inference: if the universal caustic scalings of the wave model hold generally, they give an analytic handle on interference statistics in ultralight axion dark matter without resolving the full wavefunction numerically.","Editorial inference: the weakest link in the PDF programme is the non-linear variance input; a first-principles calibration predicting this variance from the linear power spectrum would make the PDF a fully predictive cosmological probe, but this thesis does not establish such a calibration."],"forward_implications":["The matter PDF can be combined with the power spectrum in Fisher forecasts for Euclid-like survey volumes, halving the uncertainty on evolving dark energy parameters.","The PDF increases the detection significance of departures from general relativity by up to six times compared with the power spectrum alone in models like F6 and DGPw.","One-point PDF covariances, including super-sample covariance, can be obtained analytically from effective two-point PDF models, correcting a limitation of simulation-based covariances.","The wave-based forward model captures multi-streaming phase-space dynamics that a perfect-fluid closure misses, and near caustics its statistics display universal scaling.","Density statistics from wave dark matter forward models separate initial-condition effects from dynamical effects, sharpening predictions for wavelike dark matter candidates."],"supporting_citations":[{"why":"It supplies the N-body simulations and validation for the LDT PDF in f(R), DGP and w0waCDM cosmologies presented in Chapter 5.","marker":"Cataneo et al. (2022)"},{"why":"It introduces the log-density transform that extends the LDT saddle-point PDF to a wider range of densities.","marker":"Uhlemann et al. (2016)"},{"why":"It lays out the modern large-deviations formalism for the cosmic density PDF that the thesis applies.","marker":"Bernardeau and Reimberg (2016)"},{"why":"It provides the effective two-cell PDF model from which Chapter 6 derives one-point PDF covariances.","marker":"Uhlemann et al. (2023)"},{"why":"It presents the wave-based forward model and the stationary-phase 'unweaving' of interference central to Chapter 7.","marker":"Gough and Uhlemann (2022a)"},{"why":"It is the submitted work underlying Chapter 8 on density statistics in wave dark matter.","marker":"Gough and Uhlemann (2024)"}],"fun_headline_variants":["Matter PDF doubles dark energy precision in forecasts","Wave model unlocks full phase-space of dark matter","Joint PDF forecasts super-sample covariance accurately","Non-Gaussian stats sharpen modified-gravity detection","Cosmic PDF emerges as new lens on dark energy"],"cache_read_input_tokens":59008,"weakest_assumption_plain":"The prediction stands or falls on the non-linear variance of the log-density, which the theory does not derive from first principles: it must be measured from the same simulations being compared or approximated by a lognormal rescaling calibrated to a fiducial cosmology.","fun_headline_variants_meta":{"raw":{"variants":["Matter PDF doubles dark energy precision in forecasts","Wave model unlocks full phase-space of dark matter","Joint PDF forecasts super-sample covariance accurately","Non-Gaussian stats sharpen modified-gravity detection","Cosmic PDF emerges as new lens on dark energy"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000198,"raw_usage":{"total_tokens":1421,"prompt_tokens":1049,"completion_tokens":372,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":665,"completion_tokens_details":{"reasoning_tokens":302}},"tokens_in":665,"tokens_out":372,"duration_ms":4556,"temperature":1.0,"reasoning_tokens":302,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:02:43.264166+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an f(R) or w0waCDM cosmology, compute the LDT PDF using only the lognormal-rescaled non-linear variance, and compare it against N-body measurements on $10\\,h^{-1}\\,\\mathrm{Mpc}$ spheres at $z=0$ and $z=1$; if residuals exceed a few percent within two log-density standard deviations, the variance input is inadequate.","supporting_citations":[],"review_version":1}