Pith. sign in

REVIEW 4 major objections 4 minor 6 cited by

DES Y6 calibrates the MagLim++ lens sample's n(z) to 1-2 percent mean accuracy (20-30 percent error reduction over Y3) and compresses the uncertainty into three modes per bin for cosmology.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

DES Y6 calibrates MagLim++ lens galaxy redshift distributions to 1-2% mean accuracy using SOM-based photometric redshifts plus clustering redshifts, with uncertainties compressed into a few modes.

T0 review reviewed 2026-08-04 challenge →

load-bearing objection A careful, unusually honest lens-redshift calibration for the DES Y6 pipeline; the central argument holds, with one legitimate robustness question about catastrophic photo-z outliers worth pushing a referee on. the 4 major comments →

arxiv 2509.07964 v1 pith:F7AFDWH6 submitted 2025-09-09 astro-ph.CO

Dark Energy Survey Year 6 Results: Redshift Calibration of the MagLim++ Lens Sample

G. Giannini , A. Alarcon , W. d'Assignies , G. M. Bernstein , M. A. Troxel , C. Chang , B. Yin , A. Amon
show 80 more authors
J. Myles N. Weaverdyck A. Porredon D. Anbajagane S. Avila K. Bechtol M. R. Becker J. Blazek M. Crocce D. Gruen M. Rodriguez-Monroy C. S\'anchez D. Sanchez Cid I. Sevilla-Noarbe M. Aguena S. Allam O. Alves F. Andrade-Oliveira D. Bacon S. Bocquet D. Brooks R. Camilleri A. Carnero Rosell J. Carretero R. Cawthon L. N. da Costa M. E. da Silva Pereira T. M. Davis J. De Vicente D. L. DePoy S. Desai H. T. Diehl S. Dodelson P. Doel C. Doux A. Drlica-Wagner J. Elvin-Poole S. Everett A. E. Evrard B. Flaugher J. Frieman J. Garc\'ia-Bellido M. Gatti E. Gaztanaga P. Giles R. A. Gruendl G. Gutierrez K. Herner S. R. Hinton D. L. Hollowood K. Honscheid D. Huterer D. J. James K. Kuehn O. Lahav S. Lee H. Lin J. L. Marshall J. Mena-Fern\'andez F. Menanteau R. Miquel J. Muir R. L. C. Ogando D. Petravick A. A. Plazas Malag\'on J. Prat M. Raveri E. S. Rykoff S. Samuroff E. Sanchez T. Shin M. Smith E. Suchyta M. E. C. Swanson G. Tarle D. Thomas C. To D. L. Tucker V. Vikram M. Yamamoto
This is my paper
classification astro-ph.CO
keywords redshift calibrationphotometric redshiftsself-organizing mapsclustering redshiftsDark Energy Surveylens galaxiesweak gravitational lensingn(z) uncertainties
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

The Dark Energy Survey's Year 6 cosmology analysis reads the Universe's structure through the positions and lensing of a bright galaxy sample called MagLim++. This paper supplies the one thing that sample must have to be cosmologically trustworthy: a calibrated answer to 'how far away is each galaxy, and how sure are we?' The authors claim per-bin mean redshifts known to 1-2 percent, a 20-30 percent error reduction over Year 3, with an uncertainty budget covering four systematic sources from cosmic sample variance to photometric zero-point errors to residual bias in the redshift catalogs themselves. The method marries a self-organizing color map that transfers deep-field redshifts onto the wide survey with clustering-redshift measurements that cross-correlate the lens sample against spectroscopic surveys, producing 100 million equally plausible n(z) curves. Those curves are compressed to three modes per bin, chosen by how strongly they disturb the observable correlations, so cosmology runs marginalize over redshift ignorance at negligible cost. If the calibration is correct, lens redshift systematics no longer dominate the DES Y6 3x2pt error budget, and the pipeline is ready to hand to the next generation of surveys.

Core claim

MagLim++ lens n(z) is calibrated to 1-2 percent accuracy in mean redshift per bin. A two-tier Self-Organizing Map does the work: deep-field galaxies fill a 'deep SOM', survey galaxies a 'wide SOM', and the Balrog synthetic-injection catalog bridges them so each survey galaxy inherits a redshift distribution from its photometric phenotype. Dirichlet resampling adds sample variance and shot noise, zero-point perturbations add calibration error, and coherent per-bin shifts add photo-z bias, yielding 100 million n(z) realizations per bin. Clustering-redshift data reweight these; Fisher-based mode projection compresses them to three amplitudes per bin. Published means (Table 1): 0.306 to 1.011, u

What carries the argument

The central object is the two-tier Self-Organizing Map plus its transfer function: a deep SOM on eight-band deep-field photometry (ugrizJHKs) with secure redshifts, a wide SOM on survey griz photometry, and the Balrog synthetic-injection catalogue bridging them, so redshift information flows from calibrator galaxies to survey galaxies of matching photometric phenotype. Three auxiliary mechanisms carry the uncertainty model: three-step Dirichlet (3sDir) resampling separating sample variance from shot noise; Latin Hypercube Sampling over photometric zero-point shifts and over coherent magnitude-redshift-bin shifts, capturing calibration and catalog systematics; and Fisher-based mode projection

Load-bearing premise

The calibration assumes the deep-field redshift catalogs (COSMOS2020, PAUS, and spectroscopy in the COSMOS and X3 fields) are complete and unbiased across the full magnitude-color range of MagLim++; if those catalogs miss galaxies or carry photometric-redshift bias that the coherent-shift model cannot capture, the central n(z) values shift and the 1-2 percent accuracy claim fails.

What would settle it

Measure redshifts directly for a complete sample of MagLim++-like galaxies down to i about 22.2 in the COSMOS and X3 deep fields, with dense spectroscopy or narrow-band photo-z's, and compare the measured per-bin mean redshift against the published SOMPZ+WZ values (0.306, 0.435, 0.624, 0.778, 0.903, 1.011); deviations exceeding the quoted 1-2 percent would falsify the calibration. A cheaper immediate check: recompute the WZ likelihood using only the deepest spectroscopic reference samples and see whether importance sampling drags the posterior off the SOMPZ prior by more than the reported erro

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Lens n(z) systematics become subdominant: marginalizing the three redshift modes per bin costs only a small widening of the S8-Om contours relative to a fixed n(z), with S8 = 0.762 +/- 0.011 in the fiducial 3x2pt run.
  • The clustering-redshift update tightens S8 by about 5 percent in 2x2pt relative to SOMPZ alone and leaves 3x2pt essentially unchanged, since the lens n(z) is already well constrained there.
  • The old shift-and-stretch parametrization yields slightly tighter 3x2pt contours than the new modes, which the paper reads as evidence that shift-and-stretch underestimated the true n(z) uncertainty.
  • The paper flags that the highest-redshift bin's error bars grew relative to Y3, which it interprets as a more faithful accounting of systematic limits, and that two unphysical variance artifacts (negative RU in Bin 1, zeroed ZPU in Bin 5) appear only in the display decomposition, not in the propagated realizations.
  • The pipeline (deep/wide SOM, Balrog transfer, Dirichlet resampling, mode compression) is explicitly transferable to future photometric surveys such as Rubin/LSST and Euclid.

Where Pith is reading between the lines

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

  • The Y3 lens-n(z) treatment shifted cosmology by about 0.4 sigma in the S8-Om plane when replaced by SOMPZ+WZ; with mean-redshift errors now cut by 20-30 percent, the residual lens-redshift systematic should sit well below Y6 statistical errors, a prediction the final Y6 cosmology release can test directly.
  • Bins 3-6 rely on COSMOS as their only redshift calibrator (X3 is dropped beyond i about 20.5); a color- or magnitude-dependent bias common to the COSMOS system that survives the coherent-shift model would enter all high-redshift bins coherently, so an independent deep field remains the strongest external check.
  • The RU prior width is set equal to measured photo-z scatter per magnitude-redshift bin; where spectroscopy is sparse at the faint end, the quoted 1-2 percent errors inherit that prior, so denser spectroscopy in those corners of color space would directly test whether the error bars are correctly sized.
  • The mode basis is Fisher-optimized for the data vector, not for cosmological parameters, a distinction the paper notes; choosing the basis in parameter space could in principle extract more dark-energy information from the same 100 million realizations.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. This paper presents the redshift calibration of the DES Y6 MagLim++ lens sample, combining the SOMPZ photometric method with clustering-redshift information. The pipeline uses deep-field photometry, a Balrog-based transfer function, and external redshift catalogs to generate 100 million n(z) realizations per tomographic bin, propagating sample variance, shot noise, zero-point uncertainty, and redshift-sample uncertainty. The realizations are reweighted by a WZ likelihood via importance sampling and compressed into a small number of orthogonal modes for cosmological inference. The authors report 1–2% uncertainties on the mean redshift, a 20–30% average improvement over DES Y3, and a minor degradation of cosmological constraints after marginalizing over the redshift modes.

Significance. If correct, this is an important methodological and data-product paper for the DES Y6 3x2pt cosmology analysis and a useful template for future surveys. Its strengths include an unusually complete uncertainty budget (three-step Dirichlet sampling for sample variance and shot noise, LHS zero-point perturbations, coherent per-bin redshift-sample shifts), a transparent combination with WZ via importance sampling, and a mode-compression scheme that is validated against cosmological parameter inference. The public release of the n(z) ensembles and the detailed appendices are also commendable. The main risks are the treatment of catastrophic photo-z outliers in the redshift-sample uncertainty model and the lack of importance-sampling diagnostics; both are addressable with additional quantitative checks.

major comments (4)
  1. [§3.1.1, Eqs. (2) and (3)] The two displayed forms of the SOMPZ combination are inconsistent. Eq. (2) reads p(z)=Σ_ĉ Σ_c p(z|c) p(c) p(c|ĉ) p(ĉ), which contains an extra factor p(c) relative to the standard marginalization p(z)=Σ_ĉ p(ĉ) Σ_c p(z|c) p(c|ĉ). Eq. (3), by contrast, reduces to Σ p(z|c) p(c,ĉ), which is the correct combination provided p(c,ĉ) is the joint deep-wide cell probability. Please reconcile Eq. (2) with Eq. (3), or define p(c|ĉ) precisely. As written, a reader cannot tell which expression the pipeline actually implements.
  2. [§4.3 and Fig. 9] The redshift-sample uncertainty model uses coherent shifts Δz∼N(b(i), s(i)) per magnitude–redshift bin. This Gaussian-coherent model cannot represent catastrophic photo-z outliers |Δz|/(1+z)>0.1, which are by definition a secondary population rather than a shift of the whole bin. The paper itself notes in Fig. 9 that some cells have large median biases because they are dominated by Balmer/Lyman-break outlier confusion. Because the MagLim++ selection (§2.1) removes only compact high-dispersion SOM regions, residual outlier contamination in non-compact or multi-modal cells can enter the calibration. Please quantify, per tomographic bin, the MagLim++-weighted fraction of calibration weight in cells with outlier rate above the 10% threshold (or equivalently with large Balmer/Lyman-break contamination), and either demonstrate that this weight is negligible or augment the RU model with an outl
  3. [§5.1 and §5.3] The final SOMPZ+WZ n(z) ensemble and the mode priors are obtained by importance-sampling 100 million SOMPZ realizations against the WZ likelihood. No effective sample size (ESS) or stability diagnostic is reported. If the WZ likelihood strongly downweights the prior ensemble, the effective posterior sample size may be far smaller than 100M, which would make the mode coefficients and their priors noisy and would weaken the claim of a 20–30% uncertainty reduction. Please report the ESS per tomographic bin, and check that the mode coefficients and their priors are stable under bootstrap resampling or alternative WZ covariance assumptions.
  4. [§4.1] The 3sDir uncertainty model deliberately excludes non-COSMOS redshifts, while the central n(z) estimate for bins 1–2 uses X3 data (Section 2.2). The manuscript states this is conservative, but it creates a mismatch between the data that generate the central value and the data that generate the uncertainty. Please report the quantitative impact of including versus excluding X3 in the uncertainty model (e.g., the change in the mean-z variance for bins 1 and 2). If the exclusion is intended to avoid poorly characterized selection effects, a brief numerical demonstration that the bins' conclusions are unaffected would close the loop.
minor comments (4)
  1. [Table 2] The neutrino-mass row reads mν[eV] = 0.77 with prior [0.06, 0.6], which is inconsistent both with the text (mν fixed to 0.06 eV in §6) and with the prior range. Please correct the central value or the prior.
  2. [§4.2] The text first says zero-point shifts are drawn for all bands (ugrizYJHKs), then says the dominant impact arises from the u-band. Please clarify whether the fiducial ZPU realizations perturb all bands or only the u-band, since this affects how to interpret the LHS sampling and the Bin-6 ZPU contribution.
  3. [§3.1.1] There is a typo immediately after Eq. (2): 'where We can rewrite this equation' should be lowercase 'we'.
  4. [Abstract and §5] The '20–30% average reduction relative to DES Y3' is a comparison between different error models, and the paper itself notes that the Y3 Bin-6 uncertainty may have been underestimated. Please add a sentence making explicit that part of the improvement may reflect more complete error accounting rather than strictly better data.

Circularity Check

0 steps flagged

No significant circularity: the n(z) calibration is an empirical estimate from external deep-field, spectroscopic, and clustering data, with uncertainties propagated from stated inputs.

full rationale

The paper derives the MagLim++ redshift distributions from independent data products: deep-field photometry, external spectroscopic and narrow-band photo-z catalogs (COSMOS2020, PAUS, BOSS/eBOSS), and Balrog simulations. The SOMPZ mapping (Eqs. 2-3) is a direct reweighting of the redshift sample to the wide-field selection, not an inversion of the target quantity. The 3sDir uncertainty model samples Dirichlet distributions around the measured redshift sample, and the RU model applies coherent shifts whose means and widths are measured from spectroscopic comparisons (Eqs. 6-7); these are error-propagation steps, not fitted parameters renamed as predictions. The WZ likelihood (Eq. 9) uses external angular clustering measurements and is presented as an independent constraint combined via importance sampling; it is not derived from the SOMPZ n(z). The mode compression (Eqs. 10-12) is a lossy representation of the already-generated realizations, validated by the small discarded chi-square, and the cosmological tests in Section 6 are internal consistency checks using simulated data vectors, not predictions of the n(z) itself. Self-citations to companion papers (d'Assignies et al. 2025 for WZ, Bernstein et al. 2025 for mode projection) are methodological, and the underlying data are external and falsifiable; no uniqueness theorem or ansatz is imported to force the result. The paper's own caveat about Balmer/Lyman-break outlier cells (Section 4.3, Figure 9) is a modeling limitation that would affect accuracy, but it does not make the derivation circular.

Axiom & Free-Parameter Ledger

4 free parameters · 6 axioms · 0 invented entities

The central n(z) values are data-driven; the paper's own free parameters are confined to the uncertainty model (RU prior widths, ZPU amplitudes, mode truncation threshold). The calibration rests on three external pillars: completeness and unbiasedness of the redshift samples, fidelity of the Balrog transfer function, and correctness of the companion WZ model. The unspecified 3sDir concentration parameters (inherited from Sanchez et al. 2020) are a reproducibility gap rather than a fitted parameter. No new physical entities are introduced.

free parameters (4)
  • RU shift priors b(i), s(i) per magnitude-redshift bin = median bias and bootstrap scatter from spec-z comparisons (Figure 9), e.g. median biases up to ~0.05-0.075 with outlier-
    Section 4.3: the Gaussian prior for coherent photo-z shifts in each magnitude-redshift bin is centered on the measured median bias and has width equal to the measured photo-z scatter. These are fitted to spec-z data and directly set the breadth of the RU realizations.
  • Zero-point uncertainty amplitudes = 0.055 mag (u), 0.005 (g,r,i,z), 0.008 (J,H,Ks)
    Section 4.2: amplitudes of the LHS-sampled zero-point perturbations, taken from Hartley et al. (2022). They set the scale of ZPU, which dominates Bin 6.
  • Mode truncation threshold = cumulative discarded chi^2 < 0.15
    Section 5.3: threshold for the number of retained modes (3 for SOMPZ+WZ, 4 for SOMPZ). Chosen by hand; a different threshold changes the number of redshift nuisance parameters in the cosmology fit.
  • 3sDir Dirichlet concentration parameters
    Section 4.1: three-step Dirichlet draws for superphenotype, redshift-given-superphenotype, and phenotype-given-redshift. Concentration parameters are not specified in this paper; they are inherited from Sanchez et al. (2020), and they control the amplitude of sample variance and shot noise in the realizations.
axioms (6)
  • domain assumption Redshift calibration samples (COSMOS2020, PAUS, spectra) are complete and unbiased over the MagLim++ magnitude-color range in the COSMOS and X3 fields.
    Section 2.2: completeness is asserted and is the basis for excluding C3/E2 fields and restricting X3 to i<20.5. If the completeness assessment is wrong, the central n(z) is biased and the 1-2% mean claim fails.
  • domain assumption Balrog injection reproduces the wide-field detection and selection process, giving an unbiased transfer function p(c|ĉ).
    Section 2.1 and Eq. 3: the deep-to-wide connection rests entirely on the Balrog catalog. The paper delegates validation to Anbajagane et al. (2025) and does not test alternate injection models.
  • domain assumption Residual photo-z biases are captured by coherent per-bin Gaussian shifts; there is no unmodeled global redshift-scale error or catastrophic-outlier population.
    Section 4.3: the RU model draws shifts from Gaussians centered at the measured median bias with width equal to the scatter. Catastrophic outliers (|dz|/(1+z)>0.1, e.g. Balmer/Lyman confusion cells in Figure 9) are not explicitly modeled.
  • domain assumption The WZ likelihood model (galaxy bias, magnification, covariance) of d'Assignies et al. (2025) is correct.
    Section 5.1, Eqs. 8-9: the importance sampling weights depend entirely on this companion paper's model; no WZ details or validation are reproduced here.
  • domain assumption The u-band dominates deep-field zero-point error, and the transfer function is unaffected by zero-point perturbations.
    Section 4.2: only deep-field fluxes are perturbed and Balrog is not recomputed per realization. If zero-point errors are correlated across bands or the transfer function shifts, Bin 6 (ZPU-dominated) uncertainty is underestimated.
  • standard math Standard Bayesian and sampling machinery: Dirichlet resampling, importance sampling consistency, Fisher-based mode projection.
    Sections 4.1, 5.1, 5.3: the statistical machinery is standard; the specific 3sDir factorization is from Sanchez et al. (2020).

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Dark Energy Survey Year 6 Results: Redshift Calibration of the MagLim++ Lens Sample." pith.science (2026). https://pith.science/paper/F7AFDWH6

@misc{pith2026250907964,
  author       = {Pith},
  title        = {Pith review of: Dark Energy Survey Year 6 Results: Redshift Calibration of the MagLim++ Lens Sample},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F7AFDWH6}},
  note         = {Machine review of arXiv:2509.07964}
}
Share X Bluesky LinkedIn Reddit HN
abstract

In this work, we derive and calibrate the redshift distribution of the MagLim++ lens galaxy sample used in the Dark Energy Survey Year 6 (DES Y6) 3x2pt cosmology analysis. The 3x2pt analysis combines galaxy clustering from the lens galaxy sample and weak gravitational lensing. The redshift distributions are inferred using the SOMPZ method - a Self-Organizing Map framework that combines deep-field multi-band photometry, wide-field data, and a synthetic source injection (Balrog) catalog. Key improvements over the DES Year 3 (Y3) calibration include a noise-weighted SOM metric, an expanded Balrog catalogue, and an improved scheme for propagating systematic uncertainties, which allows us to generate O($10^8$) redshift realizations that collectively span the dominant sources of uncertainty. These realizations are then combined with independent clustering-redshift measurements via importance sampling. The resulting calibration achieves typical uncertainties on the mean redshift of 1-2%, corresponding to a 20-30% average reduction relative to DES Y3. We compress the $n(z)$ uncertainties into a small number of orthogonal modes for use in cosmological inference. Marginalizing over these modes leads to only a minor degradation in cosmological constraints. This analysis establishes the MagLim++ sample as a robust lens sample for precision cosmology with DES Y6 and provides a scalable framework for future surveys.

Figures

Figures reproduced from arXiv: 2509.07964 by A. Alarcon, A. Amon, A. A. Plazas Malag\'on, A. Carnero Rosell, A. Drlica-Wagner, A. E. Evrard, A. Porredon, B. Flaugher, B. Yin, C. Chang, C. Doux, C. S\'anchez, C. To, D. Anbajagane, D. Bacon, D. Brooks, D. Gruen, D. Huterer, D. J. James, D. L. DePoy, D. L. Hollowood, D. L. Tucker, D. Petravick, D. Sanchez Cid, D. Thomas, E. Gaztanaga, E. Sanchez, E. S. Rykoff, E. Suchyta, F. Andrade-Oliveira, F. Menanteau, G. Giannini, G. Gutierrez, G. M. Bernstein, G. Tarle, H. Lin, H. T. Diehl, I. Sevilla-Noarbe, J. Blazek, J. Carretero, J. De Vicente, J. Elvin-Poole, J. Frieman, J. Garc\'ia-Bellido, J. L. Marshall, J. Mena-Fern\'andez, J. Muir, J. Myles, J. Prat, K. Bechtol, K. Herner, K. Honscheid, K. Kuehn, L. N. da Costa, M. Aguena, M. A. Troxel, M. Crocce, M. E. C. Swanson, M. E. da Silva Pereira, M. Gatti, M. Raveri, M. R. Becker, M. Rodriguez-Monroy, M. Smith, M. Yamamoto, N. Weaverdyck, O. Alves, O. Lahav, P. Doel, P. Giles, R. A. Gruendl, R. Camilleri, R. Cawthon, R. L. C. Ogando, R. Miquel, S. Allam, S. Avila, S. Bocquet, S. Desai, S. Dodelson, S. Everett, S. Lee, S. R. Hinton, S. Samuroff, T. M. Davis, T. Shin, V. Vikram, W. d'Assignies.

Figure 1
Figure 1. Figure 1: Flowchart summarizing the DES Y6 3×2pt redshift calibration pipeline. The target sample of this study is MagLim++ (Section 2), with redshift information provided by the redshift sample (Section 2.2) and deep fields combined with Balrog simulations (Section 2). Photometric data are then mapped using Self-Organizing Maps (SOMs) for deep and wide fields (Section 3), transferring the redshift information to th… view at source ↗
Figure 2
Figure 2. Figure 2: The four DES deep fields used for our redshift analysis, which includes overlapping deep DES ugriz bands and VIDEO or UltraVISTA JHK bands, as compiled from Hartley et al. (2022). Black points indicate DES deep-field galaxies with no redshift information, while points with colors show galaxies with spectroscopy or from COSMOS2020, PAUS+COSMOS or PAUSW1, as indicated in the legend [PITH_FULL_IMAGE:figures/… view at source ↗
Figure 3
Figure 3. Figure 3: Distribution of redshift samples as a function of the deep field DES 𝑖-band magnitude. Each galaxy in these stacked histograms is weighted by the Balrog probability of detection and selection into each Maglim sample (different rows). For details on the definition of the ‘Bulge Plus Disk, Fixed Ratio’ (BDF) galaxy profile see Hartley et al. (2022). 2.2 The redshift catalogs Our analysis relies on the use of… view at source ↗
Figure 4
Figure 4. Figure 4: Redshift quality of different redshift catalogues used for the red￾shift calibration of MagLim++ samples, relative to spec-z measurements for the same objects, using Δ𝑧 /(1 + 𝑧spec ). We define outliers as having |Δ𝑧 |/(1+𝑧spec ) > 0.1. The redshift catalogues shown are PAUS+COSMOS; PAUS+W1; and then Classic-EAZY and Farmer-Lephare from the COS￾MOS2020 release (see text). Relative to each other, Classic-EA… view at source ↗
Figure 5
Figure 5. Figure 5: Visualisation of the 6 deep SOMs, one for each tomographic bin (ordered from left to right and top to bottom, i.e., 1–2 on the first row, 3–4 on the second, 5–6 on the third), each composed of 144 cells (12×12). Left: average redshift for each deep SOM cell 𝑐. Right: standard deviation on the redshift distribution for each deep SOM cell 𝑐. The black cells in the deep SOM are due to the lack of spectroscopi… view at source ↗
Figure 6
Figure 6. Figure 6: Visualisation of the 6 wide SOMs, one for each tomographic bin, each composed of 1024 cells (32×32). Left: average redshift for each wide SOM cell 𝑐ˆ. Right: standard deviation on the redshift distribution for each wide SOM cell 𝑐ˆ. regions, galaxies with similar photometric properties have broad or multimodal redshift distributions that cannot be cleanly separated, even with the deep multi-band photometry… view at source ↗
Figure 7
Figure 7. Figure 7: Ensemble of 100M 𝑛(𝑧) distributions for the MagLim++ lens sample, for each of the six tomographic bins. The spread across realizations reflects the four sources of uncertainty included in cosmological analyses: sample variance, shot noise, zero point, and redshift uncertainty. The legend reports also the mean redshift and its uncertainty. shifts in magnitude–redshift bins based on spectroscopic compar￾ison… view at source ↗
Figure 8
Figure 8. Figure 8: The relative contribution of each source of uncertainty for the mean (left) and the width (right) for each of the six tomographic bins. We show the individual contributions from sample variance (SV, blue), shot noise (SN, light blue), redshift sample uncertainty (RU, purple), and photometric zeropoint uncertainty (ZPU, green). No single source dominates all bins or metrics, underscoring the need to propaga… view at source ↗
Figure 9
Figure 9. Figure 9: Distribution and photometric redshift performance across magnitude and redshift bins, for the four photometric redshift samples Cosmos Classic, Cosmos Farmer, PAUS Cosmos, and PAUS W1. Left: Number of galaxies (with spectra) in each magnitude–redshift bin. Center: Median photometric redshift bias (𝑧photoz − 𝑧specz ). Right: Scatter of photometric redshift estimates. On the left column, we indicate with a w… view at source ↗
Figure 10
Figure 10. Figure 10: Cumulative discarded 𝜒 2 as a function of the number of retained modes. The sharp decline indicates that most of the cosmologically relevant variance is captured in just a few modes, justifying our truncation choice. SOMPZ with the external constraints from WZ, we adopt an im￾portance sampling scheme that reweights the SOMPZ realizations based on their consistency with the clustering-redshift measure￾ment… view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of the redshift distributions for the six tomographic bins of MagLim++, reconstructed from the modes. The blue lines represent the average SOMPZ 𝑛(𝑧) reconstructed from the modes, with the ±1𝜎 and ±2𝜎 bands showcasing the realizations. The orange lines and bands represent the SOMPZ+WZ 𝑛(𝑧) reconstructed from modes. Solid and dashed styles are used solely to differentiate tomographic bins in reg… view at source ↗
Figure 12
Figure 12. Figure 12: Comparison of marginalization methods for photometric redshift uncertainties for 2×2pt (left) and 3x2pt (right). The figure contrasts the shift and stretch approach (green) with the mode-based marginalization method used in the fiducial analysis (blue). While the resulting 𝑆8–Ωm constraints for 2×2pt are very similar, for 3x2pt the shift and stretch method yields slightly tighter contours, suggesting that… view at source ↗
Figure 13
Figure 13. Figure 13: Impact of photo-z uncertainty modeling on cosmological parameter constraints (Ωm, 𝜎8, and 𝑆8) in 2×2pt analysis. The different contours show the effect of fixing the n(z) for both lenses and sources, fixing n(z) only for lenses while marginalizing over SOMPZ+WZ modes for sources, using SOMPZ￾only modes for lenses, and marginalizing over SOMPZ+WZ modes for both lenses and sources. Increasing marginalizatio… view at source ↗
Figure 14
Figure 14. Figure 14: Impact of mismatches between the redshift distributions used in the theory data vector and the redshift modes used for data compression on cosmological constraints. We show the Ωm–𝑆8 posterior contours for three runs: using SOMPZ redshift distributions with SOMPZ+WZ compression modes (red), using SOMPZ+WZ redshift distributions with SOMPZ modes (green), and using SOMPZ+WZ redshift distributions with match… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Assessing the large-scale angular clustering of UNIONS Lyman Break Galaxies via cross-correlations

    astro-ph.CO 2026-07 conditional novelty 6.0

    UNIONS LBGs yield robust cross-power spectra with Planck CMB lensing and DESI/Quaia quasars at amplitudes matching theory, despite systematics that render the auto-spectrum unusable.

  2. Assessing the large-scale angular clustering of UNIONS Lyman Break Galaxies via cross-correlations

    astro-ph.CO 2026-07 conditional novelty 6.0

    UNIONS Lyman-break galaxy auto-clustering is unusable at large scales due to imaging systematics, but LBG x CMB-lensing and LBG x quasar cross-spectra are measured robustly, with amplitudes consistent with predictions.

  3. Consistency of DES and DESI distances and the Standard Cosmological Model

    astro-ph.CO 2026-06 unverdicted novelty 6.0

    DESI recovers the CMB-constrained parameter combination at sub-percent precision; high-redshift DES supernovae fit flat Lambda CDM while a 0.05 mag offset in the low-redshift anchor sample may be systematic, supportin...

  4. Constraints on Dynamical Dark Energy from Multiple Probes in the Full Dark Energy Survey

    astro-ph.CO 2026-05 unverdicted novelty 6.0

    Full DES data from SN+BAO+3x2pt yields w0=-0.84, wa=-0.44 with 2.2σ deviation from ΛCDM; adding DESI+CMB reaches 3.0σ while 3x2pt improves figure of merit by ~10%.

  5. Dark Energy Survey Year 6 Results: Redshift Calibration of the Weak Lensing Source Galaxies

    astro-ph.CO 2025-10 conditional novelty 6.0

    DES Y6 weak lensing source galaxies now have calibrated redshift distributions with mean-redshift uncertainties 0.008–0.024, built from SOMPZ + clustering-redshift + blending corrections.

  6. Modeling Redshift Uncertainties in Roman Weak Lensing Cosmology

    astro-ph.CO 2026-02 conditional novelty 4.0

    A PCA-based redshift uncertainty model, implemented for Roman in CoCoA, matches or outperforms the nine-bin mean-shift model at mitigating S8 and Omega_m biases in simulated cosmic shear analyses.

Reference graph

Works this paper leans on

74 extracted references · 30 canonical work pages · cited by 5 Pith papers · 12 internal anchors

  1. [1]

    Ahumada , R., Allende Prieto , C., Almeida , A., et al., 2020, , 249, 1, 3, arXiv: 1912.02905

    arXiv 2020

  2. [2]

    Alam , S., Ata , M., Bailey , S., et al., 2017, , 470, 3, 2617, arXiv: 1607.03155

    Pith/arXiv arXiv 2017

  3. [3]

    Alam , S., de Mattia , A., Tamone , A., et al., 2021, , 504, 4, 4667, arXiv: 2007.09004

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  4. [4]

    Alarcon , A., Gaztanaga , E., Eriksen , M., et al., 2021, , 501, 4, 6103, arXiv: 2007.11132

    Pith/arXiv arXiv 2021

  5. [5]

    Anbajagane , D., Tabbutt , M., Beas-Gonzalez , J., et al., 2025, arXiv e-prints, arXiv:2501.05683, arXiv: 2501.05683

    Pith/arXiv arXiv 2025

  6. [6]

    Bechtol , K., Sevilla-Noarbe , I., Drlica-Wagner , A., et al., 2025

    work page 2025

  7. [7]

    On the realistic validation of photometric redshifts, or why Teddy will never be Happy

    Beck , R., Lin , C. A., Ishida , E. E. O., et al., 2017, , 468, 4, 4323, arXiv: 1701.08748

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  8. [8]

    A., et al., 2025, arXiv e-prints, arXiv:2506.00758, arXiv: 2506.00758

    Bernstein , G., d'Assignies , W., Troxel , M. A., et al., 2025, arXiv e-prints, arXiv:2506.00758, arXiv: 2506.00758

    Pith/arXiv arXiv 2025

  9. [9]

    A., MacCrann , N., Troxel , M

    Blazek , J. A., MacCrann , N., Troxel , M. A., Fang , X., 2019, , 100, 10, 103506, arXiv: 1708.09247

    Pith/arXiv arXiv 2019

  10. [10]

    Buchs , R., Davis , C., Gruen , D., et al., 2019, , 489, 1, 820, arXiv: 1901.05005

    Pith/arXiv arXiv 2019

  11. [11]

    Campos , A., Yin , B., Dodelson , S., et al., 2024, arXiv e-prints, arXiv:2408.00922, arXiv: 2408.00922

    Pith/arXiv arXiv 2024

  12. [12]

    SOMz: photometric redshift PDFs with self organizing maps and random atlas

    Carrasco Kind , M., Brunner , R. J., 2014, , 438, 4, 3409, arXiv: 1312.5753

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  13. [13]

    Cawthon , R., Elvin-Poole , J., Porredon , A., et al., 2022, , 513, 4, 5517, arXiv: 2012.12826

    Pith/arXiv arXiv 2022

  14. [14]

    Dark Energy Survey Year 3 results: Marginalisation over redshift distribution uncertainties using ranking of discrete realisations

    Cordero , J. P., Harrison , I., Rollins , R. P., et al., 2022, , 511, 2, 2170, arXiv: 2109.09636

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  15. [15]

    Dark Energy Survey Collaboration , 2025,

    work page 2025

  16. [16]

    d'Assignies , W., et al., 2025,

    work page 2025

  17. [17]

    S., Kneib , J.-P., Percival , W

    Dawson , K. S., Kneib , J.-P., Percival , W. J., et al., 2016, , 151, 2, 44, arXiv: 1508.04473

    Pith/arXiv arXiv 2016

  18. [18]

    S., Schlegel , D

    Dawson , K. S., Schlegel , D. J., Ahn , C. P., et al., 2013, , 145, 1, 10, arXiv: 1208.0022

    Pith/arXiv arXiv 2013

  19. [19]

    De Vicente , J., S \'a nchez , E., Sevilla-Noarbe , I., 2016, , 459, 3, 3078, arXiv: 1511.07623

    Pith/arXiv arXiv 2016

  20. [20]

    DES , C., Abbott , T. M. C., Adamow , M., et al., 2024, , 110, 6, 063515, arXiv: 2402.10696

    arXiv 2024

  21. [21]

    DES Collaboration , 2022, , 105, 2, 023520, arXiv: 2105.13549

    Pith/arXiv arXiv 2022

  22. [22]

    G., Aguilar , J., et al., 2024, , 168, 2, 58, arXiv: 2306.06308

    DESI Collaboration , Adame , A. G., Aguilar , J., et al., 2024, , 168, 2, 58, arXiv: 2306.06308

    Pith/arXiv arXiv 2024

  23. [23]

    Eriksen , M., Alarcon , A., Gaztanaga , E., et al., 2019, , 484, 3, 4200, arXiv: 1809.04375

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  24. [24]

    Fang , X., Krause , E., Eifler , T., 2020, CosmoCov: Configuration space covariances for projected galaxy 2-point statistics , Astrophysics Source Code Library, record ascl:2006.005

    work page 2020

  25. [25]

    T., Honscheid , K., et al., 2015, , 150, 5, 150, arXiv: 1504.02900

    Flaugher , B., Diehl , H. T., Honscheid , K., et al., 2015, , 150, 5, 150, arXiv: 1504.02900

    Pith/arXiv arXiv 2015

  26. [26]

    M., et al., 2022, , 510, 1, 1223, arXiv: 2012.08569

    Gatti , M., Giannini , G., Bernstein , G. M., et al., 2022, , 510, 1, 1223, arXiv: 2012.08569

    arXiv 2022

  27. [27]

    Giannini , G., Alarcon , A., Gatti , M., et al., 2024, , 527, 2, 2010, arXiv: 2209.05853

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  28. [28]

    Gruen , D., Brimioulle , F., 2017, , 468, 1, 769, arXiv: 1610.01160

    Pith/arXiv arXiv 2017

  29. [29]

    C., Ogando , R

    Gschwend , J., Rossel , A. C., Ogando , R. L. C., et al., 2018, Astronomy and Computing, 25, 58, arXiv: 1708.05643

    Pith/arXiv arXiv 2018

  30. [30]

    The impact of spectroscopic incompleteness in direct calibration of redshift distributions for weak lensing surveys

    Hartley , W. G., Chang , C., Samani , S., et al., 2020, , 496, 4, 4769, arXiv: 2003.10454

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  31. [31]

    G., Choi , A., Amon , A., et al., 2022, , 509, 3, 3547, arXiv: 2012.12824

    Hartley , W. G., Choi , A., Amon , A., et al., 2022, , 509, 3, 3547, arXiv: 2012.12824

    arXiv 2022

  32. [32]

    Heymans , C., Tr \"o ster , T., Asgari , M., et al., 2021, , 646, A140, arXiv: 2007.15632

    Pith/arXiv arXiv 2021

  33. [33]

    Hildebrandt , H., Arnouts , S., Capak , P., et al., 2010, , 523, A31, arXiv: 1008.0658

    Pith/arXiv arXiv 2010

  34. [34]

    J., Ilbert , O., et al., 2016, , 224, 24

    Laigle , C., McCracken , H. J., Ilbert , O., et al., 2016, , 224, 24

    work page 2016

  35. [35]

    U., 2023, , 525, 2, 3181, arXiv: 2306.16923

    Lange , J. U., 2023, , 525, 2, 3181, arXiv: 2306.16923

    Pith/arXiv arXiv 2023

  36. [36]

    Le F \`e vre , O., Cassata , P., Cucciati , O., et al., 2013, , 559, A14, arXiv: 1307.0545

    Pith/arXiv arXiv 2013

  37. [37]

    Leauthaud , A., Saito , S., Hilbert , S., et al., 2017, , 467, 3, 3024, arXiv: 1611.08606

    Pith/arXiv arXiv 2017

  38. [38]

    J., Le Brun , V., Maier , C., et al., 2009, , 184, 2, 218

    Lilly , S. J., Le Brun , V., Maier , C., et al., 2009, , 184, 2, 218

    work page 2009

  39. [39]

    Mandelbaum , R., Slosar , A., Baldauf , T., et al., 2013, , 432, 2, 1544, arXiv: 1207.1120

    Pith/arXiv arXiv 2013

  40. [40]

    C., Stern , D

    Masters , D. C., Stern , D. K., Cohen , J. G., et al., 2017, , 841, 111

    work page 2017

  41. [41]

    C., Stern , D

    Masters , D. C., Stern , D. K., Cohen , J. G., et al., 2019, , 877, 2, 81, arXiv: 1904.06394

    Pith/arXiv arXiv 2019

  42. [42]

    J., Milvang-Jensen , B., Dunlop , J., et al., 2012, , 544, A156, arXiv: 1204.6586

    McCracken , H. J., Milvang-Jensen , B., Dunlop , J., et al., 2012, , 544, A156, arXiv: 1204.6586

    Pith/arXiv arXiv 2012

  43. [43]

    Mena-Fern \'a ndez , J., Rodr \' guez-Monroy , M., Avila , S., et al., 2024, , 110, 6, 063514, arXiv: 2402.10697

    work page arXiv 2024

  44. [44]

    Myles , J., Alarcon , A., Amon , A., et al., 2020, arXiv e-prints, arXiv:2012.08566, arXiv: 2012.08566

    Pith/arXiv arXiv 2020

  45. [45]

    Myles , J., Gruen , D., Amon , A., et al., 2023, , 519, 2, 1792, arXiv: 2210.03130

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  46. [46]

    Navarro-Giron \'e s , D., Gazta \ n aga , E., Crocce , M., et al., 2024, , 534, 2, 1504, arXiv: 2312.07581

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  47. [47]

    A., Gruen , D., 2022, , 60, 363, arXiv: 2206.13633

    Newman , J. A., Gruen , D., 2022, , 60, 363, arXiv: 2206.13633

    Pith/arXiv arXiv 2022

  48. [48]

    The Physics of the Accelerating Universe Camera

    Padilla , C., Castander , F. J., Alarc \'o n , A., et al., 2019, , 157, 6, 246, arXiv: 1902.03623

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  49. [49]

    Porredon , A., Crocce , M., Elvin-Poole , J., et al., 2022, , 106, 10, 103530, arXiv: 2105.13546

    Pith/arXiv arXiv 2022

  50. [50]

    Porredon , A., Crocce , M., Fosalba , P., et al., 2021, , 103, 4, 043503, arXiv: 2011.03411

    Pith/arXiv arXiv 2021

  51. [51]

    Rodr \'i guez-Monroy , M., Weaverdyck , N., et al., 2025,

    work page 2025

  52. [52]

    S., Abate , A., et al., 2016, , 461, 1431, arXiv: 1507.05460

    Rozo , E., Rykoff , E. S., Abate , A., et al., 2016, , 461, 1431, arXiv: 1507.05460

    Pith/arXiv arXiv 2016

  53. [53]

    N., Eifler, T., Behroozi, P., 2025, The Astrophysical Journal, 983, 2, 183

    Salcedo, A. N., Eifler, T., Behroozi, P., 2025, The Astrophysical Journal, 983, 2, 183

    work page 2025

  54. [54]

    S \'a nchez , C., Prat , J., Zacharegkas , G., et al., 2022, , 105, 8, 083529, arXiv: 2105.13542

    Pith/arXiv arXiv 2022

  55. [55]

    M., 2020, , 498, 2, 2984, arXiv: 2004.09542

    S \'a nchez , C., Raveri , M., Alarcon , A., Bernstein , G. M., 2020, , 498, 2, 2984, arXiv: 2004.09542

    Pith/arXiv arXiv 2020

  56. [56]

    Sanchez-Cid , D., et al., 2025,

    work page 2025

  57. [57]

    F., Meisner , A

    Schlafly , E. F., Meisner , A. M., Green , G. M., 2019, , 240, 2, 30, arXiv: 1901.03337

    Pith/arXiv arXiv 2019

  58. [58]

    Schutt , T., Jarvis , M., Roodman , A., et al., 2025

    work page 2025

  59. [59]

    Scodeggio , M., Guzzo , L., Garilli , B., et al., 2018, , 609, A84, arXiv: 1611.07048

    Pith/arXiv arXiv 2018

  60. [60]

    Scoville , N., Aussel , H., Brusa , M., et al., 2007, , 172, 1, 1, arXiv: astro-ph/0612305

    Pith/arXiv arXiv 2007

  61. [61]

    A., Gunn , J

    Smee , S. A., Gunn , J. E., Uomoto , A., et al., 2013, , 146, 2, 32, arXiv: 1208.2233

    Pith/arXiv arXiv 2013

  62. [62]

    Euclid Preparation: XIV. The Complete Calibration of the Color-Redshift Relation (C3R2) Survey: Data Release 3

    Stanford , S. A., Masters , D., Darvish , B., et al., 2021, , 256, 1, 9, arXiv: 2106.11367

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  63. [63]

    Sugiyama , S., Miyatake , H., More , S., et al., 2023, , 108, 12, 123521, arXiv: 2304.00705

    Pith/arXiv arXiv 2023

  64. [64]

    Tanoglidis , D., Chang , C., Frieman , J., 2020, , 491, 3, 3535, arXiv: 1908.07150

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  65. [65]

    R., Kauffmann , O

    Weaver , J. R., Kauffmann , O. B., Ilbert , O., et al., 2022, , 258, 1, 11, arXiv: 2110.13923

    Pith/arXiv arXiv 2022

  66. [66]

    Weaverdyck , N., Rodr \'i guez-Monroy , M., et al., 2025 a ,

    work page 2025

  67. [67]

    Weaverdyck , N., et al., 2025 b ,

    work page 2025

  68. [68]

    H., Mortonson , M

    Weinberg , D. H., Mortonson , M. J., Eisenstein , D. J., Hirata , C., Riess , A. G., Rozo , E., 2013, , 530, 2, 87, arXiv: 1201.2434

    Pith/arXiv arXiv 2013

  69. [69]

    R., Sheldon , E., et al., 2025

    Yamamoto , M., Becker , M. R., Sheldon , E., et al., 2025

    work page 2025

  70. [70]

    Yin , B., et al., 2025,

    work page 2025

  71. [71]

    Zuntz , J., Paterno , M., Jennings , E., et al., 2015, Astronomy and Computing, 12, 45, arXiv: 1409.3409

    Pith/arXiv arXiv 2015

  72. [72]

    , " * write output.state after.block = add.period write newline

    ENTRY address archiveprefix author booktitle chapter edition editor eid eprint howpublished institution issn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

    work page

  73. [73]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in " " * FUNCTION format....

    work page

  74. [74]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1...

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.