Dark Energy Survey Year 6 Results: fast and interpretable posterior predictive checks for correlated cosmic probes
Abstract
We present a fast and numerically stable framework for conducting posterior predictive distribution (PPD) tests of internal consistency in correlated cosmological probes. Our method, developed as a validation check for the Dark Energy Survey (DES) Year 6 combined analysis of weak lensing and galaxy clustering, employs a Gaussian-mixture model to efficiently approximate the high-dimensional PPD and introduces a scalar consistency metric, ΔPPD, defined as the fraction of the PPD at lower probability density than the observed data. This method avoids challenges related to calibration and numerical stability that impacted PPD methods previously used in DES Year 3. After demonstrating performance with a toy model, we confirm that when applied to DES Year 3 measurements, ΔPPD identifies the same inter-probe tensions as published, calibrated PPD results while requiring only minutes per test. We further validate the method on several noise realizations of simulated DES Y6 analyses, demonstrating that ΔPPD responds coherently to injected inconsistency, while highlighting two effects which can complicate the interpretation of PPD tests for increasingly complex models. First, internal tensions that can be absorbed as shifts in nuisance parameters become more difficult to detect. Second, poorly constrained directions in parameter space can induce projection effects in PPDs for disjoint sets of observables, producing reports of tension even for consistent data. We conclude that ΔPPD provides a practical PPD-based diagnostic for multiprobe unblinding, to be interpreted with care alongside other validation checks. This framework has already played an important role in DES analyses and is well suited for application to forthcoming Stage-IV surveys such as LSST, Euclid, and Roman, where fast, robust internal consistency tests will be essential.
Related
- Assessing the consistency of the CPL dark-energy model with cosmological data: Simulation-calibrated neural posterior disagreement
- Dark Energy Survey year 6 results: Cosmological constraints from galaxy clustering and weak lensing
- Extreme, transient bursts of energy in the auroral ionosphere. I. Predictive radar tracking
- Dark Energy Survey Year 6 Results: improved mitigation of spatially varying observational systematics with masking for the MagLim++ lens sample
- Searching for gravitational wave echo with group equivariant neural posterior estimation
BibTeX
@misc{doux2026dark,
title = {{Dark Energy Survey Year 6 Results: fast and interpretable posterior predictive checks for correlated cosmic probes}},
author = {C Doux and J Muir and A Ferté and M Raveri and D Sanchez Cid and N Weaverdyck and E.J Baxter and W d'Assignies and T Karwal and A Amon and D Anbajagane and F Andrade-Oliveira and M.R Becker and J Blazek and A Carnero Rosell and C Chang and J.M Coloma-Nadal and M Crocce and J Derose and S Dodelson and G Giannini and E Krause and E Legnani and N Maccrann and J Mena-Fernández and A Porredon and J Prat and C Preston and M Rodriguez-Monroy and P Rogozenski and R Rosenfeld and S Samuroff and E Sanchez and I Sevilla-Noarbe and M Soares-Santos and C To and M.A Troxel and M Yamamoto and B Yin and J Zuntz and S Allam and M Aguena and D Bacon and D Brooks and R Camilleri and R Cawthon and L.N da Costa and S Desai and P Doel and K Glazebrook and K Honscheid and D Huterer and K Kuehn and J.L Marshall and R Miquel and M Paterno and A.A Plazas Malagón and M Smith and E Suchyta and D.L Tucker},
howpublished = {HAL},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.03447},
eprint = {2610.03447},
archivePrefix = {arXiv},
url = {https://hal.science/hal-05785801v1},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Dark Energy Survey Year 6 Results: fast and interpretable posterior predictive checks for correlated cosmic probes
AU - C Doux
AU - J Muir
AU - A Ferté
AU - M Raveri
AU - D Sanchez Cid
AU - N Weaverdyck
AU - E.J Baxter
AU - W d'Assignies
AU - T Karwal
AU - A Amon
AU - D Anbajagane
AU - F Andrade-Oliveira
AU - M.R Becker
AU - J Blazek
AU - A Carnero Rosell
AU - C Chang
AU - J.M Coloma-Nadal
AU - M Crocce
AU - J Derose
AU - S Dodelson
AU - G Giannini
AU - E Krause
AU - E Legnani
AU - N Maccrann
AU - J Mena-Fernández
AU - A Porredon
AU - J Prat
AU - C Preston
AU - M Rodriguez-Monroy
AU - P Rogozenski
AU - R Rosenfeld
AU - S Samuroff
AU - E Sanchez
AU - I Sevilla-Noarbe
AU - M Soares-Santos
AU - C To
AU - M.A Troxel
AU - M Yamamoto
AU - B Yin
AU - J Zuntz
AU - S Allam
AU - M Aguena
AU - D Bacon
AU - D Brooks
AU - R Camilleri
AU - R Cawthon
AU - L.N da Costa
AU - S Desai
AU - P Doel
AU - K Glazebrook
AU - K Honscheid
AU - D Huterer
AU - K Kuehn
AU - J.L Marshall
AU - R Miquel
AU - M Paterno
AU - A.A Plazas Malagón
AU - M Smith
AU - E Suchyta
AU - D.L Tucker
PY - 2026
DA - 2026/10/02
PB - HAL
DO - 10.48550/arxiv.2610.03447
UR - https://hal.science/hal-05785801v1
AB - We present a fast and numerically stable framework for conducting posterior predictive distribution (PPD) tests of internal consistency in correlated cosmological probes. Our method, developed as a validation check for the Dark Energy Survey (DES) Year 6 combined analysis of weak lensing and galaxy clustering, employs a Gaussian-mixture model to efficiently approximate the high-dimensional PPD and introduces a scalar consistency metric, Δ_PPD, defined as the fraction of the PPD at lower probability density than the observed data. This method avoids challenges related to calibration and numerical stability that impacted PPD methods previously used in DES Year 3. After demonstrating performance with a toy model, we confirm that when applied to DES Year 3 measurements, Δ_PPD identifies the same inter-probe tensions as published, calibrated PPD results while requiring only minutes per test. We further validate the method on several noise realizations of simulated DES Y6 analyses, demonstrating that Δ_PPD responds coherently to injected inconsistency, while highlighting two effects which can complicate the interpretation of PPD tests for increasingly complex models. First, internal tensions that can be absorbed as shifts in nuisance parameters become more difficult to detect. Second, poorly constrained directions in parameter space can induce projection effects in PPDs for disjoint sets of observables, producing reports of tension even for consistent data. We conclude that Δ_PPD provides a practical PPD-based diagnostic for multiprobe unblinding, to be interpreted with care alongside other validation checks. This framework has already played an important role in DES analyses and is well suited for application to forthcoming Stage-IV surveys such as LSST, Euclid, and Roman, where fast, robust internal consistency tests will be essential.
ER -