A general non-perturbative field-level posterior is constructed and expanded around its Gaussian limit to express Fisher information in terms of connected correlators, recovering standard results for power spectrum and bispectrum while quantifying compression losses.
Bayesian physical reconstruction of initial conditions from large scale structure surveys
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
abstract
We present a fully probabilistic, physical model of the non-linearly evolved density field, as probed by realistic galaxy surveys. Our model is valid in the linear and mildly non-linear regimes and uses second order Lagrangian perturbation theory to connect the initial conditions with the final density field. Our parameter space consists of the 3D initial density field and our method allows a fully Bayesian exploration of the sets of initial conditions that are consistent with the galaxy distribution sampling the final density field. A natural byproduct of this technique is an optimal non-linear reconstruction of the present density and velocity fields, including a full propagation of the observational uncertainties. A test of these methods on simulated data mimicking the survey mask, selection function and galaxy number of the SDSS DR7 main sample shows that this physical model gives accurate reconstructions of the underlying present-day density and velocity fields on scales larger than ~6 Mpc/h. Our method naturally and accurately reconstructs non-linear features corresponding to three-point and higher order correlation functions such as walls and filaments. Simple tests of the reconstructed initial conditions show statistical consistency with the Gaussian simulation inputs. Our test demonstrates that statistical approaches based on physical models of the large scale structure distribution are now becoming feasible for realistic current and future surveys.
citation-role summary
background 2
citation-polarity summary
fields
astro-ph.CO 8years
2026 8roles
background 2polarities
background 2representative citing papers
Generative models for cosmological field-level inference can reproduce posterior means and cross-correlations yet fail to capture correct uncertainty geometry when validated against HMC reference samples.
Manticore-Deep uses tiled Bayesian field-level inference on SDSS and BOSS data to produce posterior ensembles of 3D cosmic fields that are consistent with LCDM and validated by 7.4σ CMB lensing and 3.5σ kSZ detections.
Develops a galaxy-biasing formalism for cosmic variance using perturbation theory and tests it on the non-linear BAO shift against N-body simulations.
AI trained on LambdaCDM simulations reconstructs the cosmic vorticity field from SDSS galaxies, revealing coherent vortical structures consistent with the standard model and correcting redshift-space distortions.
50 constrained simulations of Coma cluster analogues reproduce the observed radial X-ray surface brightness and Compton-y profiles within the scatter expected from environment and assembly history.
The paper defines a continuous Shannon entropy from the tidal tensor eigenvalue sign patterns of the cosmic web and shows that its redshift evolution constrains the linear growth rate f(z) complementary to redshift-space distortions.
Bayesian hierarchical modeling of photometric redshifts in KiDS+VIKING-450 raises S8 to 0.756 ± 0.039 and reduces Planck tension to 1.9σ.
citing papers explorer
-
On the Relation Between Field-Level Posteriors, Correlators, and their Likelihoods
A general non-perturbative field-level posterior is constructed and expanded around its Gaussian limit to express Fisher information in terms of connected correlators, recovering standard results for power spectrum and bispectrum while quantifying compression losses.
-
Learning the Universe: Posterior Reliability of Neural Generative Models in High-Dimensional Field-Level Inference of Cosmic Initial Conditions
Generative models for cosmological field-level inference can reproduce posterior means and cross-correlations yet fail to capture correct uncertainty geometry when validated against HMC reference samples.
-
The Manticore Project II: Bayesian digital twins of cosmic structure across the SDSS and BOSS volumes
Manticore-Deep uses tiled Bayesian field-level inference on SDSS and BOSS data to produce posterior ensembles of 3D cosmic fields that are consistent with LCDM and validated by 7.4σ CMB lensing and 3.5σ kSZ detections.
-
Cosmic variance or galaxy bias? Disentangling finite-volume and galaxy formation effects in cosmological analysis
Develops a galaxy-biasing formalism for cosmic variance using perturbation theory and tests it on the non-linear BAO shift against N-body simulations.
-
Closing the Observational Gap in Cosmic Dynamics: AI-Enabled Reconstruction of the Universe's Vorticity and Rotational Flow Morphology
AI trained on LambdaCDM simulations reconstructs the cosmic vorticity field from SDSS galaxies, revealing coherent vortical structures consistent with the standard model and correcting redshift-space distortions.
-
Learning the Universe: Constrained simulations of the Coma galaxy cluster -- I. Radial X-ray and Compton-y signatures
50 constrained simulations of Coma cluster analogues reproduce the observed radial X-ray surface brightness and Compton-y profiles within the scatter expected from environment and assembly history.
-
Information Content of the Cosmic Web
The paper defines a continuous Shannon entropy from the tidal tensor eigenvalue sign patterns of the cosmic web and shows that its redshift evolution constrains the linear growth rate f(z) complementary to redshift-space distortions.
-
KiDS+VIKING-450 cosmology with Bayesian hierarchical model redshift distributions
Bayesian hierarchical modeling of photometric redshifts in KiDS+VIKING-450 raises S8 to 0.756 ± 0.039 and reduces Planck tension to 1.9σ.