REVIEW 19 cited by
NAUTILUS: boosting Bayesian importance nested sampling with deep learning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
We introduce a novel approach to boost the efficiency of the importance nested sampling (INS) technique for Bayesian posterior and evidence estimation using deep learning. Unlike rejection-based sampling methods such as vanilla nested sampling (NS) or Markov chain Monte Carlo (MCMC) algorithms, importance sampling techniques can use all likelihood evaluations for posterior and evidence estimation. However, for efficient importance sampling, one needs proposal distributions that closely mimic the posterior distributions. We show how to combine INS with deep learning via neural network regression to accomplish this task. We also introduce NAUTILUS, a reference open-source Python implementation of this technique for Bayesian posterior and evidence estimation. We compare NAUTILUS against popular NS and MCMC packages, including EMCEE, DYNESTY, ULTRANEST and POCOMC, on a variety of challenging synthetic problems and real-world applications in exoplanet detection, galaxy SED fitting and cosmology. In all applications, the sampling efficiency of NAUTILUS is substantially higher than that of all other samplers, often by more than an order of magnitude. Simultaneously, NAUTILUS delivers highly accurate results and needs fewer likelihood evaluations than all other samplers tested. We also show that NAUTILUS has good scaling with the dimensionality of the likelihood and is easily parallelizable to many CPUs.
Forward citations
Cited by 19 Pith papers
-
Unveiling and Characterising Ubiquitous Nitrogen Enhancement in $6 \leq z \leq 10$ Galaxies with JWST Spectroscopy
Stacked JWST/NIRSpec spectra of 135 z=6-10 galaxies show supersolar N/O that is highest after a star-formation lull, consistent with delayed AGB enrichment and pristine gas inflows.
-
KiDS-1000: Improved constraints on cosmology, intrinsic alignments and baryonic feedback from clipped cosmic shear
Combining clipped and standard shear correlation functions from KiDS-1000 tightens S8 by 16% and w0 by 24%, yielding S8=0.724±0.027 and w0=-1.24+0.26-0.28, consistent with ΛCDM.
-
First detection of ultra-fast outflows in a quiescent galaxy
First detection of an ultra-fast outflow (v≈0.07c) in a quiescent galaxy quenched ~9 Gyr ago, suggesting AGN winds can maintain quiescence independent of global star formation.
-
A Physically Driven Parameterisation of Multidimensional Atmospheres: Application to the JWST Phase Curve of WASP-121b
A 3D temperature parameterization derived from energy-balance timescales, fit to JWST phase-curve spectra of WASP-121b, recovers a day–night asymmetric atmosphere with a nightside methane enhancement and hints of magn...
-
Web-Halo Model Peak-Background Split (WHM-PBS): halo bias as a distribution, not a number
Halo bias is derived as an environment-inherited probability distribution whose spread simultaneously produces halo stochasticity, EFT prior bands, and the measured bias–concentration inversion at M ≈ 1.7×10¹³ h⁻¹M☉.
-
Impact and measurability of linear relativistic effects in galaxy surveys
Neglecting linear GR effects biases f_NL at 1–3σ for Euclid/SPHEREx in SFB forecasts; multi-tracer improves Doppler detection and weakly breaks b_ϕ f_NL degeneracy.
-
Testing General Relativity with Individual Supermassive Black Hole Binaries
Beyond-GR polarization modes in a single supermassive black hole binary show up linearly (not quadratically) in pulsar-timing cross-correlations, massive-graviton dispersion shifts antenna patterns and pulsar-term pha...
-
Fingerprints of Individual Supermassive Black Hole Binaries in Pulsar Timing Arrays
A single supermassive black hole binary imprints a deterministic, direction-dependent correlation fingerprint on pulsar timing arrays, enabling identification via cross-correlations.
-
The simple way to measure evolving dark energy without prior-volume effects
Absorbing the AP and growth amplitudes into EFTofLSS nuisance parameters removes prior-volume shifts and yields Stage III LSS-only evolving dark energy constraints aligned with DESI.
-
Dark Energy Survey Year 6 Results: Redshift Calibration of the MagLim++ Lens Sample
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.
-
Sampler-free gravitational wave inference using matrix multiplication
A new algorithm computes gravitational wave posterior distributions and evidence integrals by grid evaluation with matrix multiplications, avoiding stochastic samplers and running in minutes on a single CPU.
-
Progress toward the detection of the gravitational-wave background from stellar-mass binary black holes: a mock data challenge
A mock data challenge shows that a phase-coherent search for the binary black hole background can recover injected signal fractions in realistic noise, using new treatments of noise uncertainty, finite-duration effect...
-
Growth, geometry, and early-universe split of the matter density parameter $\Omega_{\rm m}$
Splitting Ω_m into geometry, growth, and early-universe regimes yields mutually compatible values, yet ΔΩ_m^{geo,early} is 2σ from zero under combined DES, Planck (scale-cut), DESI, Pantheon+, and RSD data.
-
Searching for signatures of inflationary massive fields in DESI Imaging data and Stage-V galaxy surveys
DESI Imaging yields no robust beyond-local PNG detection, while Stage-V LBG forecasts give σ(Δ)≈0.17–0.50 near the local limit for f_NL,Δ^fid=4 and an empirical link to σ(f_NL^loc).
-
Constraints on Logarithmic Model Extensions of Symmetric Teleparallel Gravity
Two new logarithmic f(Q) gravity models fit current cosmological data and predict contrasting, testable deviations in the effective gravitational coupling and gravitational-wave damping.
-
Attention-based Neural Network Emulators for Multi-Probe Data Vectors Part III: Modeling The Next Generation Surveys
A transformer-based emulator reproduces CAMB CMB TT, TE, and EE power spectra within cosmic variance errors across a wide Lambda-CDM parameter space, with outlier fractions below 10% for future survey configurations.
-
Dynamical Dark Energy or Modified Gravity? Signatures in Gravitational Wave Propagation
Reconstructing the dark energy density from DESI BAO and DESyr5 supernovae, then recasting it as f(Q) gravity, predicts a low-redshift gravitational wave damping ν≈0.18 (≳2σ from GR) only for the DESyr5 dataset.
-
Constraints on Dark Energy Models Using Late Universe Probes
Using late-universe probes only, the authors find all dark energy models remain within 1-2 sigma of Lambda CDM, which wins the Bayesian model comparison, and confirm that DESI's LRG1 and LRG2 points drive the dynamica...
-
Effect of Peak Absolute Magnitude of Type Ia Supernovae and Sound Horizon Values on Hubble Tension using DESI results
Using DESI DR1, Pantheon+ and cosmic chronometer data, the authors show that changing the Type Ia supernova absolute magnitude and sound-horizon priors moves the inferred H0 from about 73 to 68 km/s/Mpc, reducing the ...
Discussion (0). Continue with ORCID to comment.