CONNECT: a neural network based framework for emulating cosmological observables and cosmological parameter inference
Problem
Bayesian cosmological parameter inference requires 10^5-10^6 theoretical model evaluations, each taking tens of CPU core-seconds, making the process very computationally expensive.
Why it matters
Parameter inference underpins cosmological results from observational datasets, and its computational cost limits feasible analyses.
What the AI did
CONNECT, an iteratively trained neural network, emulates the observables usually computed by CLASS as a plug-in for the MontePython sampler, using a novel algorithm to pick favourable training points in parameter space.
Results
Required CLASS evaluations reduced by two orders of magnitude compared to a traditional inference run.
Impact
Stated to cut expensive model evaluations by two orders of magnitude for cosmological inference.
Record
| Venue | Journal of Cosmology and Astroparticle Physics |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Andreas Nygaard, Emil Brinch Holm, Steen Hannestad, Thomas Tram |
| Citations | 49 (OpenAlex, at import time) |
| References | 45 |
| Model or method | Iteratively trained neural network emulator (CONNECT) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1088/1475-7516/2023/05/025 |
| Open access | hybrid |
| Keywords | cosmology, neural network emulator, Bayesian inference, computational speedup, MontePython |
Importance rating
significant — thousands to millions affected
Bayesian parameter inference is an essential tool in modern cosmology; abstract quantifies the computational burden (10^5-10^6 models, tens of CPU core-seconds each).
More in Physics & Astronomy
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | Magnetic control of tokamak plasmas through deep reinforcement learning | 2022 | 739 |
7
|
| 2 | Solving the quantum many-body problem with artificial neural networks | 2017 | 2,205 |
7
|
| 3 | Learning high-accuracy error decoding for quantum processors | 2024 | 90 |
7
|
| 4 | Optimizing Quantum Error Correction Codes with Reinforcement Learning | 2019 | 158 |
7
|
| 5 | Machine-learning-assisted correction of correlated qubit errors in a topological code | 2018 | 141 |
7
|