Automatic physical inference with information maximising neural networks Physical Review D 97 , 083004 arXiv:1802.03537 github:information_maximiser Tom Charnock Institut d’Astrophysique de Paris charnock@iap.fr DOI 10.5281/zenodo.1175196 DOI 10.5281/zenodo.1175196
How would we like to do inference? 𝓆 ( d ) arXiv:1802.03537, Physical Review D 97 , 083004, charnock@iap.fr Tom Charnock ℒ( d |𝜄) 20.0 17.5 Simulated quasar flux [Photon counts] 15.0 12.5 10.0 7.5 5.0 2.5 410 420 430 440 450 460 [nm] ( | d ) 𝒬 (𝜄| d ) = ℒ( d |𝜄) 𝓆 (𝜄) 4 2 0 2 4 Amplitude of scalar perturbation scaling ln DOI DOI 10.5281/zenodo.1175196 10.5281/zenodo.1175196
How can we do inference without a likelihood? arXiv:1802.03537, Physical Review D 97 , 083004, charnock@iap.fr Tom Charnock ∶ d → 𝓎 {𝜄|𝜄 ↶ 𝓆 (𝜄)} 20.0 20 17.5 Simulated quasar flux Simulated quasar flux [Photon counts] [Photon counts] 15.0 15 12.5 10.0 10 7.5 5 5.0 2.5 0 410 420 430 440 450 460 410 420 430 440 450 460 [nm] [nm] 30 Compressed summary 20 10 ( | x ) 0 10 20 30 4 2 0 2 4 4 2 0 2 4 Amplitude of scalar perturabtion scaling Amplitude of scalar perturbation scaling ln ln DOI DOI 10.5281/zenodo.1175196 10.5281/zenodo.1175196
How can we fjnd the summaries? Tom Charnock charnock@iap.fr Physical Review D 97 , 083004, arXiv:1802.03537, DOI DOI 10.5281/zenodo.1175196 10.5281/zenodo.1175196
Automatic physical inference with information maximising neural networks Tom Charnock charnock@iap.fr Physical Review D 97 , 083004, arXiv:1802.03537, ▶ Simulate data at a fjducial parameter value ▶ Train the network to increase the Fisher information ▶ Compress real data using trained network ▶ Do ABC with the optimally compressed summaries DOI DOI 10.5281/zenodo.1175196 10.5281/zenodo.1175196
Automatic physical inference with information maximising neural networks Physical Review D 97 , 083004 arXiv:1802.03537 github:information_maximiser Tom Charnock Institut d’Astrophysique de Paris charnock@iap.fr DOI 10.5281/zenodo.1175196 DOI 10.5281/zenodo.1175196
Recommend
More recommend