Carol Cuesta-Lazaro (IAIFI Fellow)
in collaboration with Siddarth Mishra-Sharma
Generative models for the Large Scale Structure
ML for Large Scale Structure:
Carol's wish list
Generative models
Learn p(x)
Evaluate the likelihood of a 3D map, as a function of the parameters of interest
1
Combine different galaxy properties (such as velocities and positions)
2
Sample 3D maps from the posterior distribution
3
Explicit Density
Implicit Density
Tractable Density
Approximate Density
Normalising flows
Variational Autoencoders
Diffusion models
Generative Adversarial Networks
The zoo of generative models
Reverse diffusion: Denoise previous step
Forward diffusion: Add Gaussian noise (fixed)
Diffusion models
A person half Yoda half Gandalf
Diffusion on point clouds
Reverse diffusion: Denoise previous step
Forward diffusion: Add Gaussian noise (fixed)
Cosmology
Halo Mass Function
Velocity
Mean pairwise velocity
Evidence Lower Bound
Distance to true posterior
Find
1. ELBO is a lower bound of the evidence
2. Maximising ELBO = Minimising KL
Maximise ELBO to maximise ev/likelihood
Maximise ELBO to approximate true posterior
+ Galaxy formation
+ Observational systematics (Cut-sky, Fiber collisions)
+ Lightcone, Redshift Space Distortions....
Forward Model
N-body simulations
Observations
Copy of Copy of Copy of deck
By carol cuesta
Copy of Copy of Copy of deck
- 250