Big Data Cosmology meets AI

An Invitation to Backpropagate Through the Origins of the Universe

 

IAIFI Fellow

Carol Cuesta-Lazaro

 

Video Credit: N-body simulation Francisco Villaescusa-Navarro

The era of Big Data Astrophysics

1-Dimensional

Machine Learning

Cosmic Cartography

Galaxy Clustering

Galaxy Imaging

Lensing

Cosmic Microwave Background

Gravitational Waves

Time domain

Early Universe Inflation

{\delta_\mathrm{Initial}}

Late Universe

What's the Universe made of?

Evolution

{\delta_\mathrm{Final}}
\color{darkgray}{\Omega_m}

Dark matter

Dark energy

\color{darkgreen}{w_0, w_a}
\color{purple}{f_\mathrm{NL}}

Non-Gaussianity

...

Multifield Inflation

Initial Conditions

The Universe's forward model

Observables

Why Astrophysics is hard 101

Dataset Size = 1 

Can't poke it in the lab 

Simulations

Bayesian statistics

How well can we simulate the Universe?

Very interested on ideas in the area of model mispecification!

How do we learn what is the robust information?

Simulating dark matter is easy!

"Atoms" are hard" :(

2
3

Hybrid ML - Physics Simulators

Unsupervised searches

1

Cosmological (field level) Inference for Galaxy Surveys

DESI

High dimensional data 

x

Unknown

p(x|\mathcal{C})

Simple summary statistic 

s
p(s|\mathcal{C})

estimated with Perturbation Theory

Probability pair of galaxy

Pair separation

\theta

Forward Model

Parameters

Observable

x

Likelihood

p(\mathcal{\theta}|x)

Simulator

+ MCMC hammer

\color{darkgray}{\Omega_m}, \color{darkgreen}{w_0, w_a},\color{purple}{f_\mathrm{NL}}\, ...

Dark matter

Dark energy

Inflation

Perturbation Theory

Pen and paper

p(x|\mathcal{\theta})

+ Density Estimation

+ Sampler

p(\mathcal{\theta}|x) =
p(\theta) / p(x)
"A point cloud approach to generative modeling for galaxy surveys at the field level" 
Cuesta-Lazaro and Mishra-Sharma 

arXiv:2311.17141

Base Distribution

Target Distribution

  • Sample
  • Evaluate

Siddharth Mishra-Sharma

Long range correlations

Huge pointclouds (20M)

Homogeneity and isotropy

Fixed Initial Conditions

 Varying Cosmology

Trained on only 5000 positions!

Real observations 20 Million points :(

Learning in 5000 dimensions with only 2000 simulations

Symmetries?

Julia Balla

Loss

Step

Pair counting

MP GNN

Hierarchical

Symmetries

"GalaxyBench: A Long- and Short-Range Benchmark for Symmetry-Preserving Data Processing" Balla et al (in prep.)
p_\phi(\delta_\mathrm{ICs}|\delta_\mathrm{Today})

1 to Many:

Can we run the Universe backwards?

Today

Initial Conditions

"Probabilistic Forecasting with Stochastic Interpolants and Follmer Processes" 
Chen et al 

arXiv:2403.13724

x_t = \alpha_t x_0 + \beta_t x_1 + \sigma_t W_t
d X_t = b_t(X_t,x_0) dt + \sigma_t dW_t

Sampling SDE

Interpolant

\mathcal{L} = \int_0^1 \mathbb{E} \left[|\hat{b}_t(x_t, x_0) - R_t|^2 \right]
R_t = \dot{\alpha_t} x_0 + \dot{\beta_t} x_1 + \dot{\sigma_t}W_t
x_1 \sim p(x_1|x_0)

Drift

Regression loss

"Probabilistic Forecasting with Stochastic Interpolants and Follmer Processes" 
Chen et al 

arXiv:2403.13724

Current model is not very good when ran forwards!

3D U-Nets are annoying :(

True

Initial

Final

Predicted

Can we run larger simulations? (Observable volumes)

At high resolution?

Faster?

All this works depends on simulations, but...

Thousands of them?

Hybrid Physical / ML simulators

\frac{\mathrm{d} \mathbf{x}}{\mathrm{d} a } = \frac{1}{a^3 E(a)}\mathbf{v}
\frac{\mathrm{d} \mathbf{v}}{\mathrm{d} a } = \frac{1}{a^2 E(a)}\mathbf{F}(\mathbf{x},a)
\mathbf{F}(\mathbf{x},a) = \frac{3 \Omega_m}{2} \nabla \phi^\mathrm{PM}(\mathbf{x})

Gravitational evolution ODE

Particle-mesh

"Nbodyify: Adaptive mesh corrections for PM simulations" Cuesta-Lazaro, Modi in preps

Particle-mesh

Full Nbody

\mathbf{F}_\theta(\mathbf{x},a) = \frac{3 \Omega_m}{2} \nabla \left[\phi^\mathrm{PM}(\mathbf{x}) + \phi^\mathrm{corr}_\theta(\mathbf{x}, a, \phi^\mathrm{PM}, \delta^\mathrm{PM}) \right]

Hybrid Simulator - on the fly

\frac{\mathrm{d} \mathbf{x}}{\mathrm{d} a } = \frac{1}{a^3 E(a)}\mathbf{v}
\frac{\mathrm{d} \mathbf{v}}{\mathrm{d} a } = \frac{1}{a^2 E(a)}\mathbf{F}(\mathbf{x},a)

Gravitational evolution ODE

Trained to match particle velocities and positions: DIFFERENTIABLE

\mathcal{L} = \sum_t \left(x_t^{\rm pred} - x_t^{\rm HR}\right)^2
\delta_\mathrm{LR}
\phi_\mathrm{LR}

Density

Gravitational Potential

1. CNN

2. Read features at position using attention

\mathbf{F}_\theta(\mathbf{x},a) = f_\theta(h_\theta(\mathbf{x}), a)

3. Compute force correction

4. Run corrected simulation

Learn features

h_\theta(\mathbf{x})
h

Particle-mesh

Full Nbody

Hybrid ML-Simulator

"Nbodyify: Adaptive mesh corrections for PM simulations" Cuesta-Lazaro, Modi in preps

Video credit: Francisco Villaescusa-Navarro

Gas density

Gas temperature

Finding missing physics with differentiable simulators?

What is the space of plausible solutions and how do we search it?

Differentiable Galaxies ODEs

Humans best bet

\frac{d \mathrm{Galaxies}}{dt} = \phi(\mathrm{Dark Matter}(t))
\color{blue}{+ \phi_\theta(?)}

Neural Network correction

Are there problems in cosmology that bypass a forward model?

Parity violation cannot be originated by gravity

7 \sigma
x
\mathrm{Mirror}(x)
"Measurements of parity-odd modes in the large-scale 4-point function of SDSS..." 
Hou, Slepian, Chan arXiv:2206.03625
?
1 \sigma
"Could sample variance be responsible for the parity-violating signal seen in the BOSS galaxy survey?"
 Philcox, Ereza arXiv:2401.09523
x
\mathrm{Mirror}(x)
\mathrm{max} \, \left( f_\theta(x) - f_\theta(\mathrm{Mirror}(x)) \right)

Train

Test

Me: I can't wait to work with observations

Me working with observations:

Finding interesting objects:

Very small galaxies (dwarf galaxies)

Interesting in Astrophysics: How we define an anomaly and how do we find it?

Background

Region of Interest

\frac{p_\mathrm{RoI}(x_\mathrm{star})}{p_\mathrm{BckG}(x_\mathrm{star})}

Conclusions

1. There is a lot of information in galaxy surveys that ML methods can access

2. We can tackle high dimensional inference problems so far unatainable

3. Our ability to simulate will limit the amount of information we can extract

Hybrid simulators, forward models, robustness

Unsupervised problems: parity violation

Finding anomalies for new physics?

Finding the Initial Conditions of the Universe, let's get creative!

Field level inference

MIT

By carol cuesta

MIT

  • 51