Cosmology's Midlife Crisis

IAIFI Fellow, MIT

Carolina Cuesta-Lazaro

Art: "A Bar at the Folies-Bergère" by Édouard Manet

Embracing the Machine Learning Makeover

Robust

Very few assumptions about the mechanisms of galaxy formation

Generalizable

Physical Inductive biases

Symmetries: Homogeneous and isotropic Universe on large scales

Precise

(On large scales)

Information Loss

Galaxy formation = Nuisance

Due to being fast

The perks of Perturbation Theory

Small Scales

High dimensional inference

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

p(\mathcal{G}|\delta_\mathrm{dm}, \mathcal{A}, z)

Galaxy properties

Astro parametrization

DM density + velocity fields

Target

1. Robust: Mapping dark matter into anything

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

MTNG

Astrid

UniverseMachine

CAMELS

Data

Hydro Sims

SAMS

Santa Cruz

p(x_0)
p(x_1)

Dark Matter

"Baryons"

x_t = t x_0 + (1-t) x_1
\frac{d x_t}{dt} = u_t(x_t)

Transformer

\mathcal{G} = \{x, v, \mathrm{SFR}, M_\star \}

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

t

How many parameters can all these simulations share?

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

MTNG

Astrid

UniverseMachine

CAMELS

Data

Hydro Sims

SAMS

Santa Cruz

Can we fit the remaining to observations directly?

2. Generalizable: Representation learning

Learning to represent feedback

p(\mathcal{G}|\mathcal{C}, \mathcal{A})

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

SIMBA

TNG

Astrid

Different Supernovae and AGN feedback mechanisms

+ Different parametrisations

Finding anomalies

Z

\frac{p(x)}{p_{\Lambda\mathrm{CDM}}(x)}
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

3. Physical Inductive biases

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

Same model architecture (transformer) on direct inference wouldn't work at all!

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

p(x|\mathcal{C})
\mu_\theta(z_t,t)
p(z_T)

Making homogeneous and isotropic Universes

Base Distribution

Denoiser

=
p(
)
p(
p(
)

Invariant

Equivariant

=
p(
)
p(
=
p(
)
p(

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

All learnable functions

All learnable functions constrained by your data

All Equivariant functions

More data efficient!

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

Equivariant models

Non-equivariant

SEGNN: arXiv:2110.02905

 

NequIP: arXiv:2101.03164

 

 

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

Robust

Very few assumptions about the mechanisms of galaxy formation

Generalizable

Physical Inductive biases

Symmetries: Homogeneous and isotropic Universe on large scales

Precise

(On large scales)

Information Loss

Galaxy formation = Nuisance

Due to being fast

The perks of Machine Learning

Small Scales

High dimensional inference

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

Learned

Learned representations

Symmetries: Homogeneous and isotropic Universe on large scales

as long as we can sufficiently simulate

Interesting Physics

Perturbation Theory's retirement plan:

The Physics of Machine Learning

 

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

Grokking

Double Descent

Emergent abilities

Perturbation Theory's Midlife Crisis

Embracing the Machine Learning Makeover

IAIFI Fellow

Carolina Cuesta-Lazaro

Art: "Melancholy" by Edvard Munch

Perturbation Theory's Midlife Crisis

Embracing the Machine Learning Makeover

IAIFI Fellow

Carolina Cuesta-Lazaro

Art: "Midlife crisis" by Nik Ad

Fixed Initial Conditions

 Varying Cosmology

Carolina Cuesta-Lazaro IAIFI/MIT @ Split 2024

  • "Cosmic Plot Twist: When Algorithms Crash the Perturbation Party"
  • "Perturbation Theory's Retirement Plan: 
  • Breaking Up with Perturbation Theory: It's Not You, It's AI

IAIFI Fellow, MIT

Carolina Cuesta-Lazaro

Art: "A Bar at the Folies-Bergère" by Édouard Manet

Embracing the Machine Learning Makeover