Question 1

How do I compute a Fisher matrix with jax-cosmo?

Accepted Answer

Use jax.hessian on a likelihood function defined with jax-cosmo observables, as demonstrated in the TL;DR snippet. This automatically computes exact second derivatives without finite differences, ideal for cosmological parameter inference.

Question 2

jax-cosmo vs CAMB: which is better for gradient calculations?

Accepted Answer

jax-cosmo excels in differentiable cosmology with built-in automatic differentiation, while CAMB is a mature, non-differentiable code for standard calculations. Choose jax-cosmo if you need gradients for optimization or Fisher matrices; use CAMB for validated, broad-scope simulations without JAX.

Question 3

How to install jax-cosmo with GPU support?

Accepted Answer

First install JAX with GPU support following JAX's documentation, then run 'pip install jax-cosmo'. The library leverages JAX's XLA backend, so ensure your JAX installation is configured for CUDA or TPU if needed.

Question 4

Does jax-cosmo support weak lensing probes?

Accepted Answer

Yes, it includes angular power spectra computations for probes like weak lensing, as indicated in the likelihood example. However, check the documentation for specific implementations, as coverage may evolve with community contributions.

Question 5

How can I add custom cosmological models to jax-cosmo?

Accepted Answer

Extend the API by implementing differentiable functions using JAX operations, keeping code readable per the design philosophy. Refer to the design document and contribute via GitHub, as the project encourages community additions.

Question 6

Is jax-cosmo suitable for CMB analysis?

Accepted Answer

It supports angular power spectra relevant to CMB studies, but for comprehensive CMB tools like lensing or foregrounds, you might need to integrate or extend it, as it focuses on core differentiable functions rather than full CMB pipelines.

jax-cosmo

What is jax-cosmo?

Overview

Use Cases

Best For

Related Projects

Found a gem we're missing?

Not Ideal For

Pros & Cons

Pros

Cons

Frequently Asked Questions