FlowPM Blog Post Out of Date

[bhorowitz]

The flowpm_blog.ipynb file which is lined from the main README still uses tensorflow 1.x rather than 2.6. I'm not sure what the highest 2.x flowpm has been tested on so far...

[EiffL]

so yes, the code from the blog post is out of date, but I think it should still work, no?

We just need newer examples

[EiffL]

@bhorowitz if you want to see a recent demo, have a look at this notebook: https://colab.research.google.com/drive/1k6cTniiRZGF4F_RAA4XDuBCY16FxyUcZ?usp=sharing

This is doing a lensing sim, but you'll get the picture.

You want to encapsulate your function in something like this:

@tf.function
def make_map(Omega_c=0.2589, sigma8=0.8159): 
  # Instantiates a cosmology with desired parameters
  cosmology = flowpm.cosmology.Planck15(Omega_c=Omega_c, sigma8=sigma8)
  ....

[adamrouhiainen]

The minimal example flowpm_demo.ipynb is outdated, but it should work if the 4th cell is replaced with

import flowpm
from flowpm.tfpower import linear_matter_power

@tf.function
def simulation(om, s8):
    cosmo = flowpm.cosmology.Planck15(Omega_c=om, sigma8=s8)

    stages = np.linspace(a0, af, n_steps, endpoint=True) #time-steps for the integration

    initial_conditions = flowpm.linear_field(N,          # size of the cube
                                             L,          # Physical size of the cube
                                             lambda k: tf.cast(linear_matter_power(cosmo, k), tf.complex64), # Initial powerspectrum
                                             batch_size=16)

    # Sample particles
    state = flowpm.lpt_init(cosmo, initial_conditions, a0)   

    # Evolve particles down to z=0
    final_state = flowpm.nbody(cosmo, state, stages, 32)

    # Retrieve final density field
    final_field = flowpm.cic_paint(tf.zeros_like(initial_conditions), final_state[0])

    return final_field