FlopPITy 0.0.3

normalising Flow exoPlanet Parameter Inference Toolkyt

This is a code to perform retrievals on spectra of substellar objects using SNPE-C (citation) and neural spline flows (citation).

• Known issues:

  • Requires python<3.13 and numpy<2:

    conda create -n name python==3.12

Basic usage:

• First, import FlopPITy:

from floppity import Retrieval
from floppity.simulators import read_ARCiS_input, ARCiS

• Now you can initialize the retrieval class with a simulator. A python wrapper for ARCiS comes built-in (you need to install ARCiS on your own tho):

R = Retrieval(ARCiS)

• Read in observations and define parameters to retrieve:

R.get_obs(['path/to/obs_0', 'path/to/obs_1',..., 'path/to/obs_n'])
    
R.add_parameter(par_0, min, max)
R.add_parameter(par_1, min, max)
...
R.add_parameter(par_m, min, max)

• For ARCiS, the observations and parameters can be read from the ARCiS input file:

pars, obs_list = read_ARCiS_input('path/to/ARCiS/input')
R.get_obs(obs_list)
R.parameters=pars

• For retrievals using ARCiS, the input file and output directory need to be passed in a dictionary:

ARCiS_kwargs= dict(
                    input_file = arcis_input,
                    output_dir = 'path/to/output',
                  )

• You can now run the retrieval, indicating the number of rounds and samples per round:

R.run_retrieval(n_rounds=10, n_samples=1000, simulator_kwargs=ARCiS_kwargs)

• Great! You can now inspect your posterior:

fig = R.plot_corner()

Writing a simulator

Writing a simulator to work for FlopPITy is relatively straightforward. All that's needed is a function that takes in observations and parameters and returns spectra. The spectra need to be returned in a dictionary where each key represents each of the observations simulated (e.g. simulated[0] contains PRISM spectra and simulated[1] contains MIRI/LRS spectra):

def simulator(obs, parameters, **kwargs):
    wvl_0 = obs[0][:,0]
    wvl_1 = obs[1][:,0]
    ...
    wvl_n = obs[n][:,0]

    spectra={}
    spectra[0] = # array of shape (ndims, len(wvl_0))
    spectra[1] = # array of shape (ndims, len(wvl_1))
    ...
    spectra[n] = # array of shape (ndims, len(wvl_n))

    return spectra

Advanced options:

• Additional post processing parameters (currently RV, vrot, offset and scaling) can be added, for example:

R.add_parameter('RV', -100, 100, post_process=True) # km/s