DeepPlanck

paper link : https://arxiv.org/abs/2209.10333

Contact

Please for any inquiry or comment contact the author Daniel de Andres. email: daniel.deandres@uam.es

Dependences

tensorflow 2.7.0

Introduction

This repository contains the data product of the project DeepPlanck, i.e. Deep learning mass estimates for Planck PSZ2 clusters. The file containing the estimated masses is DeepPlanck.csv. The file contains 6 columns: cluster, Planck_mass, CNN_mass, uncertainties,redsfhit, Y_sz and uncertainties. The cluster column corresponds to the name of the cluster in the PSZ2 catalog https://heasarc.gsfc.nasa.gov/W3Browse/all/plancksz2.html. You can read the csv file using, for instance, pandas in Python.

import pandas as pd
pd.read_csv('DeepPlanck.csv')

IMPORTANT: masses are given as the decimal logarithm of the mass in solar masses. Moreover, Y_500 provided is computed from M_CNN. Following eq.(4), the quantity provided in the table 'Y_500' is

$$ E(z)^{-2/3}\left[\frac{D_{A}^{2}(z)Y}{10^{-4} \text{Mpc} ^{2}}\right] $$

Weights

We also provide the trained CNN weights for other applications such us transfer learning. The clusters masses can be predicted using the module BaseModel.PredictMass(X,z). The notebook Example.ipynb explains how to use this module.