Interlopers

This dataset was prepared for the analysis presented in Cagliari et al. (2025), which explores the correction of line interloper contamination in measured summary statistics using machine learning. We simulated two types of line interlopers:

  • “Inbox” interlopers, which are highly correlated with the target sample (displacing halos within the same snapshot).

  • “Outbox” interlopers, which have low correlation with the target sample (displacing halos from a different snapshot).

We generated contaminated catalogs with varying interloper fractions, using Friend-of-Friend catalogs with a \(512^3\) resolution for \(1000\) snapshots in the fiducial cosmology, as well as for the \(\Lambda\) CDM Latin hypercube and the Big Sobol Sequence.

For each simulation, we provide all measured statistics used in the analysis, including the target power spectrum, as well as the observed power spectrum and bispectrum.

The code used to contaminate the simulations and correct the summary statistics is available at the following link.

Contaminated Catalogues and Measured Statistics

We contaminated the snapshots at \(z=1\). Therefore, we give the catalogs and the statistics corresponding to that redshift alone. The fractions are always produced in the range \([0.01, 0.11]\).

We produced two types of interlopers:

  1. Inbox interlopers: the halos are shifted within the box at \(z=1\) by \(\delta d = 90 \, {\rm Mpc}/h\).

  2. Outbox interlopers: the interloper halos are shifted inside the box at \(z=1\) from the box at \(z=2\).

There are 3 directories:

  • inbox

  • outbox

  • Pk_contaminated (may be renamed interloper_statistics?)

(inbox and outbox may be moved into a folder named FoF_contaminated?)

Contaminated FoF Catalogs

The contaminated halo catalogs have the same characteristics as the original Quijote FoF catalogs (columns, naming, redshift, etc.).

The two directories (inbox and outbox) have the same internal structure.

  • ``fiducial``: \(2000\) contaminated FoF catalogs at \(z=1\) (groups_002) produced from \(1000\) FoF original Quijote catalogs and \(100\) fractions extracted from a Latin hypercube and are stored in the file fiducial/fractions-2000.txt.

    Original number

    Contaminated number

    0-499 0-499 500-999 500-999

    0-499 500-999 1000-1499 1500-1999

  • ``latin_hypercube``: \(10000\) catalogs at \(z=1\) (groups_002) produced from the \(2000\) Quijote original Latin hypercube cosmologies and \(2000\) fractions sampled from a Latin hypercube. Each cosmology has \(5\) fraction realizations. The folders are ordered as in the original Quijote hypercube, and in each folder, there are \(5\) directories (from 0 to 4) for the different fraction realizations. The fraction values for the cosmologies are stored in each cosmology folder in the file fractions.txt.

  • ``BSQ``: \(2^{15}\) catalogs at \(z=1\) (groups_006) from the original Quijote Big Sobol Sequence (BSQ) with \(2^{15}\) different fractions sampled from a Sobol sequence. The directory structure is the same as in the Latin hypercube, but for each cosmology, there is only one fraction realization (0), whose fraction value is stored in the file fractions.txt contained in each cosmology directory.

Example to read the contaminated catalogs:

import readfof

snapdir = '/home/cagliari/Quijote/FoF_contaminated/inbox/fiducial/0' # folder hosting the catalogue
snapnum = 2  # redshift 1

# Determine the redshift of the catalogue
z_dict = {4:0.0, 3:0.5, 2:1.0, 1:2.0, 0:3.0}
redshift = z_dict[snapnum]

# Read the halo catalogue
FoF_c_r = readfof.FoF_catalog(snapdir, snapnum, long_ids=False,
                              swap=False, SFR=False, read_IDs=False, read_type=True)

pos_h = FoF_c_r.GroupPos/1e3  # Halo positions in Mpc/h
mass  = FoF_c_r.GroupMass*1e10  # Halo masses in Msun/h
vel_h = FoF_c_r.GroupVel*(1.0+redshift)  # Halo peculiar velocities in km/s
Npart = FoF_c_r.GroupLen  # Number of CDM particles in the halo
Type  = FoF_c_r.GroupType  # 0 if target, 1 if interloper

# Check interloper fraction
N_t = len(pos_h[Type==0,:])
N_i = len(pos_h[Type==1,:])

print('Targets    :', N_t)
print('Interlopers:', N_i)
print('Total      :', N_t + N_i)
print('Interloper fraction:', N_i / (N_t + N_i))

Contaminated Statistics

The statistics that we provide are:

  1. The non-contaminated power spectrum saved in Pk_pylians_no-dz.dat.

  2. The contaminated power spectrum saved in Pk_pylians_dz.dat.

  3. The contaminated bispectra saved in Bk_6k_pyspectrum_dz.dat.

The statistics are stored in the folder Pk_contaminated and its subdirectories inbox and outbox.

Acknowledgements

This work has been done thanks to the facilities offered by the Univ. Savoie Mont Blanc - CNRS/IN2P3 MUST computing center.

Team

  • Marina Silvia Cagliari (LAPTh, France)

  • Azadeh Moradinezhad (LAPTh, France)

  • Francisco Villaescusa-Navarro (Simons/Princeton, USA)