Columbia LensingWelcome!
We are the weak lensing group originated from Columbia University Department of Astronomy.
Here you can download weak lensing maps (and more!) we created using the NSF XSEDE (Stampede2) and NASA HECC (Pleiades) supercomputing facilities.
© Columbia Lensing
Our group uses numerical simulations to study the weak gravitational lensing signature of large-scale structure and to understand fundamental physics such as the nature of dark energy and the
total mass of neutrinos. In particular, we try to capture the rich information that is beyond the traditional two-point statistics, using non-Gaussian statistics and convolutional neural networks.
PUBLICATIONS: ADS
Active contributors: |
 |
 |
 |
 |
 |
| Zoltán Haiman (Columbia) |
Jia Liu (Kavli IPMU) |
Ken Osato (Chiba U) |
J. Colin Hill (Columbia/Flatiron) |
Max Lee (Columbia) |
Alina Sabyr (Columbia) |
DESCRIPTION (simulation paper: arxiv:2010.09731)
The kappaTNG is a suite of mock weak lensing maps based on the cosmological hydrodynamic simulations IllustrisTNG (TNG300-1).
We also include the kappaTNG-Dark suit of maps, generated based on the corresponding dark matter-only TNG simulations.
These mock maps are suitable for studying the effects of baryons on weak lensing.
Simulation configuration:
- 10,000 realizations
- map size: 5x5 deg^2, 1024x1024 pixels
- resolution: 0.29 arcmin/pixel
- 40 source redshifts up to zs = 2.6
Sample Jupyter Notebook for loading data: https://github.com/0satoken/kappaTNG
DESCRIPTION
The dataset includes ray-traced convergence maps generated from N-body simulations with the baryonic correction model. See Lu et al. (2021) for the description of how the maps are generated. See Petri (2016) for the information of the underlying simulations.
The convergence maps have a field-of-view of 3.5×3.5 deg2 and a resolution of 2048×2048, ray-traced from z=1. The maps do not contain shape noise and smoothing.
There are two groups files in the dataset:
(1) `[cosmo]_fiducial.f32` includes 128 realization of the maps with fiducial baryonic model;
(2) `[cosmo]_baryon.f32` includes one realization (same random seed) for each of the 160 baryonic model (see `baryonic_parameters.txt` for the table of baryonic parameters).
`[cosmo]` denotes the cosmological parameter of the simulation (Om=ΩM, si=σ8).
Each file stores the a row-major array of little-endian 32-bit floating point numbers with dimensions n×2048×2048, where n is the number of maps in the file.
Full data can be downloaded from here, please email zoltan@astro.columbia.edu for a guest username/password.
DESCRIPTION
For the full detail for our simulation pipeline see Petri 2016.
A briefer description can be found in the "simulation" section in either Zorrilla et al 2016, Liu et al 2016 , or Petri et al 2016 .
Each tar.gz file (15GB) has 1000 fits maps (17MB each).
Simulation configuration (also in the header of the fits files):
size of the box = 240 Mpc/h
source redshift z_s = 2.0
map size = 3.5 x 3.5 = 12.25 deg^2
number of particles = 512^3
resolution = 2048 x 2048 pixels
No shape noise
Fixed cosmological parameters:
h = 0.72
n_s = 0.96
Omega_b = 0.046
Tcmb = 2.725K
N_eff = 3.04
neutrino masses = 0, 0, 0
Varying parameters are in the file name (also in the fits header), for example:
"Om0.260_Ode0.740_w-0.800_wa0.000_si0.800.tar.gz" has:
Omega_m = 0.26
Omega_lambda = 0.74
w_0 = -0.8, w_a = 0, where the dark energy EoS is w(a)=w_0+(1-a)w_a.
sigma_8 = 0.80
kappa maps:
Om0.230_Ode0.770_w-1.000_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-0.800_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa-0.200_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa-0.500_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-1.200_wa0.000_si0.800.tar.gz
Om0.290_Ode0.710_w-1.000_wa0.000_si0.800.tar.gz
B-mode maps (for null tests):
Om0.230_Ode0.770_w-1.000_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-0.800_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa-0.200_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa-0.500_si0.800.tar.gz
Om0.260_Ode0.740_w-1.000_wa0.000_si0.800.tar.gz
Om0.260_Ode0.740_w-1.200_wa0.000_si0.800.tar.gz
Om0.290_Ode0.710_w-1.000_wa0.000_si0.800.tar.gz
DESCRIPTION
This is a set of simulated convergence maps
for 96 different cosmologies. Each cosmology differs only on two cosmological parameters,
the density of matter, Omega_m, and the amplitude of density fluctuations measured in the
late universe, sigma_8. Each is saved in a compressed directory. Within
the directory, there are 512 convergence maps, saved as fits files.
This dataset was used for the analyses presented in Matilla et al 2016 and Gupta et al 2018, where you
can find detailed descriptions of the data. These two papers should be cited in a publication that makes use of the maps.
For convenience, here is a brief description:
- Each convergence map covers a field of view of 3.5deg x 3.5deg, and has a
resolution of 1024 x 1024 pixels.
- Maps share the initial random seeds between cosmologies. That is, the map
0001 for cosmology a and the same map for cosmology b were generated
using the same random seed, and will exhibit similar structures in the
same regions.
- Maps represent noiseless convergence from sources at a constant redshift
z=1.0.
- Each map was generated ray-tracing the outputs of dark matter-only, N-body
simulations, using the multi plane algorithm implemented in Lenstools. The ray-tracing does not use
the Born approximation, but assumes a flat sky.
- Each past light-cone was built from snapshots from a single N-body simulation
for each cosmology. The simulations evolved a 240Mpc/h side cube with 512^3
dark matter particles, using GADGET2. The distance between planes corresponds
to 80Mpc/h on the fiducial cosmology (Omega_m=0.260, sigma_8=0.800).
- Initial conditions for the Nbody simulations were built using NGenIC, from
scaled power spectra computed with CAMB.
Sample data: Download
Full data (set of maps in all 96 cosmologies) can be downloaded from here, please email zoltan@astro.columbia.edu for a guest username/password.
Note that a few maps were lost during a file transfer to a permanent repository and these two cosmological models have fewer maps:
Om0.246_si0.926: 508 maps
Om0.251_si0.807: 455 maps
DESCRIPTION (paper: Sabyr,
Hill, & Haiman (2024)) When our maps are used in your paper:
The dataset includes simplified tSZ maps (1-halo term only), generated using hmpdf (the version used for these maps can be found here). The maps are generated by Poisson sampling from Tinker+2010 halo mass function and painting Compton-y profiles using the pressure profile
fitting function from Battaglia+2012. The redshift and mass ranges are z=0.005-6 and M=10^11-10^16 M_sun. Radial
cutoff for the profiles was set at r_out=2r_vir. We provide maps covering fsky=0.1 at 0.1 arcmin
resolution: 960 realizations at the fiducial
cosmology and 142-144 at 4 cosmologies for Fisher forecasts with Omega_c, sigma_8 perturbed by
1%. Below we provide the exact cosmological parameters that we set for
the fiducial cosmology as an
example. The rest of the cosmological parameters are set to the default values in the CLASS 3.2.0 version.
#consistent with Planck 2018 except no massive neutrinos
h = 0.674
Omega_b = 0.0493
Omega_cdm = 0.264
sigma8 = 0.811
N_ur = 3.046
n_s = 0.965
tau_reio=0.054
N_ncdm=0
m_ncdm=0
output = mPk
P_k_max_h/Mpc = 10.0
We also include fiducial noiseless and noisy summary statistic data vectors across 154 cosmologies (816,
696 × 5 × 2 = 8,166,960 files). These include power spectra, Minkowski
functionals, peaks, minima, and moments. The noisy summary statistics were computed across maps that
include a noise realization from the post-component separated tSZ noise power spectra forecasted for the
Simons Observatory goal noise levels. The settings for these statistics can be found in Table I of the
paper. The files have extension *npy and can be read as dictionaries. The scripts
for generating the maps and the summary statistic pipeline can be found here.
More details about the tSZ maps and summary statistics can be found in the paper and code repositories.
If you have any questions, please feel free to contact Alina at as6131@columbia.edu.
To download the data, please contact zoltan@astro.columbia.edu for a guest username/password.
Acknowledgement
(1) Please add the following to the Acknowledgement section of your publication if you used any data product from this website:
"We thank the Columbia Lensing group (https://columbialensing.github.io/) for making their simulations available. The creation of these simulations is supported through grants NSF AST-1210877, NSF AST-140041, and NASA ATP-80NSSC18K1093."
(2) For the kappaTNG dataset, please cite Osato, Liu, & Haiman 2020 and the 5
TNG300 presentation papers.
(3) For the MassiveNuS dataset, please cite Liu et al. 2018, and add to Acknowledgement (in addition to #1 above): "We thank New Mexico State University (USA) and Instituto de Astrofisica de Andalucia CSIC (Spain) for hosting the Skies & Universes site for cosmological simulation products."
(4) For the Dark Matter dataset, please cite Matilla et al 2016 and Gupta et al 2018.
(5) For the Baryon Painting dataset, please cite Lu et al. (2021).
(6) For tSZ maps, please cite Sabyr,
Hill, & Haiman (2024).
Background image credit: CFHTLenS