csiborgtools/scripts/concentration_fit.ipynb
Richard Stiskalek 8a56c22813
Within halo work and NFW fit (#4)
* add listing of snapshots

* change distance to comoving

* ignore cp files

* rename nb

* add str to list

* add NFW profile shapes

* add fits imports

* Rename to Nsnap

* in clumps_read only select props

* make clumpid int

* expand doc

* add import

* edit readme

* distribute halos

* add profile & posterior

* add import

* add import

* add documentation

* add rvs and init guess

* update todo

* update nb

* add file

* return end index too

* change clump_ids format to int32

* skeleton of dump particle

* update nb

* add func to drop 0 clump indxs parts

* add import

* add halo dump

* switch to float32

* Update TODO

* update TODO

* add func that loads a split

* add halo object

* Rename to clump

* make post work with a clump

* add optimiser

* add Nsplits

* ignore submission scripts

* ignore .out

* add dumppath

* add job splitting

* add split halos script

* rename file

* renaem files

* rm file

* rename imports

* edit desc

* add pick clump

* add number of particles

* update TODO

* update todo

* add script

* add dumping

* change dumpdir structure

* change dumpdir

* add import

* Remove tqdm

* Increase the number of splits

* rm shuffle option

* Change to remove split

* add emojis

* fix part counts in splits

* change num of splits

* rm with particle cut

* keep splits

* fit only if 10 part and more

* add min distance

* rm warning about not set vels

* update TODO

* calculate rho0 too

* add results collection

* add import

* add func to combine splits

* update TODO

* add extract cols

* update nb

* update TODO
2022-10-30 20:16:56 +00:00

420 KiB

Using a calibrated flow model to predict $z_{\rm cosmo}$

In [5]:
# Copyright (C) 2024 Richard Stiskalek
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 3 of the License, or (at your
# option) any later version.
#
# This program is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
# Public License for more details.
#
# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
import numpy as np
import matplotlib.pyplot as plt
from h5py import File
from tqdm import tqdm

import csiborgtools

%load_ext autoreload
%autoreload 2
%matplotlib inline

paths = csiborgtools.read.Paths(**csiborgtools.paths_glamdring)
The autoreload extension is already loaded. To reload it, use:
  %reload_ext autoreload
In [2]:
def load_calibration(catalogue, simname, nsim, ksmooth):
    fname = f"/mnt/extraspace/rstiskalek/csiborg_postprocessing/peculiar_velocity/flow_samples_{catalogue}_{simname}_smooth_{ksmooth}.hdf5"  # noqa
    keys = ["Vext_x", "Vext_y", "Vext_z", "alpha", "beta", "sigma_v"]

    # SN_keys = ['mag_cal', 'alpha_cal', 'beta_cal']
    # SN_keys = []
    calibration_samples = {}
    with File(fname, 'r') as f:
        for key in keys:
            calibration_samples[key] = f[f"sim_{nsim}/{key}"][:][::10]

        # for key in SN_keys:
        #     calibration_samples[key] = f[f"sim_{nsim}/{key}"][:]

    return calibration_samples

Test running a model

In [135]:
# fpath_data = "/mnt/extraspace/rstiskalek/catalogs/PV_compilation_Supranta2019.hdf5"
fpath_data = "/mnt/extraspace/rstiskalek/catalogs/PV_mock_CB2_17417_large.hdf5"

simname = "csiborg2_main"
catalogue = "CB2_large"

nsims = paths.get_ics(simname)[:-1]
ksmooth = 1

loaders = []
models = []
zcosmo_mean = None
zobs = None

for i, nsim in enumerate(tqdm(nsims)):
    loader = csiborgtools.flow.DataLoader(simname, i, catalogue, fpath_data, paths, ksmooth=ksmooth)
    calibration_samples = load_calibration(catalogue, simname, nsim, ksmooth)
    model = csiborgtools.flow.Observed2CosmologicalRedshift(calibration_samples, loader.rdist, loader._Omega_m)

    if i == 0:
        zcosmo_mean = loader.cat["zcosmo"]
        zobs = loader.cat["zobs"]
        vrad = loader.cat["vrad"]

    loaders.append(loader)
    models.append(model)
  0%|          | 0/19 [00:00<?, ?it/s]
10:32:19: reading the catalogue.
10:32:19: reading the interpolated field.
/mnt/users/rstiskalek/csiborgtools/csiborgtools/flow/flow_model.py:113: UserWarning: The number of radial steps is even. Skipping the first step at 0.0 because Simpson's rule requires an odd number of steps.
  warn(f"The number of radial steps is even. Skipping the first "
  5%|▌         | 1/19 [00:00<00:05,  3.38it/s]
10:32:19: calculating the radial velocity.
10:32:20: reading the catalogue.
10:32:20: reading the interpolated field.
 11%|█         | 2/19 [00:00<00:04,  3.43it/s]
10:32:20: calculating the radial velocity.
10:32:20: reading the catalogue.
10:32:20: reading the interpolated field.
 16%|█▌        | 3/19 [00:00<00:04,  3.47it/s]
10:32:20: calculating the radial velocity.
10:32:20: reading the catalogue.
10:32:20: reading the interpolated field.
 21%|██        | 4/19 [00:01<00:04,  3.44it/s]
10:32:20: calculating the radial velocity.
10:32:20: reading the catalogue.
10:32:20: reading the interpolated field.
 26%|██▋       | 5/19 [00:01<00:04,  3.41it/s]
10:32:21: calculating the radial velocity.
10:32:21: reading the catalogue.
10:32:21: reading the interpolated field.
 32%|███▏      | 6/19 [00:01<00:03,  3.44it/s]
10:32:21: calculating the radial velocity.
10:32:21: reading the catalogue.
10:32:21: reading the interpolated field.
 37%|███▋      | 7/19 [00:02<00:03,  3.33it/s]
10:32:21: calculating the radial velocity.
10:32:21: reading the catalogue.
10:32:21: reading the interpolated field.
 42%|████▏     | 8/19 [00:02<00:03,  3.31it/s]
10:32:22: calculating the radial velocity.
10:32:22: reading the catalogue.
10:32:22: reading the interpolated field.
 47%|████▋     | 9/19 [00:02<00:03,  3.14it/s]
10:32:22: calculating the radial velocity.
10:32:22: reading the catalogue.
10:32:22: reading the interpolated field.
 53%|█████▎    | 10/19 [00:03<00:02,  3.18it/s]
10:32:22: calculating the radial velocity.
10:32:22: reading the catalogue.
10:32:22: reading the interpolated field.
 58%|█████▊    | 11/19 [00:03<00:02,  3.19it/s]
10:32:22: calculating the radial velocity.
10:32:23: reading the catalogue.
10:32:23: reading the interpolated field.
 63%|██████▎   | 12/19 [00:03<00:02,  3.25it/s]
10:32:23: calculating the radial velocity.
10:32:23: reading the catalogue.
10:32:23: reading the interpolated field.
 68%|██████▊   | 13/19 [00:03<00:01,  3.23it/s]
10:32:23: calculating the radial velocity.
10:32:23: reading the catalogue.
10:32:23: reading the interpolated field.
 74%|███████▎  | 14/19 [00:04<00:01,  3.25it/s]
10:32:23: calculating the radial velocity.
10:32:23: reading the catalogue.
10:32:23: reading the interpolated field.
 79%|███████▉  | 15/19 [00:04<00:01,  3.21it/s]
10:32:24: calculating the radial velocity.
10:32:24: reading the catalogue.
10:32:24: reading the interpolated field.
 84%|████████▍ | 16/19 [00:04<00:00,  3.26it/s]
10:32:24: calculating the radial velocity.
10:32:24: reading the catalogue.
10:32:24: reading the interpolated field.
 89%|████████▉ | 17/19 [00:05<00:00,  3.31it/s]
10:32:24: calculating the radial velocity.
10:32:24: reading the catalogue.
10:32:24: reading the interpolated field.
 95%|█████████▍| 18/19 [00:05<00:00,  3.33it/s]
10:32:25: calculating the radial velocity.
10:32:25: reading the catalogue.
10:32:25: reading the interpolated field.
100%|██████████| 19/19 [00:05<00:00,  3.30it/s]
10:32:25: calculating the radial velocity.

In [143]:
n = 400
zcosmo, pzcosmo = csiborgtools.flow.stack_pzosmo_over_realizations(
    n, models, loaders, "zobs")
Stacking:   0%|          | 0/19 [00:00<?, ?it/s]Stacking: 100%|██████████| 19/19 [00:06<00:00,  3.06it/s]
In [144]:
plt.figure()

# for i in range(len(nsims)):
    # mask = pzcosmo[i] > 1e-5
    # plt.plot(zcosmo[mask], pzcosmo[i][mask], color="black", alpha=0.1)

# mu = np.nanmean(pzcosmo, axis=0)
mask = pzcosmo > 1e-5
plt.plot(zcosmo[mask], pzcosmo[mask], color="black", label=r"$p(z_{\rm cosmo})$")

plt.ylim(0)
plt.axvline(zcosmo_mean[n], color="green", label=r"$z_{\rm cosmo}$")
plt.axvline(zobs[n], color="red", label=r"$z_{\rm CMB}$")

plt.xlabel(r"$z$")
plt.ylabel(r"$p(z)$")
plt.legend()
plt.tight_layout()
# plt.savefig("../plots/zcosmo_posterior_mock_example_B.png", dpi=450)
plt.show()
No description has been provided for this image
In [ ]: