Aquila-Consortium/csiborgtools
0
0
Fork 0
mirror of https://github.com/Richard-Sti/csiborgtools_public.git synced 2025-05-13 06:01:13 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add density field plot and start preparing CSiBORG2 (#94)

Browse source 
* Add RAMSES2HDF5 conversion

* Upload changes

* Clean up

* More clean up

* updates

* Little change

* pep9

* Add basic SPH calculation for a snapshot

* Add submit script

* Remove echo

* Little changes

* Send off changes

* Little formatting

* Little updates

* Add nthreads argument

* Upload chagnes

* Add nthreads arguemnts

* Some local changes..

* Update scripts

* Add submission script

* Update script

* Update params

* Rename CSiBORGBox to CSiBORG1box

* Rename CSiBORG1 reader

* Move files

* Rename folder again

* Add basic plotting here

* Add new skeletons

* Move def

* Update nbs

* Edit directories

* Rename files

* Add units to converted snapshots

* Fix empty dataset bug

* Delete file

* Edits to submission scripts

* Edit paths

* Update .gitignore

* Fix attrs

* Update weighting

* Fix RA/dec bug

* Add FORNAX cluster

* Little edit

* Remove boxes since will no longer need

* Move func back

* Edit to include sort by membership

* Edit paths

* Purge basic things

* Start removing

* Bring file back

* Scratch

* Update the rest

* Improve the entire file

* Remove old things

* Remove old

* Delete old things

* Fully updates

* Rename file

* Edit submit script

* Little things

* Add print statement

* Add here cols_to_structured

* Edit halo cat

* Remove old import

* Add comment

* Update paths manager

* Move file

* Remove file

* Add chains
This commit is contained in:
Richard Stiskalek 2023-12-13 16:08:25 +00:00 • committed by GitHub
parent 6042a87111
commit aaa14fc880
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
30 changed files with 1682 additions and 1728 deletions
Download patch file Download diff file Expand all files Collapse all files

17
csiborgtools/__init__.py
View file

@ -20,11 +20,14 @@ from .utils import (center_of_mass, delta2ncells, number_counts,
hms_to_degrees, dms_to_degrees, great_circle_distance) # noqa
# Arguments to csiborgtools.read.Paths.
paths_glamdring = {"srcdir": "/mnt/extraspace/hdesmond/",
"postdir": "/mnt/extraspace/rstiskalek/CSiBORG/",
"borg_dir": "/users/hdesmond/BORG_final/",
"quijote_dir": "/mnt/extraspace/rstiskalek/Quijote",
}
paths_glamdring = {
"csiborg1_srcdir": "/mnt/extraspace/rstiskalek/csiborg1",
"csiborg2_main_srcdir": "/mnt/extraspace/rstiskalek/csiborg2_main",
"csiborg2_varysmall_srcdir": "/mnt/extraspace/rstiskalek/csiborg2_varysmall", # noqa
"csiborg2_random_srcdir": "/mnt/extraspace/rstiskalek/csiborg2_random", # noqa
"postdir": "/mnt/extraspace/rstiskalek/csiborg_postprocessing/",
"quijote_dir": "/mnt/extraspace/rstiskalek/quijote",
}
neighbour_kwargs = {"rmax_radial": 155 / 0.705,
@ -76,4 +79,8 @@ clusters = {"Virgo": read.ObservedCluster(RA=hms_to_degrees(12, 27),
dec=dms_to_degrees(12, 43),
dist=16.5 * 0.7,
name="Virgo"),
"Fornax": read.ObservedCluster(RA=hms_to_degrees(3, 38),
dec=dms_to_degrees(-35, 27),
dist=19 * 0.7,
name="Fornax"),
}

10
csiborgtools/field/density.py
View file

@ -69,7 +69,7 @@ class DensityField(BaseField):
Parameters
----------
box : :py:class:`csiborgtools.read.CSiBORGBox`
box : :py:class:`csiborgtools.read.CSiBORG1Box`
The simulation box information and transformations.
MAS : str
Mass assignment scheme. Options are Options are: 'NGP' (nearest grid
@ -167,7 +167,7 @@ class DensityField(BaseField):
#
# Parameters
# ----------
# box : :py:class:`csiborgtools.read.CSiBORGBox`
# box : :py:class:`csiborgtools.read.CSiBORG1Box`
# The simulation box information and transformations.
# MAS : str
# Mass assignment scheme. Options are Options are: 'NGP' (nearest grid
@ -269,7 +269,7 @@ class VelocityField(BaseField):
Parameters
----------
box : :py:class:`csiborgtools.read.CSiBORGBox`
box : :py:class:`csiborgtools.read.CSiBORG1Box`
The simulation box information and transformations.
MAS : str
Mass assignment scheme. Options are Options are: 'NGP' (nearest grid
@ -405,7 +405,7 @@ class PotentialField(BaseField):
Parameters
----------
box : :py:class:`csiborgtools.read.CSiBORGBox`
box : :py:class:`csiborgtools.read.CSiBORG1Box`
The simulation box information and transformations.
MAS : str
Mass assignment scheme. Options are Options are: 'NGP' (nearest grid
@ -444,7 +444,7 @@ class TidalTensorField(BaseField):
Parameters
----------
box : :py:class:`csiborgtools.read.CSiBORGBox`
box : :py:class:`csiborgtools.read.CSiBORG1Box`
The simulation box information and transformations.
MAS : str
Mass assignment scheme used to calculate the density field. Options

16
csiborgtools/field/interp.py
View file

@ -139,7 +139,7 @@ def observer_vobs(velocity_field):
return vobs
def make_sky(field, angpos, dist, box, volume_weight=True, verbose=True):
def make_sky(field, angpos, dist, boxsize, volume_weight=True, verbose=True):
r"""
Make a sky map of a scalar field. The observer is in the centre of the
box the field is evaluated along directions `angpos` (RA [0, 360) deg,
@ -153,9 +153,9 @@ def make_sky(field, angpos, dist, box, volume_weight=True, verbose=True):
angpos : 2-dimensional arrays of shape `(ndir, 2)`
Directions to evaluate the field.
dist : 1-dimensional array
Uniformly spaced radial distances to evaluate the field.
box : :py:class:`csiborgtools.read.CSiBORGBox`
The simulation box information and transformations.
Uniformly spaced radial distances to evaluate the field in `Mpc / h`.
boxsize : float
Box size in `Mpc / h`.
volume_weight : bool, optional
Whether to weight the field by the volume of the pixel.
verbose : bool, optional
@ -168,11 +168,11 @@ def make_sky(field, angpos, dist, box, volume_weight=True, verbose=True):
dx = dist[1] - dist[0]
assert numpy.allclose(dist[1:] - dist[:-1], dx)
assert angpos.ndim == 2 and dist.ndim == 1
# We loop over the angular directions, at each step evaluating a vector
# of distances. We pre-allocate arrays for speed.
dir_loop = numpy.full((dist.size, 3), numpy.nan, dtype=numpy.float32)
boxdist = box.mpc2box(dist)
boxsize = box.box2mpc(1.)
ndir = angpos.shape[0]
out = numpy.full(ndir, numpy.nan, dtype=numpy.float32)
for i in trange(ndir) if verbose else range(ndir):
@ -181,7 +181,7 @@ def make_sky(field, angpos, dist, box, volume_weight=True, verbose=True):
dir_loop[:, 2] = angpos[i, 1]
if volume_weight:
out[i] = numpy.sum(
boxdist**2
dist**2
* evaluate_sky(field, pos=dir_loop, mpc2box=1 / boxsize))
else:
out[i] = numpy.sum(
@ -244,7 +244,7 @@ def field2rsp(field, radvel_field, box, MAS, init_value=0.):
radvel_field : 3-dimensional array of shape `(grid, grid, grid)`
Radial velocity field in `km / s`. Expected to account for the observer
velocity.
box : :py:class:`csiborgtools.read.CSiBORGBox`
box : :py:class:`csiborgtools.read.CSiBORG1Box`
The simulation box information and transformations.
MAS : str
Mass assignment. Must be one of `NGP`, `CIC`, `TSC` or `PCS`.

7
csiborgtools/field/utils.py
View file

@ -49,5 +49,8 @@ def nside2radec(nside):
"""
pixs = numpy.arange(healpy.nside2npix(nside))
theta, phi = healpy.pix2ang(nside, pixs)
theta -= numpy.pi / 2
return 180 / numpy.pi * numpy.vstack([phi, theta]).T
ra = 180 / numpy.pi * phi
dec = 90 - 180 / numpy.pi * theta
return numpy.vstack([ra, dec]).T

6
csiborgtools/read/__init__.py
View file

@ -12,12 +12,8 @@
# 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.
from .box_units import CSiBORGBox, QuijoteBox # noqa
from .halo_cat import (CSiBORGCatalogue, QuijoteCatalogue, # noqa
CSiBORGPHEWCatalogue, fiducial_observers) # noqa
fiducial_observers) # noqa
from .obs import (SDSS, MCXCClusters, PlanckClusters, TwoMPPGalaxies, # noqa
TwoMPPGroups, ObservedCluster, match_array_to_no_masking) # noqa
from .paths import Paths # noqa
from .readsim import (CSiBORGReader, QuijoteReader, load_halo_particles, # noqa
make_halomap_dict) # noqa
from .utils import cols_to_structured, read_h5 # noqa

276
csiborgtools/read/box_units.py
View file

@ -1,276 +0,0 @@
# Copyright (C) 2022 Richard Stiskalek, Deaglan Bartlett
# 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.
"""
Simulation box unit transformations.
"""
from abc import ABC, abstractmethod, abstractproperty
import numpy
from astropy import constants, units
from astropy.cosmology import LambdaCDM
from .readsim import CSiBORGReader, QuijoteReader
###############################################################################
# Base box #
###############################################################################
class BaseBox(ABC):
_name = "box_units"
_cosmo = None
@property
def cosmo(self):
if self._cosmo is None:
raise ValueError("Cosmology not set.")
return self._cosmo
@property
def H0(self):
r"""Present Hubble parameter in :math:`\mathrm{km} \mathrm{s}^{-1}`"""
return self.cosmo.H0.value
@property
def rho_crit0(self):
"""Present-day critical density in M_sun h^2 / cMpc^3."""
rho_crit0 = self.cosmo.critical_density0
return rho_crit0.to_value(units.solMass / units.Mpc**3)
@property
def h(self):
"""The little 'h' parameter at the time of the snapshot."""
return self._h
@property
def Om0(self):
"""The present time matter density parameter."""
return self.cosmo.Om0
@abstractproperty
def boxsize(self):
"""Box size in cMpc."""
pass
@abstractmethod
def mpc2box(self, length):
r"""
Convert length from :math:`\mathrm{cMpc} / h` to box units.
Parameters
----------
length : float
Length in :math:`\mathrm{cMpc}`
Returns
-------
float
"""
pass
@abstractmethod
def box2mpc(self, length):
r"""
Convert length from box units to :math:`\mathrm{cMpc} / h`.
Parameters
----------
length : float
Length in box units.
Returns
-------
float
"""
pass
@abstractmethod
def solarmass2box(self, mass):
r"""
Convert mass from :math:`M_\odot / h` to box units.
Parameters
----------
mass : float
Mass in :math:`M_\odot / h`.
Returns
-------
float
"""
pass
@abstractmethod
def box2solarmass(self, mass):
r"""
Convert mass from box units to :math:`M_\odot / h`.
Parameters
----------
mass : float
Mass in box units.
Returns
-------
float
"""
pass
@abstractmethod
def m200c_to_r200c(self, m200c):
"""
Convert M200c to R200c in units of cMpc / h.
Parameters
----------
m200c : float
M200c in units of M_sun / h.
Returns
-------
float
"""
pass
###############################################################################
# CSiBORG box #
###############################################################################
class CSiBORGBox(BaseBox):
r"""
CSiBORG box units class for converting between box and physical units.
Paramaters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
paths : py:class`csiborgtools.read.Paths`
CSiBORG paths object.
"""
def __init__(self, nsnap, nsim, paths):
"""
Read in the snapshot info file and set the units from it.
"""
partreader = CSiBORGReader(paths)
info = partreader.read_info(nsnap, nsim)
pars = ["boxlen", "time", "aexp", "H0", "omega_m", "omega_l",
"omega_k", "omega_b", "unit_l", "unit_d", "unit_t"]
for par in pars:
setattr(self, "_" + par, info[par])
self._h = self._H0 / 100
self._cosmo = LambdaCDM(H0=100, Om0=self._omega_m,
Ode0=self._omega_l, Tcmb0=2.725 * units.K,
Ob0=self._omega_b)
self._Msuncgs = constants.M_sun.cgs.value # Solar mass in grams
def mpc2box(self, length):
conv = (self._unit_l / units.kpc.to(units.cm) / self._aexp) * 1e-3
conv *= self._h
return length / conv
def box2mpc(self, length):
conv = (self._unit_l / units.kpc.to(units.cm) / self._aexp) * 1e-3
conv *= self._h
return length * conv
def solarmass2box(self, mass):
conv = (self._unit_d * self._unit_l**3) / self._Msuncgs
conv *= self.h
return mass / conv
def box2solarmass(self, mass):
conv = (self._unit_d * self._unit_l**3) / self._Msuncgs
conv *= self.h
return mass * conv
def box2vel(self, vel):
r"""
Convert velocity from box units to :math:`\mathrm{km} \mathrm{s}^{-1}`.
Parameters
----------
vel : float
Velocity in box units.
Returns
-------
vel : float
Velocity in :math:`\mathrm{km} \mathrm{s}^{-1}`.
"""
return vel * (1e-2 * self._unit_l / self._unit_t / self._aexp) * 1e-3
@property
def boxsize(self):
return self.box2mpc(1.)
def m200c_to_r200c(self, m200c):
rho_crit = self.cosmo.critical_density(1 / self._aexp - 1)
rho_crit = rho_crit.to_value(units.solMass / units.Mpc**3)
r200c = (3 * m200c / (4 * numpy.pi * 200 * rho_crit))**(1 / 3)
return r200c / self._aexp
###############################################################################
# Quijote fiducial cosmology box #
###############################################################################
class QuijoteBox(BaseBox):
"""
Quijote cosmology box.
Parameters
----------
nsnap : int
Snapshot number.
nsim : int
IC realisation index.
paths : py:class`csiborgtools.read.Paths`
Paths manager
"""
def __init__(self, nsnap, nsim, paths):
zdict = {4: 0.0, 3: 0.5, 2: 1.0, 1: 2.0, 0: 3.0}
assert nsnap in zdict.keys(), f"`nsnap` must be in {zdict.keys()}."
info = QuijoteReader(paths).read_info(nsnap, nsim)
self._aexp = 1 / (1 + zdict[nsnap])
self._h = info["h"]
self._cosmo = LambdaCDM(H0=100, Om0=info["Omega_m"],
Ode0=info["Omega_l"], Tcmb0=2.725 * units.K)
self._info = info
@property
def boxsize(self):
return self._info["BoxSize"]
def box2mpc(self, length):
return length * self.boxsize
def mpc2box(self, length):
return length / self.boxsize
def solarmass2box(self, mass):
return mass / self._info["TotMass"]
def box2solarmass(self, mass):
return mass * self._info["TotMass"]
def m200c_to_r200c(self, m200c):
raise ValueError("Not implemented for Quijote boxes.")

112
csiborgtools/read/halo_cat.py
View file

@ -28,9 +28,9 @@ from sklearn.neighbors import NearestNeighbors
from ..utils import (cartesian_to_radec, fprint, great_circle_distance,
number_counts, periodic_distance_two_points,
real2redshift)
from .box_units import CSiBORGBox, QuijoteBox
# TODO: removing these
from .box_units import CSiBORG1Box, QuijoteBox
from .paths import Paths
from .readsim import load_halo_particles, make_halomap_dict
###############################################################################
# Base catalogue #
@ -622,75 +622,7 @@ class CSiBORGCatalogue(BaseCatalogue):
"csiborg", nsim, max(paths.get_snapshots(nsim, "csiborg")),
halo_finder, catalogue_name, paths, mass_key, bounds,
[338.85, 338.85, 338.85], observer_velocity, cache_maxsize)
self.box = CSiBORGBox(self.nsnap, self.nsim, self.paths)
###############################################################################
# Quijote PHEW without snapshot catalogue #
###############################################################################
class CSiBORGPHEWCatalogue(BaseCatalogue):
r"""
CSiBORG PHEW halo catalogue without snapshot. Units typically used are:
- Length: :math:`cMpc / h`
- Mass: :math:`M_\odot / h`
Note that the PHEW catalogue is not very reliable.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
paths : py:class`csiborgtools.read.Paths`
Paths object.
mass_key : str, optional
Mass key of the catalogue.
bounds : dict, optional
Parameter bounds; keys as parameter names, values as (min, max) or
a boolean.
cache_maxsize : int, optional
Maximum number of cached arrays.
"""
def __init__(self, nsnap, nsim, paths, mass_key=None, bounds=None,
cache_maxsize=64):
super().__init__()
self.simname = "csiborg"
self.nsnap = nsnap
self.nsim = nsim
self.paths = paths
self.mass_key = mass_key
self.observer_location = [338.85, 338.85, 338.85]
fname = paths.processed_phew(nsim)
self._data = File(fname, "r")
if str(nsnap) not in self._data.keys():
raise ValueError(f"Snapshot {nsnap} not in the catalogue. "
f"Options are {self.get_snapshots(nsim, paths)}")
self.catalogue_name = str(nsnap)
self._is_closed = False
self.cache_maxsize = cache_maxsize
if bounds is not None:
self._make_mask(bounds)
self._derived_properties = ["cartesian_pos", "spherical_pos", "dist"]
self.box = CSiBORGBox(self.nsnap, self.nsim, self.paths)
self.clear_cache()
@staticmethod
def get_snapshots(nsim, paths):
"""List of snapshots available for this simulation."""
fname = paths.processed_phew(nsim)
with File(fname, "r") as f:
snaps = [int(key) for key in f.keys() if key != "info"]
f.close()
return numpy.sort(snaps)
self.box = CSiBORG1Box(self.nsnap, self.nsim, self.paths)
###############################################################################
@ -850,3 +782,41 @@ def find_boxed(pos, center, subbox_size, boxsize):
start_index = i + 1
return indxs
###############################################################################
# Supplementary functions #
###############################################################################
def make_halomap_dict(halomap):
"""
Make a dictionary mapping halo IDs to their start and end indices in the
snapshot particle array.
"""
return {hid: (int(start), int(end)) for hid, start, end in halomap}
def load_halo_particles(hid, particles, hid2map):
"""
Load a halo's particles from a particle array. If it is not there, i.e
halo has no associated particles, return `None`.
Parameters
----------
hid : int
Halo ID.
particles : 2-dimensional array
Array of particles.
hid2map : dict
Dictionary mapping halo IDs to `halo_map` array positions.
Returns
-------
parts : 1- or 2-dimensional array
"""
try:
k0, kf = hid2map[hid]
return particles[k0:kf + 1]
except KeyError:
return None

15
csiborgtools/read/obs.py
View file

@ -26,7 +26,6 @@ from astropy.io import fits
from astropy.cosmology import FlatLambdaCDM
from scipy import constants
from .utils import cols_to_structured
###############################################################################
# Text survey base class #
@ -830,3 +829,17 @@ def match_array_to_no_masking(arr, surv):
out[indx] = arr[i]
return out
def cols_to_structured(N, cols):
"""
Allocate a structured array from `cols`, a list of (name, dtype) tuples.
"""
if not (isinstance(cols, list)
and all(isinstance(c, tuple) and len(c) == 2 for c in cols)):
raise TypeError("`cols` must be a list of (name, dtype) tuples.")
names, formats = zip(*cols)
dtype = {"names": names, "formats": formats}
return numpy.full(N, numpy.nan, dtype=dtype)

491
csiborgtools/read/paths.py
View file

@ -15,9 +15,9 @@
"""
CSiBORG paths manager.
"""
from glob import glob, iglob
from glob import glob
from os import makedirs
from os.path import isdir, join
from os.path import basename, isdir, join
from warnings import warn
import numpy
@ -40,6 +40,7 @@ class Paths:
Parameters
----------
# HERE EDIT EVERYTHING
srcdir : str, optional
Path to the folder where the RAMSES outputs are stored.
postdir: str, optional
@ -49,73 +50,158 @@ class Paths:
quiote_dir : str, optional
Path to the folder where Quijote simulations are stored.
"""
_srcdir = None
_postdir = None
_borg_dir = None
_quijote_dir = None
def __init__(self, srcdir=None, postdir=None, borg_dir=None,
quijote_dir=None):
self.srcdir = srcdir
self.postdir = postdir
self.borg_dir = borg_dir
def __init__(self,
csiborg1_srcdir=None,
csiborg2_main_srcdir=None,
csiborg2_random_srcdir=None,
csiborg2_varysmall_srcdir=None,
postdir=None,
quijote_dir=None
):
self.csiborg1_srcdir = csiborg1_srcdir
self.csiborg2_main_srcdir = csiborg2_main_srcdir
self.csiborg2_random_srcdir = csiborg2_random_srcdir
self.csiborg2_varysmall_srcdir = csiborg2_varysmall_srcdir
self.quijote_dir = quijote_dir
@property
def srcdir(self):
"""Path to the folder where CSiBORG simulations are stored."""
if self._srcdir is None:
raise ValueError("`srcdir` is not set!")
return self._srcdir
self.postdir = postdir
@srcdir.setter
def srcdir(self, path):
if path is None:
return
check_directory(path)
self._srcdir = path
def get_ics(self, simname, from_quijote_backup=False):
"""
Get available IC realisation IDs for a given simulation.
@property
def borg_dir(self):
"""Path to the folder where BORG MCMC chains are stored."""
if self._borg_dir is None:
raise ValueError("`borg_dir` is not set!")
return self._borg_dir
Parameters
----------
simname : str
Simulation name.
from_quijote_backup : bool, optional
Whether to return the ICs from the Quijote backup.
@borg_dir.setter
def borg_dir(self, path):
if path is None:
return
check_directory(path)
self._borg_dir = path
Returns
-------
ids : 1-dimensional array
"""
if simname == "csiborg":
files = glob(join(self.csiborg1_srcdir, "chain_*"))
files = [int(basename(f).replace("chain_", "") for f in files)]
elif simname == "csiborg2_main":
files = glob(join(self.csiborg2_main_srcdir, "chain_*"))
files = [int(basename(f).replace("chain_", "") for f in files)]
elif simname == "csiborg2_random":
raise NotImplementedError("TODO")
elif simname == "csiborg2_varysmall":
raise NotImplementedError("TODO")
elif simname == "quijote" or simname == "quijote_full":
if from_quijote_backup:
files = glob(join(self.quijote_dir, "halos_backup", "*"))
else:
warn(("Taking only the snapshots that also have "
"a FoF catalogue!"))
files = glob(join(self.quijote_dir, "Snapshots_fiducial", "*"))
files = [int(f.split("/")[-1]) for f in files]
files = [f for f in files if f < 100]
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
@property
def quijote_dir(self):
"""Path to the folder where Quijote simulations are stored."""
if self._quijote_dir is None:
raise ValueError("`quijote_dir` is not set!")
return self._quijote_dir
return numpy.sort(files)
@quijote_dir.setter
def quijote_dir(self, path):
if path is None:
return
check_directory(path)
self._quijote_dir = path
def snapshots(self, nsim, simname):
if simname == "csiborg":
fname = "ramses_out_{}"
if tonew:
fname += "_new"
return join(self.postdir, "output", fname.format(nsim))
return join(self.csiborg1_srcdir, fname.format(nsim))
elif simname == "csiborg2_main":
return join(self.csiborg2_main_srcdir, f"chain_{nsim}", "output")
elif simname == "csiborg2_random":
raise NotImplementedError("TODO")
elif simname == "csiborg2_varysmall":
raise NotImplementedError("TODO")
elif simname == "quijote":
return join(self.quijote_dir, "Snapshots_fiducial", str(nsim))
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
@property
def postdir(self):
"""Path to the folder where post-processed files are stored."""
if self._postdir is None:
raise ValueError("`postdir` is not set!")
return self._postdir
def snapshot(self, nsnap, nsim, simname):
"""
Path to an IC realisation snapshot.
@postdir.setter
def postdir(self, path):
if path is None:
return
check_directory(path)
self._postdir = path
Parameters
----------
nsnap : int
Snapshot index. For Quijote, `-1` indicates the IC snapshot.
nsim : inlot
IC realisation index.
simname : str
Simulation name.
Returns
-------
str
"""
if simname == "csiborg":
return join(self.csiborg1_srcdir, f"chain_{nsim}",
f"snapshot_{str(nsnap).zfill(5)}")
elif simname == "csiborg2_main":
return join(self.csiborg1_srcdir, f"chain_{nsim}",
f"snapshot_{str(nsnap).zfill(5)}")
elif simname == "csiborg2_random":
raise NotImplementedError("TODO")
elif simname == "csiborg2_varysmall":
raise NotImplementedError("TODO")
elif simname == "quijote":
raise NotImplementedError("TODO")
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
# simpath = self.snapshots(nsim, simname, tonew=nsnap == 1)
# if simname == "csiborg":
# return join(simpath, f"output_{str(nsnap).zfill(5)}")
# else:
# if nsnap == -1:
# return join(simpath, "ICs", "ics")
# nsnap = str(nsnap).zfill(3)
# return join(simpath, f"snapdir_{nsnap}", f"snap_{nsnap}")
def get_snapshots(self, nsim, simname):
"""
List of available snapshots of simulation.
Parameters
----------
nsim : int
IC realisation index.
simname : str
Simulation name.
Returns
-------
snapshots : 1-dimensional array
"""
simpath = self.snapshots(nsim, simname, tonew=False)
if simname == "csiborg":
# Get all files in simpath that start with output_
snaps = glob(join(simpath, "output_*"))
# Take just the last _00XXXX from each file and strip zeros
snaps = [int(snap.split("_")[-1].lstrip("0")) for snap in snaps]
elif simname == "csiborg2_main":
snaps = glob(join(simpath, "snapshot_*"))
snaps = [basename(snap) for snap in snaps]
snaps = [int(snap.split("_")[1]) for snap in snaps]
elif simname == "csiborg2_random":
raise NotImplementedError("TODO")
elif simname == "csiborg2_varysmall":
raise NotImplementedError("TODO")
elif simname == "quijote":
snaps = glob(join(simpath, "snapdir_*"))
snaps = [int(snap.split("/")[-1].split("snapdir_")[-1])
for snap in snaps]
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
return numpy.sort(snaps)
@staticmethod
def quijote_fiducial_nsim(nsim, nobs=None):
@ -140,268 +226,6 @@ class Paths:
return nsim
return f"{str(nobs).zfill(2)}{str(nsim).zfill(3)}"
def borg_mcmc(self, nsim):
"""
Path to the BORG MCMC chain file.
Parameters
----------
nsim : int
IC realisation index.
Returns
-------
str
"""
return join(self.borg_dir, "mcmc", f"mcmc_{nsim}.h5")
def fof_cat(self, nsnap, nsim, simname, from_quijote_backup=False):
r"""
Path to the :math:`z = 0` FoF halo catalogue.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
simname : str
Simulation name. Must be one of `csiborg` or `quijote`.
from_quijote_backup : bool, optional
Whether to return the path to the Quijote FoF catalogue from the
backup.
Returns
-------
str
"""
if simname == "csiborg":
fdir = join(self.postdir, "halo_maker", f"ramses_{nsim}",
f"output_{str(nsnap).zfill(5)}", "FOF")
try_create_directory(fdir)
return join(fdir, "fort.132")
elif simname == "quijote":
if from_quijote_backup:
return join(self.quijote_dir, "halos_backup", str(nsim))
else:
return join(self.quijote_dir, "Halos_fiducial", str(nsim))
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
def get_ics(self, simname, from_quijote_backup=False):
"""
Get available IC realisation IDs for either the CSiBORG or Quijote
simulation suite.
Parameters
----------
simname : str
Simulation name. Must be `csiborg` or `quijote`.
from_quijote_backup : bool, optional
Whether to return the ICs from the Quijote backup.
Returns
-------
ids : 1-dimensional array
"""
if simname == "csiborg":
files = glob(join(self.srcdir, "ramses_out*"))
files = [f.split("/")[-1] for f in files] # Only file names
files = [f for f in files if "_inv" not in f] # Remove inv. ICs
files = [f for f in files if "_new" not in f] # Remove _new
files = [f for f in files if "OLD" not in f] # Remove _old
files = [int(f.split("_")[-1]) for f in files]
try:
files.remove(5511)
except ValueError:
pass
elif simname == "quijote" or simname == "quijote_full":
if from_quijote_backup:
files = glob(join(self.quijote_dir, "halos_backup", "*"))
else:
warn(("Taking only the snapshots that also have "
"a FoF catalogue!"))
files = glob(join(self.quijote_dir, "Snapshots_fiducial", "*"))
files = [int(f.split("/")[-1]) for f in files]
files = [f for f in files if f < 100]
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
return numpy.sort(files)
def snapshots(self, nsim, simname, tonew=False):
"""
Path to an IC snapshots folder.
Parameters
----------
nsim : int
IC realisation index.
simname : str
Simulation name. Must be one of `csiborg` or `quijote`.
tonew : bool, optional
Whether to return the path to the '_new' IC realisation of
CSiBORG. Ignored for Quijote.
Returns
-------
str
"""
if simname == "csiborg":
fname = "ramses_out_{}"
if tonew:
fname += "_new"
return join(self.postdir, "output", fname.format(nsim))
return join(self.srcdir, fname.format(nsim))
elif simname == "quijote":
return join(self.quijote_dir, "Snapshots_fiducial", str(nsim))
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
def get_snapshots(self, nsim, simname):
"""
List of available snapshots of simulation.
Parameters
----------
nsim : int
IC realisation index.
simname : str
Simulation name. Must be one of `csiborg` or `quijote`.
Returns
-------
snapshots : 1-dimensional array
"""
simpath = self.snapshots(nsim, simname, tonew=False)
if simname == "csiborg":
# Get all files in simpath that start with output_
snaps = glob(join(simpath, "output_*"))
# Take just the last _00XXXX from each file and strip zeros
snaps = [int(snap.split("_")[-1].lstrip("0")) for snap in snaps]
elif simname == "quijote":
snaps = glob(join(simpath, "snapdir_*"))
snaps = [int(snap.split("/")[-1].split("snapdir_")[-1])
for snap in snaps]
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
return numpy.sort(snaps)
def snapshot(self, nsnap, nsim, simname):
"""
Path to an IC realisation snapshot.
Parameters
----------
nsnap : int
Snapshot index. For Quijote, `-1` indicates the IC snapshot.
nsim : int
IC realisation index.
simname : str
Simulation name. Must be one of `csiborg` or `quijote`.
Returns
-------
str
"""
simpath = self.snapshots(nsim, simname, tonew=nsnap == 1)
if simname == "csiborg":
return join(simpath, f"output_{str(nsnap).zfill(5)}")
else:
if nsnap == -1:
return join(simpath, "ICs", "ics")
nsnap = str(nsnap).zfill(3)
return join(simpath, f"snapdir_{nsnap}", f"snap_{nsnap}")
def processed_output(self, nsim, simname, halo_finder):
"""
Path to the files containing all particles of a CSiBORG realisation at
:math:`z = 0`.
Parameters
----------
nsim : int
IC realisation index.
simname : str
Simulation name. Must be one of `csiborg` or `quijote`.
halo_finder : str
Halo finder name.
Returns
-------
str
"""
if simname == "csiborg":
fdir = join(self.postdir, "processed_output")
elif simname == "quijote":
fdir = join(self.quijote_dir, "Particles_fiducial")
else:
raise ValueError(f"Unknown simulation name `{simname}`.")
try_create_directory(fdir)
fname = f"parts_{halo_finder}_{str(nsim).zfill(5)}.hdf5"
return join(fdir, fname)
def processed_phew(self, nsim):
"""
Path to the files containing PHEW CSiBORG catalogues.
Parameters
----------
nsim : int
IC realisation index.
Returns
-------
str
"""
fdir = join(self.postdir, "processed_output")
try_create_directory(fdir)
return join(fdir, f"phew_{str(nsim).zfill(5)}.hdf5")
def halomaker_particle_membership(self, nsnap, nsim, halo_finder):
"""
Path to the HaloMaker particle membership file (CSiBORG only).
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
halo_finder : str
Halo finder name.
Returns
-------
str
"""
fdir = join(self.postdir, "halo_maker", f"ramses_{nsim}",
f"output_{str(nsnap).zfill(5)}", halo_finder)
fpath = join(fdir, "*particle_membership*")
return next(iglob(fpath, recursive=True), None)
def ascii_positions(self, nsim, kind):
"""
Path to ASCII files containing the positions of particles or halos.
Parameters
----------
nsim : int
IC realisation index.
kind : str
Kind of data to extract. Must be one of `particles`,
`particles_rsp`, `halos`, `halos_rsp`.
"""
assert kind in ["particles", "particles_rsp", "halos", "halos_rsp"]
fdir = join(self.postdir, "ascii_positions")
try_create_directory(fdir)
fname = f"pos_{kind}_{str(nsim).zfill(5)}.txt"
return join(fdir, fname)
def overlap(self, simname, nsim0, nsimx, min_logmass, smoothed):
"""
Path to the overlap files between two CSiBORG simulations.
@ -569,29 +393,6 @@ class Paths:
return join(fdir, fname)
def observer_peculiar_velocity(self, MAS, grid, nsim):
"""
Path to the files containing the observer peculiar velocity.
Parameters
----------
MAS : str
Mass-assignment scheme.
grid : int
Grid size.
nsim : int
IC realisation index.
Returns
-------
str
"""
fdir = join(self.postdir, "environment")
try_create_directory(fdir)
fname = f"obs_vp_{MAS}_{str(nsim).zfill(5)}_{grid}.npz"
return join(fdir, fname)
def cross_nearest(self, simname, run, kind, nsim=None, nobs=None):
"""
Path to the files containing distance from a halo in a reference

631
csiborgtools/read/readsim.py
View file

@ -1,631 +0,0 @@
# Copyright (C) 2022 Richard Stiskalek, Harry Desmond
# 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.
"""
Functions to read in the particle and clump files.
"""
from abc import ABC, abstractmethod
from gc import collect
from os.path import getsize, isfile, join
from warnings import warn
import numpy
import pynbody
from scipy.io import FortranFile
from tqdm import tqdm
try:
import readgadget
from readfof import FoF_catalog
except ImportError:
warn("Could not import `readgadget` and `readfof`. Related routines will not be available", ImportWarning) # noqa
from tqdm import trange
from ..utils import fprint
from .paths import Paths
from .utils import add_columns, cols_to_structured, flip_cols
class BaseReader(ABC):
"""
Base class for all readers.
"""
_paths = None
@property
def paths(self):
"""Paths manager."""
return self._paths
@paths.setter
def paths(self, paths):
assert isinstance(paths, Paths)
self._paths = paths
@abstractmethod
def read_info(self, nsnap, nsim):
"""
Read simulation snapshot info.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
Returns
-------
info : dict
Dictionary of information paramaters.
"""
pass
@abstractmethod
def read_snapshot(self, nsnap, nsim, kind, sort_like_final=False):
"""
Read snapshot.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
kind : str
Information to read. Can be `pid`, `pos`, `vel`, or `mass`.
sort_like_final : bool, optional
Whether to sort the particles like the final snapshot.
Returns
-------
n-dimensional array
"""
@abstractmethod
def read_halo_id(self, nsnap, nsim, halo_finder, verbose=True):
"""
Read the (sub) halo membership of particles.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
halo_finder : str
Halo finder used when running the catalogue.
Returns
-------
out : 1-dimensional array of shape `(nparticles, )`
"""
def read_catalogue(self, nsnap, nsim, halo_finder):
"""
Read in the halo catalogue.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
halo_finder : str
Halo finder used when running the catalogue.
Returns
-------
structured array
"""
###############################################################################
# CSiBORG particle reader #
###############################################################################
class CSiBORGReader(BaseReader):
"""
Object to read in CSiBORG snapshots from the binary files and halo
catalogues.
Parameters
----------
paths : py:class`csiborgtools.read.Paths`
"""
def __init__(self, paths):
self.paths = paths
def read_info(self, nsnap, nsim):
snappath = self.paths.snapshot(nsnap, nsim, "csiborg")
filename = join(snappath, "info_{}.txt".format(str(nsnap).zfill(5)))
with open(filename, "r") as f:
info = f.read().split()
# Throw anything below ordering line out
info = numpy.asarray(info[:info.index("ordering")])
# Get indexes of lines with `=`. Indxs before/after be keys/vals
eqs = numpy.asarray([i for i in range(info.size) if info[i] == '='])
keys = info[eqs - 1]
vals = info[eqs + 1]
return {key: convert_str_to_num(val) for key, val in zip(keys, vals)}
def read_snapshot(self, nsnap, nsim, kind):
sim = pynbody.load(self.paths.snapshot(nsnap, nsim, "csiborg"))
if kind == "pid":
x = numpy.array(sim["iord"], dtype=numpy.uint64)
elif kind in ["pos", "vel", "mass"]:
x = numpy.array(sim[kind], dtype=numpy.float32)
else:
raise ValueError(f"Unknown kind `{kind}`.")
# Because of a RAMSES bug x and z are flipped.
if kind in ["pos", "vel"]:
x[:, [0, 2]] = x[:, [2, 0]]
del sim
collect()
return x
def read_halo_id(self, nsnap, nsim, halo_finder, verbose=True):
if halo_finder == "PHEW":
ids = self.read_phew_id(nsnap, nsim, verbose)
elif halo_finder in ["FOF"]:
ids = self.read_halomaker_id(nsnap, nsim, halo_finder, verbose)
else:
raise ValueError(f"Unknown halo finder `{halo_finder}`.")
return ids
def open_particle(self, nsnap, nsim, verbose=True):
"""Open particle files to a given CSiBORG simulation."""
snappath = self.paths.snapshot(nsnap, nsim, "csiborg")
ncpu = int(self.read_info(nsnap, nsim)["ncpu"])
nsnap = str(nsnap).zfill(5)
fprint(f"reading in output `{nsnap}` with ncpu = `{ncpu}`.", verbose)
# First read the headers. Reallocate arrays and fill them.
nparts = numpy.zeros(ncpu, dtype=int)
partfiles = [None] * ncpu
for cpu in range(ncpu):
cpu_str = str(cpu + 1).zfill(5)
fpath = join(snappath, "part_{}.out{}".format(nsnap, cpu_str))
f = FortranFile(fpath)
# Read in this order
ncpuloc = f.read_ints()
if ncpuloc != ncpu:
infopath = join(snappath, f"info_{nsnap}.txt")
raise ValueError(
"`ncpu = {}` of `{}` disagrees with `ncpu = {}` "
"of `{}`.".format(ncpu, infopath, ncpuloc, fpath))
ndim = f.read_ints()
nparts[cpu] = f.read_ints()
localseed = f.read_ints()
nstar_tot = f.read_ints()
mstar_tot = f.read_reals('d')
mstar_lost = f.read_reals('d')
nsink = f.read_ints()
partfiles[cpu] = f
del ndim, localseed, nstar_tot, mstar_tot, mstar_lost, nsink
return nparts, partfiles
def open_unbinding(self, nsnap, nsim, cpu):
"""Open PHEW unbinding files."""
nsnap = str(nsnap).zfill(5)
cpu = str(cpu + 1).zfill(5)
fpath = join(self.paths.snapshots(nsim, "csiborg", tonew=False),
f"output_{nsnap}", f"unbinding_{nsnap}.out{cpu}")
return FortranFile(fpath)
def read_phew_id(self, nsnap, nsim, verbose):
nparts, __ = self.open_particle(nsnap, nsim)
start_ind = numpy.hstack([[0], numpy.cumsum(nparts[:-1])])
ncpu = nparts.size
clumpid = numpy.full(numpy.sum(nparts), numpy.nan, dtype=numpy.int32)
for cpu in trange(ncpu, disable=not verbose, desc="CPU"):
i = start_ind[cpu]
j = nparts[cpu]
ff = self.open_unbinding(nsnap, nsim, cpu)
clumpid[i:i + j] = ff.read_ints()
ff.close()
return clumpid
def read_halomaker_id(self, nsnap, nsim, halo_finder, verbose):
fpath = self.paths.halomaker_particle_membership(
nsnap, nsim, halo_finder)
fprint("loading particle IDs from the snapshot.", verbose)
pids = self.read_snapshot(nsnap, nsim, "pid")
fprint("mapping particle IDs to their indices.", verbose)
pids_idx = {pid: i for i, pid in enumerate(pids)}
# Unassigned particle IDs are assigned a halo ID of 0.
fprint("mapping HIDs to their array indices.", verbose)
hids = numpy.zeros(pids.size, dtype=numpy.int32)
# Read lin-by-line to avoid loading the whole file into memory.
with open(fpath, 'r') as file:
for line in tqdm(file, disable=not verbose,
desc="Processing membership"):
hid, pid = map(int, line.split())
hids[pids_idx[pid]] = hid
del pids_idx
collect()
return hids
def read_catalogue(self, nsnap, nsim, halo_finder):
if halo_finder == "PHEW":
return self.read_phew_clumps(nsnap, nsim)
elif halo_finder == "FOF":
return self.read_fof_halos(nsnap, nsim)
else:
raise ValueError(f"Unknown halo finder `{halo_finder}`.")
def read_fof_halos(self, nsnap, nsim):
"""
Read in the FoF halo catalogue.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
Returns
-------
structured array
"""
info = self.read_info(nsnap, nsim)
h = info["H0"] / 100
fpath = self.paths.fof_cat(nsnap, nsim, "csiborg")
hid = numpy.genfromtxt(fpath, usecols=0, dtype=numpy.int32)
pos = numpy.genfromtxt(fpath, usecols=(1, 2, 3), dtype=numpy.float32)
totmass = numpy.genfromtxt(fpath, usecols=4, dtype=numpy.float32)
m200c = numpy.genfromtxt(fpath, usecols=5, dtype=numpy.float32)
dtype = {"names": ["index", "x", "y", "z", "totpartmass", "m200c"],
"formats": [numpy.int32] + [numpy.float32] * 5}
out = numpy.full(hid.size, numpy.nan, dtype=dtype)
out["index"] = hid
out["x"] = pos[:, 0] * h + 677.7 / 2
out["y"] = pos[:, 1] * h + 677.7 / 2
out["z"] = pos[:, 2] * h + 677.7 / 2
# Because of a RAMSES bug x and z are flipped.
flip_cols(out, "x", "z")
out["totpartmass"] = totmass * 1e11 * h
out["m200c"] = m200c * 1e11 * h
return out
def read_phew_clumps(self, nsnap, nsim, verbose=True):
"""
Read in a PHEW clump file `clump_XXXXX.dat`.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
verbose : bool, optional
Verbosity flag.
Returns
-------
structured array
"""
nsnap = str(nsnap).zfill(5)
fname = join(self.paths.snapshots(nsim, "csiborg", tonew=False),
"output_{}".format(nsnap),
"clump_{}.dat".format(nsnap))
if not isfile(fname) or getsize(fname) == 0:
raise FileExistsError(f"Clump file `{fname}` does not exist.")
data = numpy.genfromtxt(fname)
if data.ndim == 1:
raise FileExistsError(f"Invalid clump file `{fname}`.")
# How the data is stored in the clump file.
clump_cols = {"index": (0, numpy.int32),
"level": (1, numpy.int32),
"parent": (2, numpy.int32),
"ncell": (3, numpy.float32),
"x": (4, numpy.float32),
"y": (5, numpy.float32),
"z": (6, numpy.float32),
"rho-": (7, numpy.float32),
"rho+": (8, numpy.float32),
"rho_av": (9, numpy.float32),
"mass_cl": (10, numpy.float32),
"relevance": (11, numpy.float32),
}
cols = list(clump_cols.keys())
dtype = [(col, clump_cols[col][1]) for col in cols]
out = cols_to_structured(data.shape[0], dtype)
for col in cols:
out[col] = data[:, clump_cols[col][0]]
# Convert to cMpc / h and Msun / h
out['x'] *= 677.7
out['y'] *= 677.7
out['z'] *= 677.7
# Because of a RAMSES bug x and z are flipped.
flip_cols(out, "x", "z")
out["mass_cl"] *= 2.6543271649678946e+19
ultimate_parent, summed_mass = self.find_parents(out)
out = add_columns(out, [ultimate_parent, summed_mass],
["ultimate_parent", "summed_mass"])
return out
def find_parents(self, clumparr):
"""
Find ultimate parent haloes for every PHEW clump.
Parameters
----------
clumparr : structured array
Clump array. Must contain `index` and `parent` columns.
Returns
-------
parent_arr : 1-dimensional array of shape `(nclumps, )`
The ultimate parent halo index of every clump.
parent_mass : 1-dimensional array of shape `(nclumps, )`
The summed substructure mass of ultimate parent clumps.
"""
clindex = clumparr["index"]
parindex = clumparr["parent"]
clmass = clumparr["mass_cl"]
clindex_to_array_index = {clindex[i]: i for i in range(clindex.size)}
parent_arr = numpy.copy(parindex)
for i in range(clindex.size):
cl = clindex[i]
par = parindex[i]
while cl != par:
element = clindex_to_array_index[par]
cl = clindex[element]
par = parindex[element]
parent_arr[i] = cl
parent_mass = numpy.full(clindex.size, 0, dtype=numpy.float32)
# Assign the clump masses to the ultimate parent haloes. For each clump
# find its ultimate parent and add its mass to the parent mass.
for i in range(clindex.size):
element = clindex_to_array_index[parent_arr[i]]
parent_mass[element] += clmass[i]
# Set this to NaN for the clumps that are not ultimate parents.
parent_mass[clindex != parindex] = numpy.nan
return parent_arr, parent_mass
###############################################################################
# Quijote particle reader #
###############################################################################
class QuijoteReader(BaseReader):
"""
Object to read in Quijote snapshots from the binary files.
Parameters
----------
paths : py:class`csiborgtools.read.Paths`
"""
def __init__(self, paths):
self.paths = paths
def read_info(self, nsnap, nsim):
snapshot = self.paths.snapshot(nsnap, nsim, "quijote")
header = readgadget.header(snapshot)
out = {"BoxSize": header.boxsize / 1e3, # Mpc/h
"Nall": header.nall[1], # Tot num of particles
"PartMass": header.massarr[1] * 1e10, # Part mass in Msun/h
"Omega_m": header.omega_m,
"Omega_l": header.omega_l,
"h": header.hubble,
"redshift": header.redshift,
}
out["TotMass"] = out["Nall"] * out["PartMass"]
out["Hubble"] = (100.0 * numpy.sqrt(
header.omega_m * (1.0 + header.redshift)**3 + header.omega_l))
return out
def read_snapshot(self, nsnap, nsim, kind):
snapshot = self.paths.snapshot(nsnap, nsim, "quijote")
info = self.read_info(nsnap, nsim)
ptype = [1] # DM in Gadget speech
if kind == "pid":
return readgadget.read_block(snapshot, "ID ", ptype)
elif kind == "pos":
pos = readgadget.read_block(snapshot, "POS ", ptype) / 1e3 # Mpc/h
pos = pos.astype(numpy.float32)
pos /= info["BoxSize"] # Box units
return pos
elif kind == "vel":
vel = readgadget.read_block(snapshot, "VEL ", ptype)
vel = vel.astype(numpy.float32)
vel *= (1 + info["redshift"]) # km / s
return vel
elif kind == "mass":
return numpy.full(info["Nall"], info["PartMass"],
dtype=numpy.float32)
else:
raise ValueError(f"Unsupported kind `{kind}`.")
def read_halo_id(self, nsnap, nsim, halo_finder, verbose=True):
if halo_finder == "FOF":
path = self.paths.fof_cat(nsnap, nsim, "quijote")
cat = FoF_catalog(path, nsnap)
pids = self.read_snapshot(nsnap, nsim, kind="pid")
# Read the FoF particle membership.
fprint("reading the FoF particle membership.")
group_pids = cat.GroupIDs
group_len = cat.GroupLen
# Create a mapping from particle ID to FoF group ID.
fprint("creating the particle to FoF ID to map.")
ks = numpy.insert(numpy.cumsum(group_len), 0, 0)
pid2hid = numpy.full(
(group_pids.size, 2), numpy.nan, dtype=numpy.uint64)
for i, (k0, kf) in enumerate(zip(ks[:-1], ks[1:])):
pid2hid[k0:kf, 0] = i + 1
pid2hid[k0:kf, 1] = group_pids[k0:kf]
pid2hid = {pid: hid for hid, pid in pid2hid}
# Create the final array of hids matchign the snapshot array.
# Unassigned particles have hid 0.
fprint("creating the final hid array.")
hids = numpy.full(pids.size, 0, dtype=numpy.uint64)
for i in trange(pids.size, disable=not verbose):
hids[i] = pid2hid.get(pids[i], 0)
return hids
else:
raise ValueError(f"Unknown halo finder `{halo_finder}`.")
def read_catalogue(self, nsnap, nsim, halo_finder):
if halo_finder == "FOF":
return self.read_fof_halos(nsnap, nsim)
else:
raise ValueError(f"Unknown halo finder `{halo_finder}`.")
def read_fof_halos(self, nsnap, nsim):
"""
Read in the FoF halo catalogue.
Parameters
----------
nsnap : int
Snapshot index.
nsim : int
IC realisation index.
Returns
-------
structured array
"""
fpath = self.paths.fof_cat(nsnap, nsim, "quijote", False)
fof = FoF_catalog(fpath, nsnap, long_ids=False, swap=False,
SFR=False, read_IDs=False)
cols = [("x", numpy.float32),
("y", numpy.float32),
("z", numpy.float32),
("vx", numpy.float32),
("vy", numpy.float32),
("vz", numpy.float32),
("group_mass", numpy.float32),
("npart", numpy.int32),
("index", numpy.int32)
]
data = cols_to_structured(fof.GroupLen.size, cols)
pos = fof.GroupPos / 1e3
vel = fof.GroupVel * (1 + self.read_info(nsnap, nsim)["redshift"])
for i, p in enumerate(["x", "y", "z"]):
data[p] = pos[:, i]
data[f"v{p}"] = vel[:, i]
data["group_mass"] = fof.GroupMass * 1e10
data["npart"] = fof.GroupLen
# We want to start indexing from 1. Index 0 is reserved for
# particles unassigned to any FoF group.
data["index"] = 1 + numpy.arange(data.size, dtype=numpy.int32)
return data
###############################################################################
# Supplementary functions #
###############################################################################
def make_halomap_dict(halomap):
"""
Make a dictionary mapping halo IDs to their start and end indices in the
snapshot particle array.
"""
return {hid: (int(start), int(end)) for hid, start, end in halomap}
def load_halo_particles(hid, particles, hid2map):
"""
Load a halo's particles from a particle array. If it is not there, i.e
halo has no associated particles, return `None`.
Parameters
----------
hid : int
Halo ID.
particles : 2-dimensional array
Array of particles.
hid2map : dict
Dictionary mapping halo IDs to `halo_map` array positions.
Returns
-------
parts : 1- or 2-dimensional array
"""
try:
k0, kf = hid2map[hid]
return particles[k0:kf + 1]
except KeyError:
return None
def convert_str_to_num(s):
"""
Convert a string representation of a number to its appropriate numeric type
(int or float).
Parameters
----------
s : str
The string representation of the number.
Returns
-------
num : int or float
"""
try:
return int(s)
except ValueError:
try:
return float(s)
except ValueError:
warn(f"Cannot convert string '{s}' to number", UserWarning)
return s

126
csiborgtools/read/utils.py
View file

@ -1,126 +0,0 @@
# Copyright (C) 2022 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.
from os.path import isfile
import numpy
from h5py import File
###############################################################################
# Array manipulation #
###############################################################################
def cols_to_structured(N, cols):
"""
Allocate a structured array from `cols`, a list of (name, dtype) tuples.
"""
if not (isinstance(cols, list)
and all(isinstance(c, tuple) and len(c) == 2 for c in cols)):
raise TypeError("`cols` must be a list of (name, dtype) tuples.")
names, formats = zip(*cols)
dtype = {"names": names, "formats": formats}
return numpy.full(N, numpy.nan, dtype=dtype)
def add_columns(arr, X, cols):
"""
Add new columns `X` to a record array `arr`. Creates a new array.
"""
cols = [cols] if isinstance(cols, str) else cols
# Convert X to a list of 1D arrays for consistency
if isinstance(X, numpy.ndarray) and X.ndim == 1:
X = [X]
elif isinstance(X, numpy.ndarray):
raise ValueError("`X` should be a 1D array or a list of 1D arrays.")
if len(X) != len(cols):
raise ValueError("Mismatch between `X` and `cols` lengths.")
if not all(isinstance(x, numpy.ndarray) and x.ndim == 1 for x in X):
raise ValueError("All elements of `X` should be 1D arrays.")
if not all(x.size == arr.size for x in X):
raise ValueError("All arrays in `X` must have the same size as `arr`.")
# Define new dtype
dtype = list(arr.dtype.descr) + [(col, x.dtype) for col, x in zip(cols, X)]
# Create a new array and fill in values
out = numpy.empty(arr.size, dtype=dtype)
for col in arr.dtype.names:
out[col] = arr[col]
for col, x in zip(cols, X):
out[col] = x
return out
def rm_columns(arr, cols):
"""
Remove columns `cols` from a structured array `arr`. Allocates a new array.
"""
# Ensure cols is a list
cols = [cols] if isinstance(cols, str) else cols
# Check columns we wish to delete are in the array
missing_cols = [col for col in cols if col not in arr.dtype.names]
if missing_cols:
raise ValueError(f"Columns `{missing_cols}` not in `arr`.")
# Define new dtype without the cols to be deleted
new_dtype = [(n, dt) for n, dt in arr.dtype.descr if n not in cols]
# Allocate a new array and fill in values
out = numpy.empty(arr.size, dtype=new_dtype)
for name in out.dtype.names:
out[name] = arr[name]
return out
def flip_cols(arr, col1, col2):
"""
Flip values in columns `col1` and `col2` of a structured array `arr`.
"""
if col1 not in arr.dtype.names or col2 not in arr.dtype.names:
raise ValueError(f"Both `{col1}` and `{col2}` must exist in `arr`.")
arr[col1], arr[col2] = numpy.copy(arr[col2]), numpy.copy(arr[col1])
###############################################################################
# h5py functions #
###############################################################################
def read_h5(path):
"""
Return and return and open `h5py.File` object.
Parameters
----------
path : str
Path to the file.
Returns
-------
file : `h5py.File`
"""
if not isfile(path):
raise IOError(f"File `{path}` does not exist!")
return File(path, "r")

2
csiborgtools/utils.py
View file

@ -244,7 +244,7 @@ def real2redshift(pos, vel, observer_location, observer_velocity, box,
Observer location in `Mpc / h`.
observer_velocity: 1-dimensional array `(3,)`
Observer velocity in `km / s`.
box : py:class:`csiborg.read.CSiBORGBox`
box : py:class:`csiborg.read.CSiBORG1Box`
Box units.
periodic_wrap : bool, optional
Whether to wrap around the box, particles may be outside the default