csiborgtools/scripts/pre_dumppart.py

# 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.
"""
Script to load in the simulation particles, load them by their clump ID and
dump into a HDF5 file. Stores the first and last index of each clump in the
particle array. This can be used for fast slicing of the array to acces
particles of a single clump.
"""

from datetime import datetime
from gc import collect

import h5py
import numba
import numpy
from mpi4py import MPI
from tqdm import trange

try:
    import csiborgtools
except ModuleNotFoundError:
    import sys

    sys.path.append("../")
    import csiborgtools

from argparse import ArgumentParser

# We set up the MPI
comm = MPI.COMM_WORLD
rank = comm.Get_rank()
nproc = comm.Get_size()

# And next parse all the arguments and set up CSiBORG objects
parser = ArgumentParser()
parser.add_argument("--ics", type=int, nargs="+", default=None,
                    help="IC realisations. If `-1` processes all simulations.")
args = parser.parse_args()

verbose = nproc == 1
paths = csiborgtools.read.Paths(**csiborgtools.paths_glamdring)
partreader = csiborgtools.read.ParticleReader(paths)
# Keep "ID" as the last column!
pars_extract = ['x', 'y', 'z', 'vx', 'vy', 'vz', 'M', "ID"]

if args.ics is None or args.ics[0] == -1:
    ics = paths.get_ics()
else:
    ics = args.ics


@numba.jit(nopython=True)
def minmax_clump(clid, clump_ids, start_loop=0):
    """
    Find the start and end index of a clump in a sorted array of clump IDs.
    This is much faster than using `numpy.where` and then `numpy.min` and
    `numpy.max`.
    """
    start = None
    end = None

    for i in range(start_loop, clump_ids.size):
        n = clump_ids[i]
        if n == clid:
            if start is None:
                start = i
            end = i
        elif n > clid:
            break
    return start, end


# MPI loop over individual simulations. We read in the particles from RAMSES
# files and dump them to a HDF5 file.
jobs = csiborgtools.fits.split_jobs(len(ics), nproc)[rank]
for i in jobs:
    nsim = ics[i]
    nsnap = max(paths.get_snapshots(nsim))
    fname = paths.particles_path(nsim)
    # We first read in the clump IDs of the particles and infer the sorting.
    # Right away we dump the clump IDs to a HDF5 file and clear up memory.
    print(f"{datetime.now()}: rank {rank} loading particles {nsim}.",
          flush=True)
    part_cids = partreader.read_clumpid(nsnap, nsim, verbose=verbose)
    sort_indxs = numpy.argsort(part_cids).astype(numpy.int32)
    part_cids = part_cids[sort_indxs]
    with h5py.File(fname, "w") as f:
        f.create_dataset("clump_ids", data=part_cids)
        f.close()
    del part_cids
    collect()

    # Next we read in the particles and sort them by their clump ID.
    # We cannot directly read this as an unstructured array because the float32
    # precision is insufficient to capture the clump IDs.
    parts, pids = partreader.read_particle(
        nsnap, nsim, pars_extract, return_structured=False, verbose=verbose)
    # Now we in two steps save the particles and particle IDs.
    print(f"{datetime.now()}: rank {rank} dumping particles from {nsim}.",
          flush=True)
    parts = parts[sort_indxs]
    pids = pids[sort_indxs]
    del sort_indxs
    collect()

    with h5py.File(fname, "r+") as f:
        f.create_dataset("particle_ids", data=pids)
        f.close()
    del pids
    collect()

    with h5py.File(fname, "r+") as f:
        f.create_dataset("particles", data=parts)
        f.close()
    del parts
    collect()

    print(f"{datetime.now()}: rank {rank} creating clump mapping for {nsim}.",
          flush=True)
    # Load clump IDs back to memory
    with h5py.File(fname, "r") as f:
        part_cids = f["clump_ids"][:]
    # We loop over the unique clump IDs.
    unique_clump_ids = numpy.unique(part_cids)
    clump_map = numpy.full((unique_clump_ids.size, 3), numpy.nan,
                           dtype=numpy.int32)
    start_loop = 0
    niters = unique_clump_ids.size
    for i in trange(niters) if verbose else range(niters):
        clid = unique_clump_ids[i]
        k0, kf = minmax_clump(clid, part_cids, start_loop=start_loop)
        clump_map[i, 0] = clid
        clump_map[i, 1] = k0
        clump_map[i, 2] = kf
        start_loop = kf

    # We save the mapping to a HDF5 file
    with h5py.File(paths.particles_path(nsim), "r+") as f:
        f.create_dataset("clumpmap", data=clump_map)
        f.close()

    del part_cids
    collect()