Skip to content

Latest commit

 

History

History
258 lines (171 loc) · 9.8 KB

README.md

File metadata and controls

258 lines (171 loc) · 9.8 KB

About

This experiment is based on article Primordial Black Hole clusters, phenomenology & implications by Juan Garcia-Bellido (shortly: JGB).

The goal of this experiment is to create a self-consistent model with Plummer density profile and log-normal mass spectrum, and then evolve it for Hubble time.

Mass spectra used in the article.

How to reproduce

To reproduce the experiment, follow these steps:

  • Activate the Agama environment:

    conda activate agama

  • Start Nemo (from nemo repository root):

    source start_nemo.sh

  • Switch to custom NEMO version:

    cd $NEMO
git remote add custom https://github.com/savchenkoyana/nemo.git
git checkout nbodyx
cd $NEMO/src/nbody/evolve/aarseth/nbody0
make nmax
cd $NEMO/src/nbody/evolve/aarseth/nbody1/source
make nmax
cd $NEMO/src/nbody/reduc/Makefile
make snapbinary
cp snapbinary $NEMOLIB/
cd $NEMO
make rebuild

    If you want to go back to the default NEMO version, run:

    cd $NEMO
git checkout master
make rebuild

  • Go to the experiment root directory:

    cd /path/to/Nbody/02_Reproduce_JGB/

  • To reproduce any experiment from the original article, run the corresponding sh-script. For example:

    bash sh_scripts/run_exp_MA.sh

    You can also choose your own parameters for distributions and run the commands from sh-script one-by-one (see next section).

    Note that all scripts in this experiment overwrite the existing files. Don't forget to backup your experiments before trying to reproduce them!

    Running Nbody0 simulation takes about 3 days depending on task you choose. Running M & A experiment with N = 20000 was not reasonable because it would take us more than two weeks, so it is recommended to use N = 10000 for this experiment.

The combinations of parameters from the original article and corresponding sh-scripts are listed below:

Experiment name $\mu$, $M_{☉}$ s, $M_{☉}$ $\sigma$ Plummer radius, pc Number of particles Script to reproduce evolution
$\sigma$ = 0.5 0 1 0.5 10 $2 \times 10^4$ sh_scripts/run_exp_sigma0.5.sh
$\sigma$ = 1 0 1 1 10 $2 \times 10^4$ sh_scripts/run_exp_sigma1.0.sh
$\sigma$ = 1.5 0 1 1.5 10 $2 \times 10^4$ sh_scripts/run_exp_sigma1.5.sh
M & A 10 1.5 0.954 10 $2 \times 10^4$ sh_scripts/run_exp_MA.sh

More detailed overview

This section desctibes how to perform the evolution of PBH cluster in an external potential of SMBH.

  • Create initial coordinates of cluster:

    python create_ic.py --mu <MU> --sigma <SIGMA> --scale <SCALE> --r <PLUMMER_RADIUS> --N <N>

    Here N is the number of particles in simulation, MU, SIGMA, and SCALE are log-normal distribution parameters of PBH mass spectrum, and PLUMMER_RADIUS is a characteristic size of Plummer density distribution (type python create_ic.py --help for more details).

    The above command will automatically create (or re-create) a directory with name snap_mu<MU>_s<SCALE>_sigma<SIGMA>_r<PLUMMER_RADIUS>_N<N> containing file IC.nemo with initial coordinates for evolution.

  • Preprocess data

    JGB writes:

    Clusters are themselves immersed in a central gravitational potential with orbital radius $R_c$ = 34 kpc and central mass $M = 4.37 × 10^{10} M_{☉}$ throughout the entire evolution. This is just a point mass approximation which leads to a circular movement of period T = 2.81 Gyr

    The easiest way to create this point-mass potential is to add a new particle representing the central mass to the existing snapshot with PBH cluster data:

    python preprocess_snap.py \
  --nemo-file `DIRNAME`/IC.nemo \
  --r 10 \
  --r-shift 34 0 0 \
  --v-shift 0 -74.35014 0 \
  --add-point-source \
  --source-mass 4.37e10

    After this step, you will get a new file <DIRNAME>/IC_preprocessed.nemo with the old data concatenated with the new data (a steady point of mass $4.37\times10^{10} M_\odot$ at (0, 0, 0)).

    To convert data to units with G=1 (kpc, km/s and $\sim 232533.7 \times M_{☉}$), run:

    snapscale in=`DIRNAME`/IC_preprocessed.nemo \
  out=`DIRNAME`/IC_preprocessed_nbody.nemo \
  mscale=4.300451321727918e-06

    For more details see section Units.

  • Run evolution of this snapshot with SMBH:

    nbody0 `DIRNAME`/IC_preprocessed_nbody.nemo \
  `DIRNAME`/out_nbody.nemo \
  tcrit=14 \
  deltat=0.01 \
  eps=0.0003684031498640387

    eps is computed in code and provided by preprocess_snap.py script at the end of its output.

  • It would be useful to postprocess your data to plot profiles, spectras, etc.

    To postprocess snapshot evolved in JGB potential (the original article), run:

    python postprocess.py --snap-file <DIRNAME>/<OUT_NAME>.nemo --remove-point-source --nbody

    The postprocessed file with name <OUT_NAME>_postprocessed.nemo will be stored in <DIRNAME> folder.

Explore results

Visualize cluster evolution

  • To visualize cluster evolution, run:

    snapplot <DIRNAME>/out.nemo

    Use these options for customization:

    snapplot <DIRNAME>/out.nemo xrange=<xmin>:<xmax> yrange=<ymin>:<ymax> times=<tmin>:<tmax>

  • There is also a possibility to visulaize the evolution using glnemo2.

  • Another option is to use custom visualization script from this repository:

    python animate.py --nemo-file <DIRNAME>/out_postprocessed.nemo --add-point-source

Plot density profile $$\rho(r)$$

Plot density profile $$\rho(r)$$ for the resulting snapshot and compare it with initial density:

python plot_density_profile.py --nemo-file <DIRNAME>/<OUT_NAME>_postprocessed.nemo --times <t1> <t2> ... <tn> --mu <MU> --sigma <SIGMA> --scale <SCALE> --r <PLUMMER_RADIUS>

--times <t1> <t2> ... <tn> means that all timestamps from snapshot that you want to use to plot the graph should be separated by a space. E.g., --times 0.0 1.0 2.0. Before feeding timestamps, make sure they are present in the snapshot. To get a list of timestamps from a snapshot, run:

python stat.py --nemo-files <DIRNAME>/<OUT_NAME>_postprocessed.nemo --n-timestamps <N>

where <N> is the desired number of timestamps.

If you used sh-scripts, check txt-file in directory with your snapshot

Plot Lagrange radii

To plot Lagrange radius at different timestamps for different experiments, run:

python plot_lagrange_radius.py --nemo-files <DIRNAME1>/<OUT_NAME>_postprocessed.nemo <DIRNAME2>/<OUT_NAME>_postprocessed.nemo

As NEMO's tool for computation of cluster's density center sometimes fail and I haven't fixed it yet, it is better to add --remove-outliers at the end of the command:

python plot_lagrange_radius.py --nemo-files <DIRNAME1>/<OUT_NAME>_postprocessed.nemo <DIRNAME2>/<OUT_NAME>_postprocessed.nemo --remove-outliers

You can compare your results with plots from the article:

Note that Lagrange radius at $t=0$ should be approximately 13 pc according to analytical expression for Plummer with size 10 pc.

Plot mass spectrum $$f(M)$$

Compute and plot mass spectrum for a given snapshot along with the original distribution function:

python plot_mass_spectrum.py --nemo-file <DIRNAME>/<OUT_NAME>_postprocessed.nemo --times <t1> <t2> ... <tn> --mu <MU> --sigma <SIGMA> --scale <SCALE> --r <PLUMMER_RADIUS>

The mass distribution for your snapshot (the resulting histograms) and original pdf (the line plot) should look like a log-normal distribution with your parameters at $t=0$. You can compare your results with the picture of log-normal distributions at the beginning of this README document.

To plot the distribution of masses only for particles inside the half-mass radius, run:

python plot_mass_spectrum.py --nemo-file <DIRNAME>/<OUT_NAME>_postprocessed.nemo --times <t1> <t2> ... <tn> --mu <MU> --sigma <SIGMA> --scale <SCALE> --r <PLUMMER_RADIUS> --lagrange --remove-outliers

Test pipeline

There are several ways to test your pipeline:

  1. You may evolve a cluster in its own gravitational field. The final density after the evolution should look like the initial density — this would indicate that your model is truly self-consistent. In our case it means that we should not add point mass and shifts when preprocessing.
  2. You can check how energy and virial ratio evolve during the simulation: 
    python stat.py --nemo-files <DIRNAME>/<OUT_NAME>.nemo --eps <eps> --virial
    Use the same <eps> as during the simulation. Do not use postprocessed snapshot with removed central mass.

Compare with other N-body methods

The comparison with other methods is descriped in details in README_NBODY.md.

Units

We use non-usual units in our experiments:

  • We use pc (length), km/s (velocity) and $M_{☉}$ (mass) units for creating a cluster model because of convenience
  • We use units with G=1 for evolution: kpc for lenght, km/s for velocity and $\sim 232533.7 \times M_{☉}$ for mass

Checklist

Here is a list of what we need to fully reproduce the article:

  • N-body simulation (simple version)
  • Comparison with other methods (in process)
  • Nbody6 (simple run)
  • Gravitational waves
  • Black hole mergers