Skip to content

Setting Up the Halo 569 Zoom Simulation

This documents how the CosmicGrain zoom region (Halo 569, in a 50 Mpc parent box) was originally selected and built. It's based on contemporaneous notes from November, cross-checked line-by-line against the actual scripts to confirm they match what was really run.

The entire pipeline only needed three Python scripts. A much larger number of similarly-named scripts exist in this folder (get_best_halo*.py, find_good_zoom_halo.py, trace_ids_to_lagrangian*.py, find_lagrangian_center.py, extract_halo_particles.py, and others) — those were exploratory alternatives written along the way and are not part of the documented, actually-used pipeline below. See Scripts not used at the bottom.

1. Run a parent simulation

  • 128³ particles, 50 Mpc box, dark matter only.
  • Initial conditions generated with MUSIC2: ./MUSIC make_parent.conf
  • Transfer functions (how density fluctuations evolve — encodes baryon physics, dark matter behavior, and cosmic evolution turning primordial fluctuations into galaxy seeds) generated with CAMB: ./camb mycamb.ini
  • FOF and SUBFIND must be enabled in the parent run — the next step reads the resulting halo catalog directly.

Config files: mycamb.ini, make_parent.conf (kept alongside this doc / in the ICs directory).

2. Choose the target halo

python choose_halo.py

Reads the parent run's z=0 SUBFIND catalog (fof_subhalo_tab_049.hdf5) and selects candidates with: - Mass 2–5 × 10¹² Msun/h - More than 500 particles (well-resolved)

For each candidate it reports mass, particle count, distance from the box edge, and distance to the nearest neighboring halo — so you can pick one that's both Milky Way-mass and reasonably isolated (avoiding both edge effects and contamination from nearby massive structures). Halo 569 in this catalog was the one selected.

This script has the SUBFIND file path and mass-cut thresholds hardcoded at the top — edit those directly for a different box/halo search rather than passing CLI arguments.

3. Extract the halo's particle IDs

python extract_zoom_ids.py

Reads Subhalo/SubhaloLenType and Subhalo/SubhaloOffsetType for the chosen halo (edit halo_id = 569 at the top of the script for a different target) from the same SUBFIND catalog, then pulls the corresponding contiguous block of dark matter particle IDs from the z=0 snapshot (PartType1/ParticleIDs).

Output: halo_569_particles.txt — one particle ID per line.

4. Trace those particles back to the initial conditions

python get_zoom_id_positions.py

This is the key step. It takes the particle ID list from step 3 and looks those same IDs up directly in the parent IC file (ICs/IC_parent_50Mpc_128_music.hdf5) — since Gadget particle IDs are conserved from the initial conditions through the entire simulation, this is a direct ID-matched lookup, not a fitted or approximate trace-back. The resulting positions are exactly where those particles started out at the beginning of the simulation — the Lagrangian region that will collapse into Halo 569 by z=0.

Positions are then normalized to the unit cube [0, 1], which is the coordinate convention MUSIC2 expects for its region_point_file / particle-based region definitions.

Output: halo_569_IC_positions_normalized.txt

Like extract_zoom_ids.py, the halo ID, input file, and parent IC path are hardcoded at the top of the script rather than passed as arguments.

5. Generate the zoom initial conditions

Add the normalized position file to the zoom MUSIC2 config:

# in make_zoom.conf
region_point_file = halo_569_IC_positions_normalized.txt

Then run:

./MUSIC make_zoom.conf

This produces the actual zoom initial conditions — high resolution in the Lagrangian region that collapses into Halo 569, progressively lower resolution further away — ready to hand to Gadget-4/CosmicGrain.

Config file: make_zoom.conf (kept alongside this doc / in the ICs directory).

Scripts not used in the final pipeline

These were written at various points while exploring halo selection and zoom-region definition, but the actual pipeline above didn't end up using them. Kept for reference/history rather than as a maintained interface; safe to move to an archive subfolder if you want to tidy the scripts directory further.

Alternative halo-ranking approaches (superseded by the simpler choose_halo.py mass+isolation cut): get_best_halo.py, get_best_halo_newer.py, find_good_zoom_halo.py, find_zoom_candidates.py

Alternative particle-ID / Lagrangian-tracing approaches (superseded by the simpler direct-ID-lookup method in get_zoom_id_positions.py): extract_halo_particles.py, extract_halo_ids_for_zoom.py, save_particleids_for_desired_halo.py, trace_ids_to_lagrangian.py, trace_ids_to_lagrangian_fixed.py, find_lagrangian_center.py

IC/zoom quality-check tools (not part of the core pipeline, but potentially still useful if you want to sanity-check a future zoom setup — worth deciding case by case whether to keep): check_parent_run.py, debug_ic_positions.py, check_ic_particles.py, ic_slice_viewer.py, ic_visualizer.py, check_zoom_quality.py, verify_halos.py, test_zoom_viability.py, extract_halo_manual.py (byte-identical duplicate of test_zoom_viability.py)

Ongoing zoom-run visualization/tracking (used throughout later development, not part of initial setup — see the main scripts/README.md for the confirmed paper-figure scripts instead): studyzoom.py, view_zoom.py, zoom_halo_report.py, zoom_halo_track.py, find_halo_center.py, find_zoom_halos.py