-
Notifications
You must be signed in to change notification settings - Fork 1
Home
GRANITE (General‑Relativistic Adaptive N‑body Integrated Tool for Extreme Astrophysics) is a high‑performance, open‑source numerical relativity engine designed to simulate the most violent events in the universe: the simultaneous inspiral, tidal disruption, and merger of multiple supermassive black holes in the presence of massive stars, magnetic fields, radiation, and neutrino microphysics.
Unlike traditional numerical relativity codes that focus on binary black hole or binary neutron star mergers, GRANITE is built from the ground up for N‑body supermassive black hole coalescence (N ≥ 3) with full physics:
- CCZ4 spacetime evolution – 22 evolved variables, 4th‑order finite differences, moving‑puncture gauge, Kreiss–Oliger dissipation.
- GRMHD with advanced Riemann solvers – HLLE and HLLD (Miyoshi & Kusano 2005), MP5/PPM/PLM reconstruction, constrained transport (∇·B = 0).
- Multi‑physics matter – tabulated nuclear EOS, M1 grey radiation transport, hybrid neutrino leakage + moments.
- Block‑structured AMR – Berger–Oliger subcycling, gradient‑based tagging, tracking spheres for moving horizons (full multi‑level AMR is under active development).
- Post‑processing suite – Ψ₄ gravitational‑wave extraction, spin‑weighted spherical harmonics, recoil velocity, Blandford–Znajek jet power, and EM light curves.
- HPC ready – MPI + OpenMP + HDF5 parallel I/O. GPU backends (CUDA/HIP) are on the roadmap.
- Latest stable release: v0.6.5 (“The Stability Update”)
- Test suite: 92 unit tests, 100% pass rate (GoogleTest)
-
Validated benchmarks:
single_puncture(Schwarzschild stability),B2_eq(equal‑mass binary black hole merger),gauge_wave(CCZ4 code validation) - Production‑grade components: CCZ4 core, GRMHD Valencia formulation, HLLD+CT, tabulated EOS, horizon finder, Python analysis tools.
| If you want to... | Go to |
|---|---|
| Install GRANITE | Installation Guide |
| Understand the engine architecture | Architecture Overview |
| Configure a simulation | Parameter Reference |
| Debug a failing simulation | Simulation Health & Debugging |
| Understand the physics equations | Physics Formulations |
| See benchmark results | Benchmarks & Validation |
| Deploy on HPC | HPC Deployment |
| Contribute code | Developer Guide |
| See what bugs have been fixed | Known Fixed Bugs |
| Understand initial data options | Initial Data |
| Learn about AMR design | AMR Design |
| Understand GW extraction | GW Extraction |
| See what's coming next | Roadmap |
| Get answers to common questions | FAQ |
| See the scientific context | Scientific Context |
| Navigate all documents | Documentation Index |
| Version history | CHANGELOG Summary |
| VORTEX N-Body Interactive WebGL Sim | VORTEX Engine |
- Follow the Installation Guide (Linux / WSL2 only).
- Run the health check:
python3 scripts/health_check.py - Launch the single‑puncture benchmark:
python3 scripts/run_granite.py run --benchmark single_punctureYou can also run the simulation on Diagnostics Mode:python3 scripts/sim_tracker.py build/bin/granite_main benchmarks/single_puncture/params.yaml - Explore the
dev_benchmark.pyAndsim_tracker.pylive dashboard for constraint monitoring, NaN forensics, and phase classification.
Follow these steps for a Quick Start:
git clone https://github.com/LiranOG/Granite.git && cd Granite
# Install dependencies (Ubuntu/WSL2)
sudo apt update && sudo apt install -y build-essential cmake libhdf5-dev libopenmpi-dev libyaml-cpp-dev
# Build
python3 scripts/run_granite.py build --release
# Pre-flight check
python3 scripts/health_check.py
# Run single-puncture benchmark
python3 scripts/run_granite.py run --benchmark single_puncture
# Run BBH benchmark (~99 min)
python3 scripts/run_granite.py run --benchmark B2_eqGRANITE welcomes contributions from the scientific and open-source communities. See Developer Guide and .github/CONTRIBUTING.md.
For institutional partnership, HPC allocations, or academic collaboration: open a GitHub Issue tagged [partnership].
"Simulate the unimaginable."
GRANITE v0.6.5 — April 10, 2026 — LiranOG
v0.6.8 · Repository · Issues
- 🩺 Simulation Health & Debugging
- 📊 Benchmarks & Validation
- 🗂️ AMR Design
- 🖥️ HPC Deployment
- 🌀 VORTEX Engine
"Simulate the unimaginable."