{
  "markdown": "# Determinantal quantum Monte Carlo for the Hubbard model\n\nHigh-performance DQMC (C) for the Hubbard model, plus Python utilities (`dqmc_util`) for generating inputs and analyzing outputs.\n\n## Supported platforms\n\n- Linux and macOS\n- Windows is not supported natively; use WSL2 or GitHub Codespaces if needed.\n\n## Dependencies\n\n**Build (Linux):**\n- Intel compiler `icx`\n- Intel MKL headers and libraries\n- HDF5 headers and libraries (downloaded via `make deps`)\n\n**Build (macOS):**\n- `clang` (Xcode Command Line Tools)\n- HDF5 headers and libraries (downloaded via `make deps`)\n\n**Python (`dqmc_util`):**\n- Installed via `pip install -e .` (included in `environment-*.yml`)\n- Runtime deps: `numpy`, `scipy`, `h5py`\n\n## Install and build\n\n1. Create the conda environment:\n   ```bash\n   # macOS\n   conda env create -f environment-macos.yml\n\n   # Linux\n   conda env create -f environment-linux.yml\n   ```\n2. Activate it:\n   ```bash\n   conda activate dqmc\n   ```\n3. Download/extract the bundled HDF5 dependency (first time only):\n   ```bash\n   make deps\n   ```\n4. Build:\n   ```bash\n   make\n   ```\n   The binary is `build/dqmc`.\n\n## Quickstart (smoke run)\n\n```bash\nconda activate dqmc\nmake deps  # first time only\nmake\n\ndqmc-util gen Nx=4 Ny=4 U=4 dt=0.1 L=20 n_sweep_warm=50 n_sweep_meas=100\nbuild/dqmc sim_0.h5\ndqmc-util summary sim_0.h5\n```\n\n## Workflow\n\n1. Generate HDF5 simulation files: `dqmc-util gen` (alias: `dqmc-util gen-1band-hub`)\n2. Run Monte Carlo sweeps: `build/dqmc sim_0.h5` (optionally with checkpointing)\n3. Analyze results: `dqmc-util summary`, `dqmc-util print-n`, or `dqmc_util.analyze_hub` in Python\n\n### Run one file\n\n```bash\nbuild/dqmc sim_0.h5\ndqmc-util summary sim_0.h5\n```\n\n### Run many files (sharded queue)\n\nFor 2+ files, prefer the queue/worker system (`dqmc-util enqueue` + `dqmc-util worker`).\n\n```bash\ndqmc-util gen Nx=6 Ny=6 U=4 mu=0 dt=0.1 L=50 Nfiles=160 prefix=runs/U4_T0.25/bin\ndqmc-util enqueue queue 'runs/U4_T0.25/bin_*.h5'\ndqmc-util worker queue ./build/dqmc -n 8 -s 300 -t 3600\ndqmc-util queue-status queue\ndqmc-util print-n runs/U4_T0.25/\n```\n\n## Examples\n\nSee `examples/README.md` and:\n- `examples/mz2_vs_T/` - magnetic moment vs temperature\n- `examples/n_vs_mu/` - density vs chemical potential\n\n## HDF5 layout\n\nOne Markov chain = one HDF5 file = one “bin”.\n\n- `/metadata`: model parameters for analysis (e.g., `beta`, `mu`, `Nx`, `Ny`)\n- `/params`: simulation parameters and matrices (often stored in a shared `*.h5.params`)\n- `/state`: sweep counter, RNG state, auxiliary field configuration\n- `/meas_eqlt`: equal-time measurements\n- `/meas_uneqlt`: unequal-time measurements (optional)\n\nFor details and authoritative defaults, see `dqmc_util/gen_1band_hub.py` (`SimParams`).\n\n## Parameter notes\n\n- Trotter error rule of thumb: ensure `U * dt^2 <= 0.05`\n- `L` must be divisible by `n_matmul` and `period_eqlt`\n\n## AI agent docs\n\n- `AGENTS.md`: working on this codebase (build, architecture, guardrails)\n- `.claude/skills/`: task runbooks (generate, run, analyze, scans, dev, advanced)\n",
  "bytes": 3029,
  "sha": "3083b64b2619a151e5d13d1d550c25a0f274b72e80fb38e16f0adcc02d4744ce",
  "repo_slug": "edwnh/dqmc",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_edwnh_dqmc_claude_skills_dqmc_run_main_222bb4f7/readme"
}