# Tools

Auxiliary scripts for profiling, legacy comparisons, and scale probes.

The main CPU expectation entrypoint is `../benchmark_cpu_expectation.py`.
For the current Vidal/MPS 1D-chain tests, prefer `../run_vidal_mps_cases.sh`.

Files here are intentionally secondary:

- `compare_vidal_backend_qmatchatea.py`: diagnostic comparison against QMatchaTea.
- `profile_vidal_chrome.py`: PyTorch CPU profiler for the Vidal path.
- `run_cpu_single_cases.sh`: single-node scale probes.
- `run_cpu_large_cases.sh`: two-node MPI scale probes.
- `run_vidal_segment_mpi_scan.sh`: rank/thread scaling scan for Vidal segmented MPI.
- `baseline_mps_expectation.py`: legacy MPS comparison CLI kept for old commands.
- `benchmark_tn_mpi.py`, `benchmark_search.py`, `benchmark_slice.py`, `benchmark_contract_sliced.py`, `check_tree.py`: old TN path-search/slicing experiments.
- `qibojit_reference_expectation.py`: state-vector reference helper.
- `validate_vidal_mpi_correctness.py`: focused Vidal MPI correctness helper.