dqmc-dev

edwnh's avatarfrom edwnh

Workflow for modifying DQMC C code, adding new measurements, and running tests. Use when editing source code, implementing new observables, debugging, or validating code changes.

⭐16stars🔀5forks📁View on GitHub🕐Updated Dec 31, 2025⬇️2downloads

When & Why to Use This Skill

This Claude skill provides a comprehensive workflow for developing and maintaining Deterministic Quantum Monte Carlo (DQMC) software in C. It streamlines the process of modifying source code, implementing new physical observables, and performing rigorous validation through automated build systems and smoke tests, ensuring high-performance scientific computing standards.

Use Cases

  • Scenario 1: Implementing new physical measurements by updating X-macro lists and measurement kernels in C source files.
  • Scenario 2: Debugging and validating complex code changes using integrated test suites and Green's function verification.
  • Scenario 3: Automating the build and smoke test cycle to ensure simulation stability and performance after code modifications.
  • Scenario 4: Extending data analysis capabilities by adding Python-based Jackknife analysis functions for new observables.
namedqmc-dev
descriptionWorkflow for modifying DQMC C code, adding new measurements, and running tests. Use when editing source code, implementing new observables, debugging, or validating code changes.

Code Development

Build & Test

# Build
make               # -> build/dqmc
make clean         # remove build/

# Test
cd test && make && ./test_greens

Smoke Test

After any code change:

dqmc-util gen Nx=4 Ny=4 U=4 dt=0.1 L=20 n_sweep_warm=50 n_sweep_meas=100
build/dqmc sim_0.h5
dqmc-util summary sim_0.h5

Adding New Measurements

  1. Add to X-macro list in src/rc/sim_types.h:

    // Add to MEAS_EQLT_LIST (or MEAS_UNEQLT_LIST for unequal-time)
    #define MEAS_EQLT_LIST \
        /* ... */ \
        X(my_meas, num_ij)  // pick an appropriate size expression
    
  2. Implement measurement kernel in src/rc/meas.c

  3. Add analysis function in dqmc_util/analyze_hub.py (and update reshaping/symmetrization if needed):

    @observable(
        description="my new observable",
        requires=("meas_eqlt/n_sample", "meas_eqlt/sign", "meas_eqlt/my_meas"),
    )
    def _my_meas(ctx):
        return core.jackknife_noniid(
            ctx.data["meas_eqlt/n_sample"],
            ctx.data["meas_eqlt/sign"],
            ctx.data["meas_eqlt/my_meas"],
        )
    

Key Code Patterns

Dual compilation: Files in src/rc/ compile twice. Use RC(name) macro for real/complex dispatch.

Memory: Use my_calloc() for 64-byte aligned allocations.

Profiling: Wrap with profile_begin(name) / profile_end(name).

Definition of Done

  • make succeeds without warnings
  • cd test && make && ./test_greens passes
  • Smoke run completes
  • If touching measurements: verify via dqmc-util summary