refactor: Consolidate element data, DRY up codebase, fix packaging

[ericchansen]

Summary

DRY up the codebase, consolidate element data into a single annotated source of truth, fix packaging gaps, and reduce test fixture duplication. Net result: -243 lines with zero test regressions.

Highlights

q2mm/elements.py — Single source of truth for periodic table data

• Element NamedTuple with (z, symbol, mass, covalent_radius) and cited provenance (NUBASE2020, Cordero 2008)
• All consumers (constants.py, identifiers.py, objective.py, molecule.py) now import from this one module
• Zero q2mm imports — breaks all circular dependency chains

DRY refactors

• forcefield.py: Generic _build_param_maps() + _match_for_export() replace 5 near-identical functions
• seminario.py: Generic _match_mode() + _collect_matching() replace 4 near-identical functions
• report.py/cli.py: Shared build_leaderboard_rows() helper
• test/_shared.py: 16 path constants + 3 molecule factories replace duplicates across 13 test files

Bug fixes

• from_molecules() feature drift: Silently lacked include_eigenmatrix and eigenmatrix_diagonal_only params that from_molecule() had — now delegates properly
• Duplicate Rhodium: Removed legacy "RH" entry from MASSES (104 entries for 103 elements)

Packaging & defensive imports

• [all] extras now include jax/jaxlib; [dev] includes openmm
• scipy import moved to lazy (inside methods) so module loads without scipy installed
• Benchmark CLI: added --data-dir for wheel installs where example data isn't bundled
• Tinker bond_tolerance now configurable (was hardcoded 1.3)

Dead code removed

• Deleted q2mm/parsers/_patterns.py (nothing imported from it)
• Removed duplicate MacroModel constants from constants.py
• Removed dead Q2MMMolecule import in forcefield.py

Testing

• 356 passed, 44 skipped — identical to baseline
• ruff check and ruff format clean

[Copilot]

Pull request overview

This PR refactors Q2MM to reduce duplication by centralizing element data and shared test fixtures, while also tightening packaging/extras and making some imports more defensive to improve install-time behavior.

Changes:

• Add q2mm/elements.py as a single source of truth for element symbols/masses/radii and update multiple consumers to import from it.
• DRY up repeated logic in core modules (e.g., forcefield/seminario) and consolidate duplicated test paths/molecule factories into test/_shared.py.
• Improve packaging/runtime robustness (extras updates, lazy SciPy imports, benchmark CLI --data-dir, configurable Tinker bond tolerance) and remove dead code.

Reviewed changes

Copilot reviewed 27 out of 27 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
test/test_reference_data.py	Uses shared fixture paths from test._shared instead of local constants.
test/test_models.py	Switches common fixture paths to test._shared; keeps test-specific paths local.
test/test_method_e_pipeline.py	Reuses shared SN2/Ethane paths/flags from test._shared.
test/test_method_d_e.py	Reuses shared SN2 Hessian path from test._shared.
test/test_eigenmatrix.py	Reuses shared SN2/Ethane paths/flags from test._shared.
test/integration/test_seminario_parity.py	Imports repo root + SN2 paths from test._shared to avoid duplication.
test/integration/test_scipy_optimizer.py	Uses shared molecule factories (make_water, make_diatomic).
test/integration/test_optimization_validation.py	Uses shared fixture paths and make_water factory.
test/integration/test_openmm_export.py	Uses shared fixture paths and shared molecule factories.
test/integration/test_openmm_backend.py	Uses shared fixture paths and shared molecule factories.
test/integration/test_jax_backend.py	Uses shared repo/SN2 paths and shared molecule factories.
test/integration/test_e2e_sn2_validation.py	Imports all SN2/CH3F fixture paths from test._shared.
test/conftest.py	Imports/re-exports shared constants/factories for fixture access.
test/_shared.py	New shared module defining common example paths, availability flags, and molecule factories.
q2mm/parsers/_patterns.py	Deleted unused shared regex/util module.
q2mm/optimizers/scipy_opt.py	Makes SciPy imports lazy (import inside methods) to allow module import without SciPy.
q2mm/optimizers/objective.py	Fixes from_molecules() feature drift by delegating to from_molecule() and centralizes atomic symbol lookup.
q2mm/models/seminario.py	Consolidates matching helpers and introduces DEFAULT_DFT_SCALING.
q2mm/models/molecule.py	Imports covalent radii from centralized q2mm.elements.
q2mm/models/identifiers.py	Imports two-letter element set from centralized q2mm.elements.
q2mm/models/forcefield.py	Adds generic param-map builder and export-matching helper to reduce duplication.
q2mm/elements.py	New centralized periodic table data and derived lookup tables.
q2mm/diagnostics/report.py	Extracts shared leaderboard row-building logic into build_leaderboard_rows().
q2mm/diagnostics/cli.py	Adds --data-dir support and reuses shared leaderboard row builder.
q2mm/constants.py	Re-exports MASSES from q2mm.elements and removes duplicated element data.
q2mm/backends/mm/tinker.py	Makes bond detection tolerance configurable via bond_tolerance.
pyproject.toml	Updates extras (all includes jax/jaxlib; dev includes openmm).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.