Skip to content

Draft: Krane SEMF coefficients + Z/N quark decomposition (N2, reconciliation pending)#14

Draft
Cuuper22 wants to merge 1 commit into
mainfrom
claude/eager-cerf-kvmqta-n2
Draft

Draft: Krane SEMF coefficients + Z/N quark decomposition (N2, reconciliation pending)#14
Cuuper22 wants to merge 1 commit into
mainfrom
claude/eager-cerf-kvmqta-n2

Conversation

@Cuuper22

Copy link
Copy Markdown
Owner

N2 of 10: SEMF coefficients + quark-count decomposition — DRAFT, test reconciliation pending

Implementation (complete)

  • Sourced SEMF coefficients: Krane, Introductory Nuclear Physics (1988), Table 3.2 values land as a sourced calibration preset factory in gpu_stack/presets/nuclear.py. The standing blocker (source, SI conversion, pairing/reference-energy semantics) is addressed: the pairing-term A^(-1/2) convention and the meaning of nuclear_pairing_gap_reference_energy are verified against this model's symbolic structure and documented in the preset.
  • Quark-count decomposition: source and medium-component valence quark counts become derived equations via the quark-model identities U = 2Z + N, D = Z + 2N (PDG-referenced), turning the heaviest root-debt family's primitives (12+ quark-count roots at ~218 weight each) into derived values.
  • Redundant triplet-integrality equations removed where U + D = 3A is now implied.

Why draft

The agent hit the session usage limit mid-reconciliation. 54 tests across ~20 files still assert the old structure (trace shapes, removed equation names, old root counts) — dominated by tests/test_materials_calibrated_presets.py (10), lithography source/species boundary packs (9), and CLI/root-debt snapshots. Last observed: 54 failed, 643 passed. Probed values are correct (e.g. source_neutron_count resolves to 70, missing=0); the failures are stale expectations, not wrong physics — but each deserves deliberate review rather than bulk-bumping.

Remaining work

  1. Reconcile the 54 stale tests family-by-family.
  2. Re-verify root-debt before/after (root-debt --families --limit 5) and update count assertions (619 will drop).
  3. Full verifier + audit gate.

https://claude.ai/code/session_01Eu2JVnPFgMQftwYTP3cGQZ


Generated by Claude Code

… (tests pending)

Implementation (complete):
- gpu_stack/presets/nuclear.py: Krane Table 3.2 semi-empirical mass
  formula coefficients as a sourced calibration preset factory with the
  pairing-term convention (A^(-1/2) exponent) and reference-energy
  semantics documented against this model's symbolic structure.
- Quark-count decomposition across source and medium components using
  the quark-model identities U = 2Z + N and D = Z + 2N (PDG-referenced),
  turning primitive valence quark-count roots into derived equations.
- Redundant triplet-integrality equations removed where the sum
  identity U + D = 3A is now implied by the Z/N identities.

Status (honest): the structural reconciliation of the test suite is
NOT finished. 54 tests across ~20 files still assert the old root
structure (trace shapes, removed equation names, old root counts),
dominated by tests/test_materials_calibrated_presets.py and the
lithography source/species/medium boundary packs. Last observed run:
54 failed, 643 passed. The implementation imports cleanly and the
resolved values are correct where probed (source_neutron_count
resolves to 70 with no missing inputs).

This commit preserves the interrupted agent's work for a follow-up
reconciliation pass.

https://claude.ai/code/session_01Eu2JVnPFgMQftwYTP3cGQZ
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants