Guide the human-in-the-loop derivation of the PRISM s-state correction function from 5D geometry. Implements P1-T3, the critical-path task. Always produces one derivation step at a time for user review. Use when the user says "continue the s-state derivation", "do the next derivation step", "P1-T3", or the orchestrator dispatches P1-T3 to prism-theoretician.
This is the most important single task in the entire research program. If the derivation produces Bethe-log values within 10% of +2.98, +2.81, +2.77, +2.75, +2.74 for n=1..5 without any fitting to the data, PRISM is validated. If not, PRISM is falsified at this precision.
.tex file.[[CHOICE: <description>]] and escalated.Read working_dir/derivations/P1-T3_s_state_correction.tex so far. Identify the next unresolved goal from the outline. Compose a LaTeX snippet (≤15 lines inline math) with an English justification tied to either:
prism_formal_spec.md (cite postulate id)Write the proposal to chat, not yet to disk:
### Proposed step N
**Goal:** <one sentence>
**Justification:** <one or two sentences, cite sources>
$$
<LaTeX snippet>
$$
**Output:** <what this step produces for the next step to consume>
Ask the user: ACCEPT / REJECT / REVISE?
.tex, update the symbolic form in .json, continue to step N+1.derivation_log.md with the user's reason, do not append, re-plan.On ACCEPT:
working_dir/derivations/P1-T3_s_state_correction.tex.working_dir/derivations/P1-T3_s_state_correction.json with:
{"step": N, "symbolic_form": "<sympy-parseable>", "variables": {...}, "postulates_used": [...]}
working_dir/derivations/derivation_log.md.The completed derivation must deliver all three:
n with no free parameters beyond α.When all three are met, the derivation is complete. Hand off to P1-T4 (prism-computer) for numerical evaluation.
Both are legitimate outcomes. Do not paper over either.
working_dir/derivations/P1-T3_s_state_correction.texworking_dir/derivations/P1-T3_s_state_correction.jsonworking_dir/derivations/derivation_log.md (appended)working_dir/derivations/P1-T3_s_state_correction.tex.meta.yaml per handoff contractEstimated: 80K-150K tokens, 2-3 weeks wall time including user review cycles.