v69_tomographic_redshift_evolution.py

"""V69 — Tomographic-bin redshift evolution of (Om, s8) direction.

X6 prereg locked at commit 026d1e9. Asks whether the (Omega_m,
sigma_8) banana orientation rotates with effective redshift across
tomographic bins WITHIN each shear survey.

Implementation note: per-tomographic-bin (Omega_m, sigma_8) chains
or marginalized posteriors are NOT a standard data product in the
KiDS-1000, DES Y3, or HSC Y3 public releases. The released chains
are joint-over-bins (or joint with statistic separation, e.g.
COSEBIs vs BP vs xi+/-, all joint over bins). Per-bin posteriors
would require re-running the likelihood code with bin masks applied,
which the X6 stop condition explicitly forbids ("Do not attempt to
extraction methods not in the original survey release. ...Do not
re-run likelihoods.").

This script performs the data-availability audit, documents the
inventory, and fires the locked outcome D (data unavailable).
"""
from __future__ import annotations
import json
from pathlib import Path

ROOT = Path(__file__).resolve().parent.parent
DATA = ROOT / "data"
RESULTS = ROOT / "results"


def _audit_des_y3() -> dict:
    info = {"survey": "DES Y3", "expected_n_bins": 4,
            "expected_zeff": [0.34, 0.52, 0.74, 0.96]}
    chain = DATA / "des_y3" / "chain_3x2pt_lcdm_SR_maglim.txt"
    info["available_chain"] = str(chain.relative_to(ROOT)) if chain.exists() else None
    info["chain_kind"] = ("joint 3x2pt (cosmic shear + GGL + galaxy clustering); "
                         "joint over all tomographic bins (per-bin chains NOT released)")
    info["per_bin_chain_available"] = False
    info["per_bin_summary_available"] = False
    info["reason_unavailable"] = (
        "DES Y3 cosmic-shear-only and 3x2pt chain releases provide joint "
        "posteriors marginalized over all source tomographic bins; per-source-bin "
        "(Omega_m, sigma_8) marginalized posteriors are not separately released "
        "in machine-readable form. Reconstructing per-bin posteriors would "
        "require re-running the DES Y3 likelihood with bin masks, which is "
        "forbidden by X6 stop condition."
    )
    return info


def _audit_kids() -> dict:
    info = {"survey": "KiDS-1000", "expected_n_bins": 5,
            "expected_zeff": [0.26, 0.40, 0.56, 0.79, 0.98]}
    chain_root = (DATA / "kids1000" / "KiDS1000_cosmis_shear_data_release"
                  / "chains_and_config_files" / "main_chains_iterative_covariance")
    if chain_root.exists():
        statistics = sorted([p.name for p in chain_root.iterdir() if p.is_dir()])
    else:
        statistics = []
    info["available_chain_statistics"] = statistics  # cosebis, bp, xipm
    info["chain_kind"] = ("per-statistic (COSEBIs, BP, xi+/-) chains, each "
                         "joint over the 5 tomographic bins; per-bin chains "
                         "NOT released")
    info["per_bin_chain_available"] = False
    info["per_bin_summary_available"] = False
    info["reason_unavailable"] = (
        "KiDS-1000 public release provides per-statistic chains "
        "(COSEBIs, band powers, xi+/-) each marginalized over all 5 tomographic "
        "bins. Per-bin (Omega_m, sigma_8) posteriors are not separately released. "
        "Reconstructing them requires re-running CosmoSIS/MontePython likelihoods "
        "with bin masks, forbidden by X6 stop condition."
    )
    return info


def _audit_hsc() -> dict:
    info = {"survey": "HSC Y3", "expected_n_bins": 4,
            "expected_zeff": [0.30, 0.61, 0.82, 1.18]}
    chain_dir = DATA / "hsc_y3_chains"
    if chain_dir.exists():
        candidates = sorted([p.name for p in chain_dir.iterdir()])
    else:
        candidates = []
    info["chain_dir_contents"] = candidates
    info["available_chain"] = None  # checked below
    summary = DATA / "hsc_y3" / "summary_lcdm.json"
    info["available_summary"] = str(summary.relative_to(ROOT)) if summary.exists() else None
    info["chain_kind"] = ("Gaussian summary only (literature mean + sigma + correlation "
                         "from Li+23 / Dalal+23)")
    info["per_bin_chain_available"] = False
    info["per_bin_summary_available"] = False
    info["reason_unavailable"] = (
        "HSC Y3 chains are not on disk in this repository; the catalog entry "
        "is a literature Gaussian summary from Li+23 / Dalal+23. Even if the "
        "joint chain were available, per-bin posteriors are not in the public "
        "release. Reconstructing requires re-running the HSC likelihood, "
        "forbidden by X6 stop condition."
    )
    return info


def main() -> None:
    print("V69: tomographic-bin redshift evolution audit (X6 prereg).")
    print("Checking per-survey per-bin (Omega_m, sigma_8) posterior availability...")

    surveys = [_audit_des_y3(), _audit_kids(), _audit_hsc()]
    n_accessible = sum(1 for s in surveys
                       if s["per_bin_chain_available"] or s["per_bin_summary_available"])

    for s in surveys:
        print(f"\n  {s['survey']}: expected {s['expected_n_bins']} bins at "
              f"z_eff={s['expected_zeff']}")
        print(f"    chain kind:                 {s['chain_kind']}")
        print(f"    per-bin chain available:    {s['per_bin_chain_available']}")
        print(f"    per-bin summary available:  {s['per_bin_summary_available']}")
        print(f"    reason: {s['reason_unavailable']}")

    print(f"\n  surveys with per-bin posteriors: {n_accessible} / 3")
    if n_accessible == 0:
        outcome = "D-data_unavailable_per_stop_condition"
        msg = ("Per-bin (Omega_m, sigma_8) posteriors are not in machine-"
               "readable form in any of the three shear surveys' public "
               "releases. X6 outcome D fires per the locked prereg's stop "
               "condition. The within-shear question cannot be answered "
               "with publicly-released data alone.")
    else:
        outcome = "X6-incomplete_partial_data"
        msg = (f"Only {n_accessible}/3 surveys have per-bin posteriors; "
               "X6 prereg requires N>=10 bins for outcome A, so partial "
               "data does not suffice for the locked rule.")
    print(f"\n=== OUTCOME: {outcome} ===")
    print(msg)

    out = {
        "battery": "V69",
        "prereg": "X6 (preregs/X6_v69_tomographic_redshift_evolution_prereg.md, locked at 026d1e9)",
        "branch": "optimal_transport",
        "description": (
            "Tomographic-bin redshift evolution audit. Asks whether per-bin "
            "(Omega_m, sigma_8) banana orientation rotates with z_eff_bin "
            "within each shear survey."
        ),
        "audit_per_survey": surveys,
        "n_surveys_with_per_bin_data": n_accessible,
        "outcome_category": outcome,
        "interpretation": msg,
        "remediation_required_for_outcome_ABC": (
            "Re-run the per-survey likelihood codes (CosmoSIS for DES Y3/KiDS, "
            "the HSC pipeline for HSC Y3) with bin masks applied to extract "
            "per-bin (Omega_m, sigma_8) posteriors. This is outside the scope "
            "of the V63-V72 paper program and the X6 stop condition forbids "
            "it as part of V69."
        ),
    }
    RESULTS.mkdir(parents=True, exist_ok=True)
    out_path = RESULTS / "v69_tomographic_redshift_evolution_results.json"
    out_path.write_text(json.dumps(out, indent=2, default=float))
    print(f"\nWrote {out_path.relative_to(ROOT)}")


if __name__ == "__main__":
    main()