36-EFT.WP.EDX.Current v1.0 | 附录C. 数据集卡与管线卡

目录 ／ 文档-技术白皮书 ／ 36-EFT.WP.EDX.Current v1.0

附录C. 数据集卡与管线卡

I. 目的与覆盖（强制）

1. 本附录给出 EDX.Current 的**数据集卡（Dataset Card v1.0）与管线卡（Pipeline Card v1.0）**之字段、模板与校核规则，覆盖采集/仿真→对齐→路径改正→计算→反演→PPC→导出/发布全流程。
2. 统一口径（两式等价，显式路径/测度并记录 delta_form）：
   • 常量外提：T_arr = ( 1 / c_ref ) * ( ∫ n_eff d ell )
   • 一般口径：T_arr = ( ∫ ( n_eff / c_ref ) d ell )
3. SI 单位；硬门：check_dim=pass、Re{Z_eft}≥0、KK_consistency=pass、两口径 T_arr 一致（≤ u_Tarr）。

II. 命名与版本策略

• dataset_id / pipeline_id：[域]-[主题]-[序号]（示例：EDXCUR-001、EDXCUR-PIPE-001）。
• 版本：semver（如 1.0.0）；每一工件附 sha256 摘要。
• 时间指纹：created_at 使用 ISO 8601；工具链附 toolchain 与 env_lock。

III. 数据集卡（Dataset Card v1.0）字段定义

IV. 数据集卡——最小模板（可直接粘贴）

dataset_card:

dataset_id: "EDXCUR-001"

version: "1.0.0"

source: "measurement"

instruments:

- {type:"VNA", model:"—", cal:"2025-08-01", range:"10 MHz–20 GHz"}

deemb: {method:"TRL", version:"1.2", file:"/artifacts/deemb.json"}

sync: {ref:"10MHz", scheme:"shared_ref+trigger", dt_sync_s: 2.0e-12}

binding_ref: "LAY2PATH-xxxx"

arrival:

form: "n_over_c" # or "one_over_c_times_n"

gamma: "explicit"

measure: "d_ell"

c_ref: 299792458.0

Tarr_s: 1.234e-09

u_Tarr_s: 6.0e-12

delta_form: "n_over_c"

freq_grid_Hz: [ ... ]

Z_eft: {real:[...], imag:[...]}

argZ: [ ... ]

weights: {γ_main:[...], γ_side:[...] }

n_eff: [ ... ] # per segment or effective

qa_gates: {check_dim:"pass", passivity:"pass", KK:"pass"}

hashes:

files:

- {path:"/data/raw.s2p", sha256:"..."}

- {path:"/data/aligned.parquet", sha256:"..."}

created_at: "2025-09-15T09:30:00Z"

toolchain: {simstack:"1.0.0", kernels:"0.9.3"}

V. 管线卡（Pipeline Card v1.0）阶段与字段

• 标准阶段：ingest → align(I30-3) → path_correct → compute(Z_eft,T_arr,w_p) → invert(Mx-*) → ppc → export(cards)。
• 每阶段记录：id、命令/参数、输入/输出、版本与重跑策略。

字段定义

VI. 管线卡——最小模板（可直接粘贴）

pipeline_card:

pipeline_id: "EDXCUR-PIPE-001"

version: "1.0.0"

steps:

- {id:"ingest", cmd:"load_raw --input /data/raw.s2p"}

- {id:"align", cmd:"align --deemb /artifacts/deemb.json --sync /artifacts/sync.json"}

- {id:"path_correct", cmd:"pathfix --binding LAY2PATH-xxxx"}

- {id:"compute", cmd:"compute_zeft --arrival n_over_c"}

- {id:"invert", cmd:"invert --priors /cfg/priors.yaml --sampler NUTS"}

- {id:"ppc", cmd:"ppc --band ω1,ω2"}

- {id:"export", cmd:"export_cards --out /release"}

env_lock:

python: "3.11.6"

packages:

- {name:"numpy", version:"1.26.4"}

- {name:"scipy", version:"1.13.1"}

- {name:"pandas", version:"2.2.2"}

container:

image: "registry/edxcurrent:1.0.0"

digest: "sha256:..."

hardware: {cpu:"x86_64", gpu:"none", ram_GB:32}

seeds: {global: 20250915}

artifacts: ["/release/dataset_card.yaml","/release/pipeline_card.yaml"]

VII. 环境锁（Env Lock v1.0）字段与模板

env_lock:

os: {name:"Ubuntu", version:"22.04"}

interpreter: {name:"Python", version:"3.11.6"}

packages:

- {name:"numpy", version:"1.26.4", hash:"sha256:..."}

- {name:"scipy", version:"1.13.1", hash:"sha256:..."}

- {name:"pandas", version:"2.2.2", hash:"sha256:..."}

container:

image: "registry/edxcurrent:1.0.0"

digest: "sha256:..."

hardware: {cpu:"x86_64", gpu:"none", ram_GB:32}

VIII. QA 门与一致性检查（统一记录）

qa_gates:

check_dim: "pass" # 关键等式量纲一致

passivity: "pass" # min(Re{Z_eft}) ≥ 0

KK: "pass" # K–K 一致性

Tarr_dual:

diff_s: 3.0e-12 # |T_arr(两口径)差|

u_Tarr_s:6.0e-12

pass: true

IX. 审计轨与发布清单（模板）

audit_manifest:

run_id: "RUN-2025-09-15-001"

toolchain: {simstack:"1.0.0", kernels:"0.9.3"}

diffs: {params_changed: ["Ks_tau"], reason:"tuning within prior"}

seeds: {global: 20250915}

qa:

check_dim: "pass"

passivity: "pass"

KK: "pass"

logs: "/logs/run_001.txt"

release_manifest:

normative_refs: ["EFT.WP.Core.DataSpec v1.0","Methods.Repro v1.0"]

artifacts:

- "/release/dataset_card.yaml"

- "/release/pipeline_card.yaml"

- "/release/env_lock.yaml"

- "/release/audit_manifest.yaml"

hashes:

dataset_card: "sha256:..."

pipeline_card: "sha256:..."

env_lock: "sha256:..."

audit: "sha256:..."

X. 等价性与复现度量（阈值字段与伪代码）

equivalence_gates:

eps_Z_gate: 0.03 # 3% 相对误差

eps_phi_gate: 0.05 # rad

eps_w_gate: 0.10 # 权重总差

# 1) 两口径 T_arr 一致性

T1 = (1/c_ref) * sum(n_eff[i]*Δell[i]) # n_over_c

T2 = sum((n_eff[i]/c_ref) * Δell[i]) # one_over_c_times_n

assert abs(T1 - T2) <= u_Tarr

# 2) 快速门

assert min(Re(Z_eft)) >= 0.0

assert KK_consistency(Z_eft) is True

# 3) 等价性

eps_Z = max_abs(Z_eft_rep - Z_eft_pub) / max_abs(Z_eft_pub)

eps_phi= max_abs(phi_rep - phi_pub)

eps_w = max_over_ω(sum_abs(w_rep - w_pub))

assert eps_Z <= eps_Z_gate and eps_phi <= eps_phi_gate and eps_w <= eps_w_gate

XI. 目录结构建议（发布包）

/release/

dataset_card.yaml

pipeline_card.yaml

env_lock.yaml

audit_manifest.yaml

qa_report.json

artifacts/

deemb.json

sync.json

aligned.parquet

Z_eft.npy

XII. 错误代码（统一返回）

errors:

E_BAD_UNITS: "单位或量纲不一致"

E_KK_FAIL: "K–K 一致性失败"

E_PASSIVITY: "被动性失败 (Re{Z}<0)"

E_BINDING_MISSING: "缺失 binding_ref 或路径段"

E_DELTA_FORM_MISSING: "到达时 delta_form 未记录"

E_TARR_MISMATCH: "两口径 T_arr 不一致"

E_EQUIV_FAIL: "等价性阈值未通过"

E_ENVLOCK_FAIL: "环境锁重建失败或指纹不匹配"

E_QA_FAIL: "QA 门未通过"

XIII. 合规模板（一次性粘贴）

dataset_card: { ...按 IV 节模板... }

pipeline_card: { ...按 VI 节模板... }

env_lock: { ...按 VII 节模板... }

qa_gates: { ...按 VIII 节模板... }

audit_manifest: { ...按 IX 节模板... }

release_manifest:

normative_refs: ["EFT.WP.Core.DataSpec v1.0","Methods.Repro v1.0"]

artifacts: ["/release/dataset_card.yaml","/release/pipeline_card.yaml","/release/env_lock.yaml","/release/audit_manifest.yaml"]

hashes: {dataset_card:"sha256:...", pipeline_card:"sha256:...", env_lock:"sha256:...", audit:"sha256:..."}

equivalence_gates: {eps_Z_gate:0.03, eps_phi_gate:0.05, eps_w_gate:0.10}

XIV. 发布硬门（汇总）