37-EFT.WP.EDX.HighSpeed v1.0 | 附录 B. 实现绑定与函数原型

目录 ／ 文档-技术白皮书 ／ 37-EFT.WP.EDX.HighSpeed v1.0

附录 B. 实现绑定与函数原型

I. 目标与覆盖

1. 绑定/实现范围：I30-HF（布局-路径-模态绑定）、I40-HF（S↔Z/3D-EM 协同）、Methods.SimStackHF（仿真栈）、Mx-*（先验/似然/证据/反演）、M10-HF（计量与对齐）。
2. 强制口径（两式等价，须显式路径/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. 发布硬门：check_dim=pass、passivity(Re{Z_eft}≥0)、KK_consistency=pass、两口径 T_arr 一致（≤ u(T_arr)）。

II. 命名空间与版本

1. 命名空间：
   • edx.highspeed.binding（I30-HF）
   • edx.highspeed.circuit（I40-HF）
   • edx.highspeed.simstack（Methods.SimStackHF）
   • edx.highspeed.inference（Mx-*）
   • edx.highspeed.metrology（M10-HF）
2. 版本策略：api_version: "1.0"（语义化 semver）；所有响应附 {api_version, impl_version, checksum}。

III. 公共数据结构（最小集合）

types:

PathSegment:

fields: {layer:str, len_m:float, n_eff:float, neigh:str}

PathSpec:

fields: {id:str, segments:list[PathSegment], weight_init:float} # 0..1

ModeSpec:

fields: {name:str, Zc_ohm:list[float], alpha_per_m:list[float], beta_per_m:list[float]}

ArrivalRecord:

fields:

{form:enum["n_over_c","one_over_c_times_n"], gamma:"explicit", measure:"d_ell",

c_ref:float, Tarr_s:float, u_Tarr_s:float, delta_form:str}

QAGates:

fields: {check_dim:enum["pass","fail"], passivity:enum["pass","fail"], KK:enum["pass","fail"]}

Ports: {names:list[str], zref_ohm:list[float]}

Deemb: {method:str, version:str, artifact:str, baseline_id:str}

Sync: {ref:str, scheme:str, dt_sync_s:float}

BindingHF: {id:str, paths:list[PathSpec], modes:list[ModeSpec]}

Weights: {w:map[(path_id,mode)->list[float]]} # w_{p,m}(ω)

Grid: {freq_grid_Hz:list[float]}

MixedMode: {enabled:bool, T_mm:str, Z0_mm_ohm:list[float]}

IV. I30-HF｜布局-路径-模态绑定函数原型

edx.highspeed.binding:

- id: "I30-HF.bind_layout_hf"

proto: "bind_layout_hf(layout:any, stackup:any, returns:any, constraints:any) -> BindingHF"

requires: ["显式路径分段与 n_eff(seg)", "回流/跨接标注"]

qa: ["check_dim"]

- id: "I30-HF.mode_project"

proto: "mode_project(dataset:any, M_omega:any) -> x_m"

notes: ["保持功率守恒"]

- id: "I30-HF.mode_merge"

proto: "mode_merge(x_m:any, rules:any) -> dataset'"

qa: ["passivity","KK"]

- id: "I30-HF.path_correct_hf"

proto: "path_correct_hf(dataset:any, binding:BindingHF, arrival:ArrivalRecord) -> aligned"

post: ["写入 arrival、两口径 T_arr 一致性校核"]

V. I40-HF｜S↔Z/协同接口函数原型

edx.highspeed.circuit:

- id: "I40-HF.map_S_to_Z"

proto: "map_S_to_Z(S:any, Znorm:list[float], mode:enum['single','mixed'], T_mm?:any, Z0_mm?:any) -> {Z_eft:any, argZ:list[float], Zc_ohm:list[float], qa:QAGates}"

- id: "I40-HF.em_port_align"

proto: "em_port_align(em:any, Znorm:list[float], T_mm?:any) -> S_ren:any"

- id: "I40-HF.cosimulate"

proto: "cosimulate(em:any, circuit_netlist:any, binding:BindingHF, options:any) -> cosim_handle"

VI. Methods.SimStackHF｜仿真栈函数原型

edx.highspeed.simstack:

- id: "SimStackHF.build"

proto: "build(netlist:any, layout:any, binding:BindingHF, options:any) -> sim_handle"

- id: "SimStackHF.forward"

proto: "forward(sim_handle:any, theta:map, grid:Grid) -> {Z_eft:any, argZ:list[float], T_group_s:list[float], w:Weights, deltaZ_rad?:any}"

- id: "SimStackHF.invert"

proto: "invert(sim_handle:any, data:any, priors:any, sampler:str='NUTS') -> {posterior:any, logZ:float, summary:any}"

- id: "SimStackHF.ppc"

proto: "ppc(sim_handle:any, posterior:any, grid:Grid) -> {residual_spectrum:list[float], gates:QAGates}"

- id: "SimStackHF.export"

proto: "export(sim_handle:any, format:enum['cards','json']) -> artifact_paths:list[str]"

VII. Mx-*｜先验/似然/证据（HF 约束）

edx.highspeed.inference:

priors:

n_eff_seg: {type:"normal", mu:"n_eff_hat", sigma:"0.05*mu"}

sigma_eff: {type:"lognormal", mu:-12.0, s:0.8, unit:"S/m"}

Ks_amp: {type:"halfnormal", sigma:0.5}

Ks_tau: {type:"loguniform", low:1e-12, high:1e-7, unit:"s"}

w_dirichlet: {type:"dirichlet", alpha:[0.3,0.3,0.3]} # Σ≤1

dt_sync: {type:"normal", mu:0.0, sigma:2.0e-12, unit:"s"}

deltaZ_rad_Re: {type:"halfnormal", sigma:1.0, constraint:"Re≥0"}

likelihoods:

complex_Z: "joint_gaussian([Re,Im]; W_ω)" # ε=Z_meas−Z_model

kpi_terms: "penalty(E_phase,GDR,T_group)"

radiation: "penalty(P_rad − 0.5*Re(ΔZ_rad)*|I_port|^2)"

evidence:

compute: ["laplace","nested_sampling"]

gates: ["passivity","KK","check_dim"]

VIII. M10-HF｜计量与对齐函数原型

edx.highspeed.metrology:

- id: "M10-HF.align"

proto: "align(raw:any, deemb:Deemb, sync:Sync, binding:BindingHF, arrival:ArrivalRecord, Znorm:list[float], mixed?:MixedMode) -> aligned"

post: ["输出 Z_eft/argZ/T_group", "写入 QAGates 与 KPI"]

- id: "M10-HF.arrival_consistency"

proto: "arrival_consistency(arrival:ArrivalRecord, binding:BindingHF) -> {diff_s:float, pass:bool}"

rule: "|T_arr(n_over_c) − T_arr(one_over_c_times_n)| ≤ u_Tarr"

IX. QA 门与检查函数

qa:

- id: "check_dim"

proto: "check_dim(obj:any) -> enum['pass','fail']"

- id: "check_passivity"

proto: "check_passivity(Z:any) -> {min_Re:float, pass:bool}"

- id: "check_KK"

proto: "check_KK(Z:any, grid:Grid) -> enum['pass','fail']"

- id: "check_weights"

proto: "check_weights(w:Weights) -> {sum_max:float, pass:bool}" # Σ_{p,m} w_{p,m} ≤ 1

- id: "check_radiation_positive"

proto: "check_radiation_positive(deltaZ_rad:any) -> {min_Re:float, pass:bool}"

X. 错误码（统一）

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_QA_FAIL: "QA 门未通过"

E_PRIOR_INVALID: "先验或边界不合法"

E_PORT_NORM: "端口归一/混合模基不一致"

E_RADIATION_NEG: "ΔZ_rad 非正实"

XI. 请求/响应模式（最小示例）

request:

api_version: "1.0"

op: "s2z_forward"

payload:

S_path: "/artifacts/S.s2p"

Znorm_ohm: [50.0, 50.0]

mixed_mode: {enabled:true, T_mm:"/cfg/T_mm.yaml", Z0_mm_ohm:[100.0,25.0]}

grid: {freq_grid_Hz: [1e9, 1.01e9, ...]}

response:

api_version: "1.0"

result:

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

argZ: [...]

Zc_ohm: [...]

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

checksum: "sha256:..."

XII. 合规模板（到达时与绑定）

arrival:

form: "n_over_c" # 或 "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"

binding_hf:

id: "LAY2PATH-HF-0001"

paths:

- id: "γ_main"

segments:

- {layer:"L3", len_m:0.0100, n_eff:2.12, neigh:"GND_L2"}

- {layer:"L4", len_m:0.0020, n_eff:2.18, neigh:"GND_L5"}

weight_init: 0.88

- id: "γ_side"

segments:

- {layer:"L3", len_m:0.0035, n_eff:2.40, neigh:"slot_L2"}

weight_init: 0.12

modes:

- {name:"DM", Zc_ohm:[...], alpha_per_m:[...], beta_per_m:[...]}

- {name:"CM", Zc_ohm:[...], alpha_per_m:[...], beta_per_m:[...]}

XIII. 端到端调用顺序（伪代码）

# 1) 绑定与对齐

binding = bind_layout_hf(layout, stackup, returns, constraints)

aligned = edx.highspeed.metrology.align(raw, deemb, sync, binding, arrival, Znorm, mixed_mode)

# 2) S→Z 与协同

Zpack = map_S_to_Z(S_aligned, Znorm, mode="mixed", T_mm=Tmm, Z0_mm=Z0mm)

# 3) 仿真栈

sim = SimStackHF.build(netlist, layout, binding, options)

out = SimStackHF.forward(sim, theta0, grid)

# 4) 反演与 PPC

post = SimStackHF.invert(sim, aligned, priors, sampler="NUTS")

pp = SimStackHF.ppc(sim, post.posterior, grid)

# 5) 导出

SimStackHF.export(sim, format="cards")

XIV. 约束与门限（HF 发布硬门）