25-EFT.WP.STG.Dynamics v1.0 | Appendix C — Manifest Templates and Examples (stg manifest)

Home ／ Docs-Technical WhitePaper ／ 25-EFT.WP.STG.Dynamics v1.0

Appendix C — Manifest Templates and Examples (stg manifest)

One-Sentence Goal
Provide copy-ready manifest templates and examples for STG dynamical systems—covering data → graph → operators → kernels → propagation/assimilation → uncertainty → runtime → publication—so results are reproducible, auditable, and comparable.

I. Scope and Objects

• Objects: graph, operators (L / A / H), kernel K = g(L), dynamics/solver, filter (Q / R / x0 / P0), uncertainty (u_c, g, nu_eff, α), contracts, runtime (window/watermark), artifacts, signature.
• Inputs: per-stage *.hash, RefCond, unit dictionary units.*, key statistics, and dual-form gaps.
• Outputs: manifest.stg.* (case and runtime variants).
• Constraints: non_decreasing(tau_mono); all metrics pass check_dim( y − f(x) ); manifest includes signature and a digest of all fields.

II. Terms and Fields

• Identity & time: TraceID, ts_publish, tau_mono, wm(t) (watermark).
• Graph & operators: graph.hash, L.hash, A.hash, H.hash, spectrum = [ λ_min, λ_max ].
• Kernel & stepping: K.hash, mode ∈ { diff, wave, mix }, Δt, solver.
• Assimilation: filter ∈ { KF, UKF, PF }, Q.hash, R.hash, x0.hash, P0.hash.
• Uncertainty: u_num, u_model, u_c, k, g, nu_eff, α.
• Contracts & dual forms: rules.hash, report.hash, delta_form_stream_p95, delta_form_unc_p95.
• Ref conditions & units: RefCond, units.*.
• Artifacts & signature: artifacts.*, signature.

III. Axioms P70C-*

• P70C-1: Manifests use hash(*) as primary indices, traceable to raw data and code versions.
• P70C-2: Every numeric field declares unit(field) and dim(field) and passes check_dim.
• P70C-3: Run dual forms and log gaps: statistical (event-time vs processing-time) and uncertainty (LPU vs MC).
• P70C-4: signature must cover all fields except the signature block itself.

IV. Minimal Equations S70C-*

• Signature domain:
  signature = Sign_{key_id}( SHA256( normalize( manifest \ signature ) ) ).
• Composite uncertainty:
  u_c^2 = J V_x J^T + u_num^2 + u_model^2, g = k * u_c.
• Dual-form gap:
  delta_form_unc = | u_c^{LPU} − u_c^{MC} | with rolling p95.
• Watermark monotonicity:
  wm(t_{i+1}) ≥ wm(t_i).

V. Workflow M70-3 (Generate → Verify → Sign → Publish)

• Aggregate stage outputs & metadata to form manifest_draft.
• Run assert_runtime_contracts(state, rules) to produce contracts.report and percentile stats.
• Verify units/dimensions, spectral bounds, coverage and dual-form gaps; failures block publication.
• Normalize field order/encoding, compute digest, sign_manifest to write signature.
• Publish to object storage & panel index; persist manifest.hash and replay pointers.

VI. Manifest-Level Contracts C70C-*

• C70C-01: Presence of minimal top keys { TraceID, ts_publish, graph.hash, L.hash, RefCond, units, artifacts, signature }.
• C70C-02: Spectrum exists and satisfies λ_max ≤ λ_cap.
• C70C-03: coverage ≥ 1 − α − ε_cov; delta_form_stream_p95 ≤ tol_stream; delta_form_unc_p95 ≤ tol_unc.
• C70C-04: Every *.hash resolves to a concrete artifact (code/data/model) in the registry.

VII. I70- Bindings (Manifest Functions)*

• emit_runtime_manifest(results, policy) -> manifest.stg.runtime
• emit_case_manifest(results, policy) -> manifest.stg.case
• sign_manifest(manifest, key_id) -> signature; verify_manifest(manifest, keyring) -> ok
• assert_runtime_contracts(state, rules) -> report
  (See Appendix A: I70-14B / 161 / 162 / 163 / 14A)

VIII. Cross-References

• Contract thresholds & playbooks: Appendix B (C70-*).
• Filtering & metrics: Chs. 12 & 13.
• Runtime windowing & replay: Ch. 14.
• Use-case field deltas & panel mapping: Ch. 15.

IX. Quality & Risk Control

• SLO: manifest generation latency_p95 ≤ B_mod; failures enter degraded path with backfill.
• Drift monitoring: manifest.diff (key-level delta vs prior version) is archived.
• Rollback: on verification failure, revert to the latest verify_manifest == ok version and annotate rollback.from.

Template (Copy-Ready)

manifest_version: "stg.v1.0"

TraceID: "<uuid>"

ts_publish: "<ISO8601>"

tau_mono: "<monotonic_time_ref>"

producer: { org: "<org>", service: "<svc>", version: "<semver>" }

RefCond:

timezone: "UTC"

units_policy: "SI"

sensors: ["<sensor_id>", "..."]

notes: "<free text>"

units:

x: "<unit(x)>"

y: "<unit(y)>"

u: "<unit(u)>"

Δt: "s"

u_c: "<unit(state)>"

g: "<unit(state)>"

graph:

hash: "<graph.hash>"

|V|: <int>

|E|: <int>

directed: false

weights: "nonnegative"

meta: { build_policy: "<id>", sources: ["<src>", "..."] }

operators:

L:

hash: "<L.hash>"

spectrum: { λ_min: <float>, λ_max: <float> }

A: { hash: "<A.hash>" }

H: { hash: "<H.hash>", rank: <int>, sensors_online: <float> } # fraction 0..1

kernel:

mode: "diff|wave|mix"

hyper: { τ: <float>, ω: <float>, α: <float>, order: <int> }

approx: { scheme: "chebyshev|pade|exact", order: <int>, ε_approx: <float> }

hash: "<K.hash>"

dynamics:

solver: "RK45|BDF|StrangSplit"

Δt: <float>

stiff_flag: <bool>

events: ["<event_name>", "..."]

guards: { norm_growth_γ: <float>, bound_b: <float> }

filter:

type: "KF|UKF|PF"

Q: { hash: "<Q.hash>" }

R: { hash: "<R.hash>" }

x0: { hash: "<x0.hash>" }

P0: { hash: "<P0.hash>" }

stats:

NIS_p95: <float>

ESS_over_Np: <float> # PF only

residual_p95: <float>

uncertainty:

method: { LPU: true, MC: true }

u_num: <float>

u_model: <float>

u_c: <float>

k: <float>

g: <float>

nu_eff: <float>

α: <float>

delta_form_unc_p95: <float>

contracts:

rules: { hash: "<rules.hash>", version: "<semver>" }

report: { hash: "<report.hash>", pass: <bool>, violations: <int> }

delta_form_stream_p95: <float>

coverage: <float>

runtime:

window: { type: "sliding|tumbling", Δt_win: "<dur>", Δt_slide: "<dur>" }

watermark: { wm: "<ISO8601>", policy: "<id>" }

latency_ms:

p50: <int>

p95: <int>

p99: <int>

cache: { hit_rate: <float>, policy: "ttl|lru", ttl: "<dur>" }

artifacts:

datasets:

- { name: "<id>", hash: "<sha256>", uri: "stg://..." }

models:

- { name: "<id>", hash: "<sha256>", uri: "stg://..." }

code:

- { repo: "<url>", commit: "<sha1>" }

panels:

- { name: "<id>", url: "<url-or-id>" }

provenance:

build_host: "<hostname>"

container: { image: "<name:tag>", digest: "<sha256>" }

os: "<id>", arch: "<id>"

signature:

key_id: "<kid>"

alg: "ed25519|ecdsa-p256"

value: "<base64>"

Example (Urban Traffic — Diffusion Kernel + UKF, Runtime Publication)

manifest_version: "stg.v1.0"

TraceID: "7a5ac1d0-9e6a-4a1e-8fe3-41a9f7c0e2c4"

ts_publish: "2025-09-02T10:15:03Z"

tau_mono: "807453.551"

producer: { org: "EFT-Labs", service: "stg-traffic", version: "1.6.2" }

RefCond:

timezone: "Europe/Amsterdam"

units_policy: "SI"

sensors: ["cam-A12", "loop-R7", "gps-bus-021"]

notes: "Morning peak; rain"

units:

x: "veh/s"

y: "veh/s"

u: "veh/s"

Δt: "s"

u_c: "veh/s"

g: "veh/s"

graph:

hash: "graph:5d7d1c3e"

|V|: 1120

|E|: 1437

directed: true

weights: "nonnegative"

meta: { build_policy: "traffic-v4", sources: ["osm-2025Q3","city-iot-1.2"] }

operators:

L:

hash: "L:9c2a77ab"

spectrum: { λ_min: 0.0, λ_max: 3.84 }

A: { hash: "A:11f5a0cd" }

H: { hash: "H:07bb2913", rank: 468, sensors_online: 0.93 }

kernel:

mode: "diff"

hyper: { τ: 120.0, ω: 0.0, α: 0.85, order: 0 }

approx: { scheme: "chebyshev", order: 15, ε_approx: 3.1e-3 }

hash: "K:6f0dbe21"

dynamics:

solver: "RK45"

Δt: 5.0

stiff_flag: false

events: ["gate-closure-R7@2025-09-02T07:58Z"]

guards: { norm_growth_γ: 1.02, bound_b: 3.0 }

filter:

type: "UKF"

Q: { hash: "Q:4a77f2b1" }

R: { hash: "R:0b61e0e9" }

x0: { hash: "x0:2d8f9c77" }

P0: { hash: "P0:3c1185d0" }

stats:

NIS_p95: 0.91

ESS_over_Np: 0.0

residual_p95: 7.4

uncertainty:

method: { LPU: true, MC: true }

u_num: 0.9

u_model: 3.2

u_c: 4.0

k: 2.0

g: 8.0

nu_eff: 46.0

α: 0.05

delta_form_unc_p95: 0.7

contracts:

rules: { hash: "rules:stg-traffic-2025.08", version: "2.3.0" }

report: { hash: "rep:0f77a9aa", pass: true, violations: 0 }

delta_form_stream_p95: 0.6

coverage: 0.96

runtime:

window: { type: "sliding", Δt_win: "30m", Δt_slide: "5m" }

watermark: { wm: "2025-09-02T10:14:50Z", policy: "wm-lag-20s" }

latency_ms:

p50: 420

p95: 980

p99: 1710

cache: { hit_rate: 0.87, policy: "ttl", ttl: "10m" }

artifacts:

datasets:

- { name: "iot-traffic-2025-09-02T09Z", hash: "ds:8f0c7d21", uri: "stg://city/ds/8f0c7d21" }

models:

- { name: "diff-kernel-ukf-1.6.2", hash: "mdl:5bcf112e", uri: "stg://models/5bcf112e" }

code:

- { repo: "https://git.example/eft/stg-traffic", commit: "d1e9c4a" }

panels:

- { name: "traffic-ops", url: "panel://ops/traffic" }

provenance:

build_host: "runner-17"

container: { image: "ghcr.io/eft/stg-traffic:1.6.2", digest: "sha256:ab92..." }

os: "linux", arch: "x86_64"

signature:

key_id: "k-ed25519-ops-01"

alg: "ed25519"

value: "Z5hQ7v3...=="

Summary
This appendix defines the required keys, generation workflow, and compliance assertions for stg manifests, and provides a ready-to-run template and example.
Deliverables include manifest.stg.runtime / manifest.stg.case, contracts.report, and manifest.hash/signature for comparison, replay, and dashboard mounting.