NERVA · v11.0
Starpoint LLC
miglanthony@gmail.com

Data field around the state space.

How API inputs become a Bloch-vector magnitude — and why the answer is magnitude, not amplitude.
DATA ARCHITECTURE
SHEET 01 · OF 01
REV 11.0.1 · JUNE 2026
§00 · PRIMER

NERVA is a decision-integrity layer for autonomous systems. It sits between a proposed automated action — a trade, a strike, a lane-change, a media buy — and its execution. It accepts five normalized inputs (urgency, strategy, risk, support, stability), maps them onto a Bloch-vector geometry drawn from canonical quantum-information mathematics, and returns one of five verdicts COMMIT · HOLD · WAIT · CONSULT · TOXIC with full mathematical provenance for every call.

NERVA is not a predictor. It is a gate. The point is that the math is verifiable against published textbooks (Nielsen & Chuang 2010, Shannon 1948, Baumgratz · Cramer · Plenio 2014) rather than against a proprietary black box no regulator can audit. v11 adds a second axis: each input carries an optional confidence score, so two decisions with identical point estimates produce different verdicts depending on whether the data was instrumented or estimated. This sheet shows how API data becomes that confidence — and how confidence becomes geometry.

§01 · VERTICAL A
Ad tech
media buyside · flight integrity
  • DSP / SSP bid stream streaming
  • Verified imp + viewability audited
  • Brand-safety classifier model
  • Flight-pacing telemetry streaming
§01 · VERTICAL B
Medical
surgical pre-authorization
  • EHR / HL7 FHIR feed audited
  • Imaging confidence score model
  • Lab result timestamps instrumented
  • Protocol compliance log audited
§01 · VERTICAL C
Defense / ROE
autonomous platform release
  • Sensor fusion feed instrumented
  • IFF transponder data instrumented
  • Geofence / zone API structured
  • Comms latency telemetry streaming
§01 · VERTICAL D
Finance / HFT
execution integrity
  • L2 order-book feed streaming
  • Risk engine output model
  • Position / margin API audited
  • Volatility surface feed instrumented
§02
Data provenance layer
freshness decay · credibility · validation coverage · source class
→ normalized in [0, 1]
§03
scoreFromSource() → confidence vector
one confidence score per input · auditable per call
cE, cS, cR, cSp, cSt ∈ [0, 1]
EUrgency
intent axisw = 0.15
SStrategy
intent axisw = 0.15
RRisk
risk axisw = 0.15
SpSupport
integrity axisw = 0.275
StStability
integrity axisw = 0.275
↑ Integrity-axis inputs carry highest weight. They govern the One-Way Brake.
§04
Aggregate C = Σ wi · ci
shrinkage factor · range [0, 1] · 1.0 reproduces v10 behavior
rmixed = C · rpure
§05 · STATE SPACE

Bloch-sphere cross-section.

The data field is the envelope. Better API data → higher C → smaller envelope, longer |r|.
COMMIT HOLD |r| = 1 · PURE STATE surface of Bloch sphere · maximal information HIGH-C ENVELOPE data quality ≈ 90% tight cloud · |r| near surface LOW-C ENVELOPE data quality ≈ 30% wide cloud · |r| shrunk toward origin τ = 0.52 · ZONE BOUNDARY commit if |r| ≥ τ τ widens with entropy + risk |r| = 0 maximally mixed · N&C: no information unphysical region · |r| > 1 cannot occur by construction |r| GROWS → better data, longer vector
§06 · PHYSICS NOTE

Magnitude, not amplitude.

Q · what does the data field add to?
It's a magnitude.
|r| The Bloch vector |r| is a scalar length from 0 to 1 — a geometric magnitude, not a wave-function amplitude. Amplitude would imply a complex coefficient on a probability (the quantum-mechanics cousin); we're operating on the corresponding density matrix. What grows as data quality improves is the magnitude of the Bloch vector: better data → higher C → less shrinkage → |r| moves outward toward the unit sphere surface.
Q · what is "the data field around the state space"?
An isotropic envelope, sized by (1 − C).
The cloud surrounding the point in state space is the uncertainty envelope. In v11 it is isotropic — the envelope radius is set uniformly by the scalar aggregate C = Σ wi·ci, so shrinkage is the same in every direction: rmixed = C · rpure. The per-axis confidences (cE, cS, cR, cSp, cSt) are preserved in the audit trail for every call and visible to the operator, but they aggregate to a single shrinkage factor rather than stretching the envelope per-axis. More and better API data raises C, collapses the envelope toward a point, and slides that point outward along its radial direction toward the unit sphere. Per-axis (anisotropic) shrinkage — where each confidence independently deforms the envelope along its own axis — is a planned v12 extension.
Q · what do |r| = 1 and |r| = 0 mean for a decision?
Decisions live in the interior.
In Nielsen & Chuang Ch. 11 terms: a pure state (|r| = 1, on the surface of the Bloch sphere) represents complete information about the state — not a predetermined outcome. Measurement in a non-eigenbasis is still probabilistic; what is maximal is the information, not the certainty of any particular result. A maximally mixed state (|r| = 0, at the origin) represents zero information — the state is fully unknown. NERVA lives in between. Every real decision starts somewhere in the interior. Pilot data moves it outward toward the surface. The audit trail records exactly how much information was available when the call was made.

The Nielsen & Chuang contract. Ch. 2 (pure states) · Ch. 11 (mixed states, von Neumann entropy)

§07 · A
Pure state · |r| = 1
ρ = ½ (I + r · σ⃗)
Full information about the state. On the surface of the Bloch sphere. |r| = 1 is maximal information — not a predetermined outcome. Measurement in a non-eigenbasis remains probabilistic; what is complete is the knowledge of the state, not the result of any particular measurement.
Nielsen & Chuang Ch. 2 · pure states are |ψ⟩⟨ψ|
§07 · B
Mixed state · |r| < 1
ρ = C · ρpure + (1 − C) · I/2
Partial information — where decisions actually live. In the interior. The shrinkage factor C is set by the data field: the less you know, the closer to the origin. Every NERVA verdict carries the receipt of how much information was available when the call was made.
Nielsen & Chuang Ch. 11 · convex combination with maximally mixed state I/2