Methods

How the experiments are measured

Each experiment has its own frozen contract, measurement definitions and limits.

CP-001 measurement method ยท PF-001 numerical method and reproduction

CP-001 revision 2

The entries below come from its frozen contract and identity-bound plan.

Time

Unit
ns
Epoch
scenario-start
Tick semantics
Integer ticks denote left-closed scheduler release instants from scenario start.
Sample semantics
Sensor and plant values at tick k are sampled before commands released at tick k are applied.
Jitter bound
0

Signals

Declared contract signals
SignalKindUnitFrame
q-bi-wstate1body-frd
q-bi-xstate1body-frd
q-bi-ystate1body-frd
q-bi-zstate1body-frd
omega-b-xstaterad/sbody-frd
omega-b-ystaterad/sbody-frd
omega-b-zstaterad/sbody-frd
firmware-wheel-torque-commandcommandN*mbody-frd
plant-wheel-torque-inputcommandN*mbody-frd
gyro-ratesensorrad/sbody-frd
physical-sensor-validsensor1dimensionless
firmware-sensor-valid-sampledsensor1dimensionless
controller-modemode1dimensionless
watchdog-expiredfault1dimensionless
wheel-saturatedfault1dimensionless
attitude-errorpropertyradbody-frd

Mappings

Declared cross-domain mappings
MappingSourceTargetTransformationInformation loss
MAP-TORQUEfirmware-wheel-torque-commandplant-wheel-torque-inputcomponentwise identity in N*m at the same nanosecond ticknone declared
MAP-SENSOR-VALIDphysical-sensor-validfirmware-sensor-valid-sampledsample at each 10000000 ns sensor releasePhysical validity changes between sample instants are not represented.

Assumptions

  • ASM-001: The rigid body has diagonal constant inertia within the declared uncertainty envelope.
  • ASM-002: External torque is piecewise constant per plant step and remains within its declared component bound.
  • ASM-003: The software clock and physical trace share scenario-start epoch and exact nanosecond ticks.
  • ASM-004: Quatopsy and the independent baseline consume active wxyz body-to-ECI quaternions.

Guarantees

  • OBL-DEADLINE: Every required controller job completes before its next declared release.
  • OBL-WATCHDOG: Three consecutive invalid sensor samples enter safe mode no later than the watchdog timeout.
  • OBL-TORQUE: Every commanded reaction-wheel torque remains within 0.002 N*m per body axis.
  • OBL-MOMENTUM: Reaction-wheel momentum magnitude remains within 0.02 N*m*s per wheel.
  • OBL-ATTITUDE: Attitude error remains at or below 0.05 rad over the registered horizon.
  • OBL-RATE: Body-rate magnitude remains at or below 0.1 rad/s over the registered horizon.

Execution

Stopping rule
Qualify historical inputs, then execute each declared revision2 scenario once in order. Retain refusal and incomplete resource-bound evidence and continue to the next declared scenario. Unexpected arithmetic, identity, time, or output-bound failure aborts without emitting a canonical result. No retries or hidden repeat trials.
Primary executions per scenario
1
Retries
0
Seed use
scenario-label-only; no random generator or stochastic uncertainty sampling
Step
10000000 ns
Horizon
1000 steps
Scope
Single-axis X rotation from rest and identity attitude; zero gyro bias, noise, and wheel scale error. Integer midpoint rigid-body plant with unclipped wheel momentum and momentum-limit safe mode. Independent floating-point RK4 quaternion replay uses the same applied torque, inertia, initial state, epoch, and step. Inertia sensitivity is a separate open-loop replay, not implementation disagreement. Watchdog latches safe mode after the second consecutive invalid sample and records expiration on the third. All physical maxima include the terminal state.