E02 conditional reset demands
============================

This is a bounded postprocessing study and ideal prescribed-motion calculation.
It reuses eight saved E01 fine traces, computes no new dynamic integrations and
does not simulate a repeating impact device. No physical trials have occurred.

The fixed plan is E02-PLAN.json. It remains byte-identical with its original
awaiting-review status; root execution approval is a separate receipt, not a
silent rewrite. PRE-RUN-DIAGNOSTIC-ADDENDUM.txt adds interpretive reset kinetic
energy and bidirectional-force diagnostics without changing matrix or thresholds.

Model and meaning
The inherited0.60s release-to-pickup observation contains flight, impact and
settling. It is not added to the first-contact time again. A50ms near-rest window
must pass both E01 every-step gates and a50us retained-sample reconstruction.
Boundary failures, moving captures and all target-present cases remain ineligible.
Near-rest is not exact rest, gripper engagement or physically reliable capture.

Only conditional missing-target cases get a separate hypothetical lift starting
at exact static buffer equilibrium x_eq=2mm+m*g/k. The actual terminal22-state
vector, energy/storage and its difference from equilibrium are preserved. No
projection is performed on an E01 state and no transition/capture phase is modeled.

A minimum-jerk quintic raises that assumed rest position to the original release
height at zero final velocity and acceleration. It includes gravity, inherited
smooth guide drag, buffer unloading and an optional pickup-only drag hypothesis.
Downward signed actuator force is m*xddot-m*g+F_buffer+F_guide. Negative/upward and
positive/downward demands are retained. Any force reversal requires an ideal
bidirectional positively attached actuator, not a tension-only cable claim.

Reset actuator work closes gravitational increase minus released buffer storage
plus guide/pickup dissipation. Positive and absorbed work stay separate. There is
no assumed battery/electrical efficiency, regeneration credit or holding power.
Fast lifts can exceed the20mJ gravity-drop kinetic-energy reference. A numerical
pass is not a rated actuator, strength, retention or safety pass.

The seven commanded cadences use lift duration60/BPM-0.60-0.02-0.02seconds.
The20ms engagement and20ms release allowances are explicit optimistic hypotheses.
Nonpositive durations fail only under this prescribed pickup policy; they do not
prove a fundamental maximum repetition frequency. The release pin's separate
>=34mm lateral withdrawal, grip forces and actuator disengagement are unmodeled.
Cadence is not specimen resonance. An attached/reset drive during free drop would
change the inherited moving mass, friction and energy model.

Reproduce with Python3.12and NumPy (the recorded runtime is in the summary):
  python -m unittest -v test_reset_cycle.py

Compact inputs are supplied so the reset calculations can run without duplicating
the parent traces. To re-extract them, obtain the unchanged E01 evidence identified
by the plan's exact public URLs/hashes and use a fresh output path:
  python extract_inputs.py --e01-root ../operating-models --output fresh-inputs.json

Run from the supplied or freshly extracted compact input, always into a new folder:
  python run_study.py --inputs inputs/E01-PICKUP-INPUTS.json --output fresh-results

Generate actual data figures into another fresh folder:
  python make_figures.py --results fresh-results --output fresh-figures
Matplotlib is required only for figures. Figure manifests bind plot source and
result hashes; the numerical engine itself needs only NumPy.

The extractor requires parent impact_model.py, E01-PLAN.json, original summary and
eight original fine JSON/NPZ pairs. It verifies all frozen parent hashes before and
after extraction. It reads sources only after those hashes match. The runner freezes
its source and compact-input hashes before calculations and checks them afterward.
All112 rows are retained, including separate pickup and cadence rejection reasons.

Quadrature is64/128-point Gauss-Legendre on intervals split at buffer unloading and
signed-force roots. Peak force/power uses8193/16385time nodes and is not a certified
continuous maximum. Peak quintic speed/acceleration/kinetic energy is analytical.
Failure thresholds are in the frozen plan; no result-dependent retuning is allowed.

Current scope does not select a powered reset actuator or its electronics. An
existing acoustic amplifier or wired speaker chain does not constitute that drive.
No purchases, hardware commands, publication or CAD changes occur in these scripts.
