# SIM2C/SIM3C Practical Pass

Status:

```text
SIM_PRACTICAL_POLICIES_READY_FOR_REAL_SIMULATION_PILOT__ADAPTIVE_UPDATE_TEMPORAL_JET_AND_OBSERVABLE_CAPSULE__NO_PHYSICAL_CLAIM
```

SIM2C/SIM3C is the practical scope upgrade after the SIM2/SIM3 boundary
results. It does not erase those earlier failures:

- SIM2/SIM2B remain negative results for bare checkpoint placement.
- SIM3/SIM3B remain boundary results for mask-only compression.

The new result says something narrower and more useful:

```text
For structured toy simulation worlds, practical checkpoint records and
observable-capsule compression can be made useful under frozen engineering
rules.
```

No QPU was run. No physical claim is added.

## SIM2C: Temporal-Jet Checkpoint Records

SIM2 failed because event-only checkpoint placement loses to uniform-time
coverage when reconstruction is simple linear interpolation.

SIM2C changes the engineering object, not the physics claim:

- a checkpoint stores the state and a local time derivative;
- reconstruction uses temporal Hermite interpolation;
- checkpoint times are placed by a conservative information-arc clock.

Result:

| world | counted | target error | best control | margin |
|---|---:|---:|---|---:|
| moving_dual | yes | 0.001297685 | uniform_temporal_jet | 0.000009900 |
| split_front | yes | 0.000192301 | uniform_temporal_jet | 0.000009539 |
| record_pulse | yes | 0.000352026 | uniform_temporal_jet | 0.000099263 |
| noise_drive | no | 0.005976951 | uniform_temporal_jet | -0.000266159 |

Interpretation:

- The large practical gain is the temporal-jet record format versus bare
  linear checkpoints.
- Information-arc placement is a modest but consistent improvement over
  uniform temporal-jet placement in counted structured worlds.
- The noise-drive stress row does not pass and is not counted.

Allowed use:

```text
Use temporal-jet checkpoint records as a candidate engineering policy in a
real/semi-real simulator pilot, with uniform temporal-jet and linear
checkpoint baselines retained as controls.
```

## SIM3C: Observable-Capsule Compression

SIM3/SIM3B showed that mask-only compression was close but not broad enough.

SIM3C changes the compression format:

- keep a hybrid source/gradient/delta detail mask;
- store tiny observable-capsule metadata per compressed frame;
- apply mean/peak calibration during reconstruction.

Result:

| world | counted | target error | best control | margin |
|---|---:|---:|---|---:|
| moving_dual | yes | 0.024078341 | gradient_capsule | 0.001566794 |
| split_front | yes | 0.029164813 | source_capsule | 0.004004496 |
| record_pulse | yes | 0.016746139 | delta_capsule | 0.001955203 |
| noise_drive | no | 0.009167417 | delta_capsule | -0.000138914 |

Interpretation:

- Observable capsules make the compression policy useful across all counted
  structured worlds.
- The target beats source, gradient, delta, uniform, random and shuffled
  capsule controls in counted worlds.
- The noise-drive stress row does not pass and is not counted.

Allowed use:

```text
Use observable-capsule compression as a candidate engineering policy in a
real/semi-real simulator pilot, with source/gradient/delta/uniform/random and
shuffled controls retained.
```

## Practical Readiness

After SIM2C/SIM3C, the practical SIM pilot scope is:

1. adaptive update budget from SIM1/SIM4;
2. temporal-jet checkpoint records from SIM2C;
3. observable-capsule compression from SIM3C.

The first real/semi-real pilot must still be an engineering benchmark:

- fixed simulator and observables;
- frozen policy parameters;
- explicit compute/error tradeoff;
- stress/noise rows reported separately;
- no claim of better physical truth.

## Blocked Claims

SIM2C/SIM3C does not allow:

- a claim that ITHKOR-SIM is a universal simulator;
- a claim that it improves physical correctness in real systems;
- a claim that the universe is a simulation;
- spacetime, gravity, physical time or quantum-gravity claims.