Open constants
Values deliberately left unfilled, and why
Snapshot of this document in the GEMs repository. The repository itself is the current version. Bản chụp tài liệu này trong repo GEMs. Bản đang chạy nằm ở chính repo.
Every value this specification cannot yet derive is left marked rather than filled with a guess. A specification that invents its own constants trades honesty for the appearance of completeness.
Two kinds of blank, and they are not the same:
⟦CTRL⟧— not ours to fill. A decision belonging to whoever operates the body. Filling it here would be the body overruling its operator.⟦IMPL⟧— not yet derivable. A value that depends on choices not made, or on measurements not taken.
⟦CTRL⟧ — operator decisions
| Value | Where | Bounded by |
|---|---|---|
| Operating endurance | 03.8 | The ceiling t_max in 02.3 |
| Protection level and coverage | 02.4 | The mass loop; each kilogram costs γ kg |
| Supercapacitor fitted or not | 03.1 | Determines whether peak power is actuator-limited or source-limited |
| Sleep depth and schedule | 03.3 | Depth against wake latency |
| Charging method and timing | 03.5 | — |
| Area allocated to each shell layer | 04.5 | The same m_ext budget as armour |
| Compute split, body against external | 05.6 | Two hard ends; the middle is adjustable |
| Use made of spatial RF sensing | 01, 05.3 | Capability retained in full at the hardware level |
| Log signing period, synchronisation frequency, flag tolerances | 06.7 | — |
| Acceptable state loss when a body is lost | 01 | — |
⟦IMPL⟧ — not yet derivable
| Value | Where | Why not filled |
|---|---|---|
f_str, f_act — structural and actuator mass fractions |
02.2 | Depend on material and gearing choices not yet made. Ranges are narrow enough that no conclusion changes sign |
| Fixed non-scaling mass (compute, sensors, hands, skin, harness) | 02.5 | Depends on component selection |
| Areal density of the sense-and-heal and variable-stiffness layers | 04.5 | No reliable published figure found. The constraint is firm even though the number is not |
| Fire layer temperature and duration | 04.4 | Must anchor to an existing firefighting-garment standard, not be derived here |
| RAM retention and root-of-trust clock power | 03.6 | Depends on the parts chosen. Immaterial against self-discharge |
| Full-CSI spatial sensing power | 03.6 | The one floor term that matters; depends on implementation |
| Merkle batch period for log signing | 06.4 | Trades trace granularity against signing load |
| Durable-tier cycle life, discharge rate and pack abuse test | 03.1 | Set when the cell is chosen (plan D-5). The density, e, is the one figure the mass loop needs, and it is established |
| What must survive the pack's removal in service | 03.1 | The always-on rail sits in the power assembly as the bill of materials stands; depends on the power architecture (plan E-2, E-4) |
| Heartbeat threshold that triggers the OBC | 07.5 | Depends on the link and on how long the Real-time Control Layer can hold without new setpoints |
| Exact self-discharge rate | 03.6 | Depends on the chemistry chosen. The order of magnitude and the dominant term are established, which is what the sleep ceiling needs |
What is deliberately not here
Some things are absent because they belong to no one in this repository:
| Absent | Whose |
|---|---|
| Any specification of the controller | Not this repository |
| The standard judging a body's conduct | A third party: the deploying or certifying party |
| Conformance test procedures | protocol/ — v0.1 draft |
Reading this chapter
A blank here is not an oversight and not a placeholder waiting for a plausible number. Where the specification could establish the shape of a constraint without its exact value, it did so and said which is which — the sleep ceiling in 03.6 is the clearest example: the precise self-discharge figure is open, but knowing which term dominates was enough to close the question that mattered.