Contextual Gain-Scheduling Table¶
Decision artifact — instantiates Adaptive Gain Retuning
A lookup table that pairs each named operating regime with a pre-chosen gain, so the system switches sensitivity by recognizing its context rather than by watching a signal.
A Contextual Gain-Scheduling Table carves the operating world into a handful of named regimes — modes, load bands, risk classes, times of day — and stores the right gain for each one in advance. At run time the system identifies which cell of the map it is in and reads off the corresponding gain; there is no online estimation and no feedback chasing. Its defining trait is that it is open-loop and precomputed: the intelligence lives in the partition and the values chosen offline by someone who understood each regime, not in a loop reacting to the moment. It is the archetype's answer for when the contexts are known and few, even if the transitions between them are abrupt.
Example¶
A fly-by-wire flight control system faces a control surface that behaves completely differently at low-speed approach than at high-altitude cruise: the same stick deflection that is gentle at 250 knots would be violent at 500. Rather than let a loop hunt for the right responsiveness in flight, engineers build a gain-scheduling table indexed by airspeed and altitude bands. Each cell holds a control gain validated for that flight condition, and the system interpolates between neighbouring cells as the aircraft moves through the envelope.
On an approach, the flight computer reads its current airspeed-altitude cell and applies the softer gain scheduled there; as the aircraft accelerates and climbs, it slides to adjacent cells and the effective sensitivity firms up — all from the table, with no in-flight adaptation. The behaviour is repeatable and certifiable precisely because every value was chosen and checked before anyone flew, and the table can be read and reviewed on the ground.
How it works¶
Its logic is selection, not adaptation:
- Partition the operating space. Divide the world into named regimes along the variables that actually change the right gain — mode, load, risk tier, regime — coarse enough to be legible.
- Assign a gain per cell. For each regime, fix the gain offline using analysis, prior data, or expert judgement, and record it in the cell.
- Recognize, then look up. At run time, classify the current context into a cell and apply its gain; interpolate at boundaries so the switch between neighbouring cells isn't a jolt.
The table itself is the mechanism — a durable artifact you can print, review, and version, not a running estimator.
Tuning parameters¶
- Partition granularity — how many regimes the space is cut into. Finer cells fit each condition better but multiply the values to choose and validate, and thin the evidence behind each.
- Indexing variables — which observables define "which regime am I in." The whole scheme fails if the chosen index doesn't actually track what changes the right gain.
- Interpolation vs. hard switch — whether the boundary between cells is blended or stepped. Blending avoids jolts; hard switches are simpler to certify and reason about.
- Default / off-map cell — the gain used when the current context matches no defined regime — the safety net for the situation nobody tabulated.
When it helps, and when it misleads¶
Its strength is legibility and repeatability. When the regimes are known and the right gain in each is understood, a schedule gives behaviour that is fixed, auditable, and identical every time the same context recurs — which is why safety-critical and certified systems favour it over an adaptive loop.[n1] Nothing drifts, because nothing adapts.
Its weakness is the flip side: the table only knows the regimes its authors anticipated. A genuinely novel condition falls into a cell that fits it poorly, or off the map entirely, and the schedule applies a confidently wrong gain with no mechanism to notice. Schedules also go stale — the world the cells were tuned for moves on while the values sit frozen — and a mis-chosen index variable can put the system in the wrong cell without any error being visible. The discipline is to define an explicit off-map default, and to treat the table as something to be re-validated on a cadence rather than a set-and-forget artifact; where conditions are continuous and unpredictable, a closed loop is the better tool.
How it implements the components¶
This artifact fills the context-and-mapping components of the archetype — the partition subset, not the sensing or feedback subset:
context_partition_map— its defining structure: the division of the operating world into named, recognizable regimes.local_gain_map— the per-regime gain values the table holds and hands back on lookup.gain_parameter— the concrete gain each cell supplies to the fast pathway.
It contains no online sensing or feedback — it needs a monitor to tell it which regime it is in (Saturation Occupancy Dashboard, Adaptive Normalization Layer) and never self-corrects the way Automatic Gain Control Loop does; the hard limits on any tabled value belong to Gain Floor/Ceiling Rule.
Related¶
- Instantiates: Adaptive Gain Retuning — the precomputed, context-indexed alternative to an adaptive loop.
- Sibling mechanisms: Automatic Gain Control Loop · Adaptive Normalization Layer · Contrast Adaptation Protocol · Exposure or Alarm Sensitivity Adjuster · Fixed-Gain Degraded Mode · Gain Floor/Ceiling Rule · Saturation Occupancy Dashboard · Gain-Change Review Log
Editorial Notes¶
Form Classification¶
Form family: Control, Automation & Runtime
Rationale: A lookup table that pairs each named operating regime with a pre-chosen gain, so the system switches sensitivity by recognizing its context rather than by watching a signal, making its operative form a live operational control that automatically routes, enforces, adapts, or responds during execution.
Independent corroboration: The frozen evidence defines Contextual Gain-Scheduling Table as 'A lookup table that pairs each named operating regime with a pre-chosen gain, so the system switches sensitivity by recognizing its context rather than by watching a signal', so its operative form is Control, Automation & Runtime.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Systems Thinking & Cybernetics
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Nonlinear control theory cohered gain scheduling: controllers are designed at selected operating points and their gains switched or interpolated by a scheduling variable.
Related originating lineages:
- Engineering & Design — Aerospace and process-control practice institutionalized lookup tables for operating-envelope-specific control settings.
Review resolution: Gain scheduling is a named control-systems method, so systems/cybernetics is the precise primary and aerospace and process engineering are implementation lineages; the lookup-table form is established.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
A schedule and a closed loop are complements, not rivals: many real systems use a table to set the coarse gain for the current regime and let a bounded loop trim around it. Read that way, this artifact supplies the operating point and Automatic Gain Control Loop handles the residual — which keeps the schedule certifiable while still absorbing within-regime variation.
[n1] Gain scheduling — a standard technique from nonlinear control: rather than derive one controller that works everywhere, engineers design gains for a set of operating points and switch or interpolate between them by a measured scheduling variable. It is the direct control-theory ancestor of this artifact, and its known caveat — validity only near the tabulated points — is exactly this mechanism's failure mode. ↩