ConstantExtrapolationPredictor

Source: models/common/ConstantExtrapolationPredictor.py

Initial guess for an implicit update: each unknown takes its ~1 value.

For an ImplicitUpdate with unknowns {u_i}, this predictor reads u_i~1 and outputs \(u_i\). Used by ImplicitUpdate to seed Newton. Matches the C++ ConstantExtrapolationPredictor HIT signature.

Other options

unknowns_SR2 — list

The unknowns to extrapolate of type SR2

unknowns_Scalar — list

The unknowns to extrapolate of type Scalar

unknowns_MRP — list

The unknowns to extrapolate of type MRP

unknowns_R2 — list

The unknowns to extrapolate of type R2

cold — list

Optional unknown:variable pairs naming a COLD value per unknown. On the step where there is no history – the first – that unknown is seeded from the named variable instead of from u~1, which lets a physics-based predictor supply the cold start while this model keeps every later step. Name only the unknowns you want seeded; the rest fall back to u~1 exactly as before. The type comes from whichever unknowns_* list the name appears in, so there is no second list to keep in step.

time — str · default t

Time. Only consumed when a cold mapping is given – the no-history test needs the time history.