Tuning (bristlecone) | bristlecone

bristlecone

Tuning Module

Namespace: Bristlecone.Optimisation
Assembly: Bristlecone.dll
Parent Module: SimulatedAnnealing

Types

Type	Description
ParamTuningState<'s>
TuningSettings

Functions and values

Function or value	Description
`perturb initialScale settings` Full Usage: `perturb initialScale settings` Parameters: initialScale : `float<MeasureOne>` settings : `TuningSettings` Returns: `Optimiser`	An exploration method that perturbs around a point in optimisation-space. Used for profile likelihood in the Confidence module. initialScale : `float<MeasureOne>` settings : `TuningSettings` Returns: `Optimiser`
`tuneStepSizes writeOut random domain draw machine f initialScale settings (arg9, arg9)` Full Usage: `tuneStepSizes writeOut random domain draw machine f initialScale settings (arg9, arg9)` Parameters: writeOut : `LogEvent -> unit` random : `Random` domain : `Domain` draw : `float<MeasureProduct<optim-space, MeasureOne>> -> float -> unit -> float<MeasureProduct<optim-space, MeasureOne>>` machine : `float -> float<MeasureProduct<-logL, MeasureOne>> -> float` f : `Objective` initialScale : `float<MeasureProduct<optim-space, MeasureOne>>[]` settings : `TuningSettings` arg8 : `float<MeasureProduct<-logL, MeasureOne>>` arg9 : `Point` Returns: `float<MeasureProduct<optim-space, MeasureOne>>[] * float<MeasureProduct<-logL, MeasureOne>> * Point * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`	Diminishing, parameter-wise adaptation of individual parameter step sizes based on acceptance rate. Conforms to traditional simulated annealing tuning procedure but introduces an early-stop mechanism for if scales settle earlier than the configured maximum iteration. Returns tuned scales and the final point. writeOut : `LogEvent -> unit` random : `Random` domain : `Domain` draw : `float<MeasureProduct<optim-space, MeasureOne>> -> float -> unit -> float<MeasureProduct<optim-space, MeasureOne>>` machine : `float -> float<MeasureProduct<-logL, MeasureOne>> -> float` f : `Objective` initialScale : `float<MeasureProduct<optim-space, MeasureOne>>[]` settings : `TuningSettings` arg8 : `float<MeasureProduct<-logL, MeasureOne>>` arg9 : `Point` Returns: `float<MeasureProduct<optim-space, MeasureOne>>[] * float<MeasureProduct<-logL, MeasureOne>> * Point * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`