EndConditions (bristlecone)

EndConditions Module

Namespace: Bristlecone.Optimisation
Assembly: Bristlecone.dll

Composable end conditions to specify when optimisation
 routines should end.

Nested modules

Modules	Description
Ensemble
Profiles

Functions and values

Function or value	Description
`atIteration iteration arg2 currentIteration` Full Usage: `atIteration iteration arg2 currentIteration` Parameters: iteration : `int<MeasureProduct<iteration, MeasureOne>>` arg1 : `Solution list` currentIteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	End on or after a minimum number of iterations. iteration : `int<MeasureProduct<iteration, MeasureOne>>` arg1 : `Solution list` currentIteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`combineAll conditions results iter` Full Usage: `combineAll conditions results iter` Parameters: conditions : `EndCondition list` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Stop when all conditions have non-continue reasons. conditions : `EndCondition list` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`combineAny conditions results iter` Full Usage: `combineAny conditions results iter` Parameters: conditions : `EndCondition list` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Stop when any condition has a non-continue reason. conditions : `EndCondition list` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`stationarySquaredJumpDistance log` Full Usage: `stationarySquaredJumpDistance log` Parameters: log : `LogEvent -> unit` Returns: `EndCondition`	True if there is no significant slope in mean squared jumping distances (MSJD), binned per 200 iterations and a regression of five bins. log : `LogEvent -> unit` Returns: `EndCondition`
`stationarySquaredJumpDistance' fixedBin pointsRequired slopeTol log results i` Full Usage: `stationarySquaredJumpDistance' fixedBin pointsRequired slopeTol log results i` Parameters: fixedBin : `int` pointsRequired : `int` slopeTol : `float` log : `LogEvent -> unit` results : `Solution list` i : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	An `EndCondition` that calculates that segregates the most recent n results into five bins, and runs a regression to detect a temporal variation in the mean squared jumping distance (MSJD). The significance of the slope coefficient of a linear regression is assessed to determine if the MSJD is increasing through time for every parameter sequentially: if all p-values are >0.1, then the `EndCondition` is true. fixedBin : `int` pointsRequired : `int` slopeTol : `float` log : `LogEvent -> unit` results : `Solution list` i : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenAcceptanceRateOutside min max interval intervalsRequired log results iteration` Full Usage: `whenAcceptanceRateOutside min max interval intervalsRequired log results iteration` Parameters: min : `float` max : `float` interval : `int<MeasureProduct<iteration, MeasureOne>>` intervalsRequired : `int<MeasureOne>` log : `LogEvent -> unit` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	min : `float` max : `float` interval : `int<MeasureProduct<iteration, MeasureOne>>` intervalsRequired : `int<MeasureOne>` log : `LogEvent -> unit` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenImprovementsMade count interval results iteration` Full Usage: `whenImprovementsMade count interval results iteration` Parameters: count : `int` interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Ends when the overall number of improvements made is greater than `count`. count : `int` interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenNoBestValueImprovement window results iteration` Full Usage: `whenNoBestValueImprovement window results iteration` Parameters: window : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Given a list of solutions, which are ordered most recent first, returns `true` if there are at least `window` recent results, and the change within the recent results is no more than `tolerance`. window : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenObjectiveFlat interval results iteration` Full Usage: `whenObjectiveFlat interval results iteration` Parameters: interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Ends if the mean change in the objective's value is less than a tolerance of 1e-6. interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenStableAcceptanceRate min max interval intervalsRequired log` Full Usage: `whenStableAcceptanceRate min max interval intervalsRequired log` Parameters: min : `float` max : `float` interval : `int<MeasureProduct<iteration, MeasureOne>>` intervalsRequired : `int<MeasureOne>` log : `LogEvent -> unit` Returns: `EndCondition`	Stops when acceptance rate is consistently within the defined range. Used to avoid stopping when not mixing. min : `float` max : `float` interval : `int<MeasureProduct<iteration, MeasureOne>>` intervalsRequired : `int<MeasureOne>` log : `LogEvent -> unit` Returns: `EndCondition`
`whenStationary threshold nSteps results iter` Full Usage: `whenStationary threshold nSteps results iter` Parameters: threshold : `float<MeasureProduct<MeasureProduct<optim-space, optim-space>, MeasureOne>>` nSteps : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	Stops when there is a lack of movement, as indicated by squared jump distance below threshold across the last n steps. threshold : `float<MeasureProduct<MeasureProduct<optim-space, optim-space>, MeasureOne>>` nSteps : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iter : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenStuck movementFloor interval results iteration` Full Usage: `whenStuck movementFloor interval results iteration` Parameters: movementFloor : `float<MeasureProduct<-logL, MeasureOne>>` interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	movementFloor : `float<MeasureProduct<-logL, MeasureOne>>` interval : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`
`whenVarianceStabilised window relTol results iteration` Full Usage: `whenVarianceStabilised window relTol results iteration` Parameters: window : `int<MeasureProduct<iteration, MeasureOne>>` - number of recent samples to compare relTol : `float<MeasureOne>` - relative change threshold (e.g. 0.05 = 5%) results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason` A stop condition or Continue.	Stop when the rolling variance of the objective stabilises. window : `int<MeasureProduct<iteration, MeasureOne>>` number of recent samples to compare relTol : `float<MeasureOne>` relative change threshold (e.g. 0.05 = 5%) results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason` A stop condition or Continue.
`whenWorstValuePlateaued window results iteration` Full Usage: `whenWorstValuePlateaued window results iteration` Parameters: window : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`	End when exploration is not detecting any new worse results. For use in heating stages of simulated annealing. window : `int<MeasureProduct<iteration, MeasureOne>>` results : `Solution list` iteration : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `OptimStopReason`