Core (bristlecone) | bristlecone

bristlecone

Core Module

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

Types

Type	Description
TrendResult

Functions and values

Function or value	Description
`acceptanceFromCounts accepts stepsPerParam` Full Usage: `acceptanceFromCounts accepts stepsPerParam` Parameters: accepts : `int[]` stepsPerParam : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `float[]`	accepts : `int[]` stepsPerParam : `int<MeasureProduct<iteration, MeasureOne>>` Returns: `float[]`
`adaptiveStep sigmas acceptanceRates batchNumber` Full Usage: `adaptiveStep sigmas acceptanceRates batchNumber` Parameters: sigmas : `LogSigma[]` acceptanceRates : `float[]` batchNumber : `int<MeasureProduct<batch, MeasureOne>>` Returns: `LogSigma[]`	Adaptive tuning step. Tune variance of a parameter based on its acceptance rate. The magnitude of tuning reduces as more batches have been run. sigmas : `LogSigma[]` acceptanceRates : `float[]` batchNumber : `int<MeasureProduct<batch, MeasureOne>>` Returns: `LogSigma[]`
`mhStep1D random domain f j lsj (theta, l)` Full Usage: `mhStep1D random domain f j lsj (theta, l)` Parameters: random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` j : `int` lsj : `LogSigma` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * float<MeasureProduct<-logL, MeasureOne>> * bool`	One MH step updating only coordinate j, returning accepted theta, -logL, and accepted flag. random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` j : `int` lsj : `LogSigma` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * float<MeasureProduct<-logL, MeasureOne>> * bool`
`propose theta j lsj domain random` Full Usage: `propose theta j lsj domain random` Parameters: theta : `float<MeasureProduct<optim-space, MeasureOne>>[]` j : `int` lsj : `LogSigma` domain : `Domain` random : `Random` Returns: `float<MeasureProduct<optim-space, MeasureOne>>[]`	Propose a jump drawn from a Normal(0, exp(lsj)²) distribution, while leaving all but one parameter value fixed. theta : `float<MeasureProduct<optim-space, MeasureOne>>[]` j : `int` lsj : `LogSigma` domain : `Domain` random : `Random` Returns: `float<MeasureProduct<optim-space, MeasureOne>>[]`
`runBatchMWG random domain f batchLength sigmas (theta, l)` Full Usage: `runBatchMWG random domain f batchLength sigmas (theta, l)` Parameters: random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` batchLength : `int<'u>` sigmas : `LogSigma[]` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>> * int[] * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`	Run a batch of 'n' sweeps. random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` batchLength : `int<'u>` sigmas : `LogSigma[]` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>> * int[] * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`
`sweepOnce random domain f sigmas (theta, l)` Full Usage: `sweepOnce random domain f sigmas (theta, l)` Parameters: random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` sigmas : `LogSigma[]` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>> * int[] * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`	One full MWG sweep across all coordinates, cumulatively updating theta. random : `Random` domain : `Domain` f : `TypedTensor<Vector, MeasureProduct<optim-space, MeasureOne>> -> TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` sigmas : `LogSigma[]` theta : `Point` l : `TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>>` Returns: `Point * TypedTensor<Scalar, MeasureProduct<-logL, MeasureOne>> * int[] * (float<MeasureProduct<-logL, MeasureOne>> * Point) list`
`trendCheckStep fullResults batchLength paramCount` Full Usage: `trendCheckStep fullResults batchLength paramCount` Parameters: fullResults : `Solution list` batchLength : `int<MeasureProduct<iteration, MeasureOne>>` paramCount : `int` Returns: `(int * TrendResult) list`	Trend check step (non‑adaptive) on the last five batches. fullResults : `Solution list` batchLength : `int<MeasureProduct<iteration, MeasureOne>>` paramCount : `int` Returns: `(int * TrendResult) list`