BACCO (version 1.0-50)

genie: Genie datasets for approximator package

Description

Genie datasets that illustrate the package.

Usage

data(genie)
  D1.genie
  hpa.genie
  z.genie
  subsets.genie
  basis.genie(x)
  hpa.fun.genie(x)
  hpa.genie.start
  hpa.genie.optimal

Arguments

x

A 4-element vector (for basis.genie()); a 19-element vector (for hpa.fun.genie())

format

The genie example is a case with three levels. The D1.genie matrix is 36 rows of code run points, corresponding to the observations of the level 1 code. It has four columns, one per parameter.

hpa.genie is a hyperparameter object.

subsets.genie is a list of three elements. Element $i$ corresponds to the rows of D1.genie at which level $i$ has been observed. z.genie is a three element list. Each element is a vector; element $i$ corresponds to observations of level $i$. The lengths will match those of subsets.genie. Function basis.genie() is a suitable basis function. Function hpa.fun.genie() creates a hyperparameter object in a form suitable for passing to the other functions in the library.

References

M. C. Kennedy and A. O'Hagan 2000. Predicting the output from a complex computer code when fast approximations are available Biometrika, 87(1): pp1-13

Examples

data(genie)
z.genie


jj <- list(trace=100,maxit=10)

hpa.genie.level1 <- opt.1(D=D1.genie, z=z.genie, basis=basis.genie, subsets=subsets.genie, hpa.start=hpa.genie.start,control=jj)
 
hpa.genie.level2 <- opt.gt.1(level=2, D=D1.genie, z=z.genie, basis=basis.genie, subsets=subsets.genie, hpa.start=hpa.genie.level1,control=jj)

hpa.genie.level3 <- opt.gt.1(level=3, D=D1.genie, z=z.genie, basis=basis.genie, subsets=subsets.genie, hpa.start=hpa.genie.level2,control=jj)