# unscalingFun1d: Unscaling in one dimension In IRSN/RobustInv: "Robust inversion of expensive black-box functions"

## Description

Unscaling in one dimension using the knots of a km object

## Usage

 ```1 2``` ```unscalingFun1d(y, knots, eta, standardize = FALSE, lower = NULL, upper = NULL) ```

## Arguments

 `y` Array with values to be unscaled `knots` Array obtained from the field `[email protected]@knots` of a `km` object `eta` Array obtained from the field `[email protected]@eta` of a km object `standardize` If the initial values y are not in [0,S], with S the integral of the piecewise linear function equal to eta at points knots, then there is the possibility to rescale the values y by indicating in which interval they are. `lower` If `standardize=TRUE`, this is the lower bound of the interval where the yi's are. `upper` If `standardize=TRUE`, this is the upper bound of the interval where the yi's are.

## Value

Array with size `length(y)` containing the scaled uncoordinates.

## Author(s)

Clement Chevalier [email protected]

## Examples

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19``` ```library(DiceKriging) knots <- c(1,2,3) eta <- c(2,1,4) t <- seq(from = 1, to = 3, length = 101) scaled_t <- scalingFun1d(x = t,knots = knots,eta = eta) result <- unscalingFun1d(scaled_t,knots=knots,eta=eta) # now result is equal to t ! # an example of unscaling of uniformly distributed points # to have more points in regions where eta is large. myrands <- matrix( runif(2000),ncol=2 ) knots <- c(0,0.5,1) eta <- c(5,1,5) # large on the bounds, low in the middle res1 <- unscalingFun1d(y = myrands[,1] , knots=knots , eta = eta , standardize = TRUE, lower=0, upper = 1) res2 <- unscalingFun1d(y = myrands[,2] , knots=knots , eta = eta , standardize = TRUE, lower=0, upper = 1) plot(x=res1 , y=res2, type="p") # more points in the corners ```

IRSN/RobustInv documentation built on Dec. 8, 2018, 2:17 a.m.