downhillsimplex: downhillsimplex function

In gpHist: Gaussian Process with Histogram Intersection Kernel

Description

Impementation of the downhillsimplex function for optimisation

Usage

downhillsimplex(fc,nParams,bp=NULL,lower=0,upper=1,it=1000,tol=1e-10,
control= list(alpha=1,gamma=2, beta=0.5,sigma = 0.5) )
  

Arguments

fc	Function to be minimized
nParams	Number of parameters that the function to be optimized requires
bp	Matrix used to start the optimisation. If no parameters is given, this matrix will be randamly initialized.
lower	Vector for the lower bound of the parameters
upper	Vector for the upper bound of the parameters
it	Maximium number of iterations done by the downhillsimplex algortihm
tol	tolearnce that is used to test for convergence
control	list of variables that control the behavior of the downhillsimpley algorithm. alpha = Reflexion gamma = Expansion beta = Contraktion sigma = Compression

Details

Implemented in C, will call you R function

Value

Returns estimated parameters

Note

Implementation according to Wikipedia page

Author(s)

Dennis Becker

References

Implementation according to describtion on wiki: https://en.wikipedia.org/wiki/Nelder

See Also

Package Overview: gpHist-Package

Function for hyperparameter estimation: estimateHyperParameters

Examples

#simple function to be optimized
fn = function(x){
 x^2 
}


#call downhillsimplex
res = downhillsimplex(fn,1,lower=-10,upper=10,it=1000,tol=1e-10)

#plot results
x = seq(-10,10,0.01)
plot(x,fn(x),type='l')
points(res[1],fn(res[1]),col='red')


legend('topleft',legend=c('Function', 'Estimated minimum'), col=c('black','red'),
lty=c(1,NA),pch=c(NA,1))

gpHist documentation built on Nov. 24, 2017, 5:03 p.m.