optimx (version 2018-7.10)

hjn: Compact R Implementation of Hooke and Jeeves Pattern Search Optimization

Description

The purpose of hjn is to minimize an unconstrained or bounds (box) and mask constrained function of several parameters by a Hooke and Jeeves pattern search. This code is entirely in R to allow users to explore and understand the method. It also allows bounds (or box) constraints and masks (equality constraints) to be imposed on parameters.

Usage

hjn(par, fn, lower=-Inf, upper=Inf, bdmsk=NULL, control = list(trace=0), …)

Arguments

par

A numeric vector of starting estimates.

fn

A function that returns the value of the objective at the supplied set of parameters par using auxiliary data in …. The first argument of fn must be par.

lower

A vector of lower bounds on the parameters.

upper

A vector of upper bounds on the parameters.

bdmsk

An indicator vector, having 1 for each parameter that is "free" or unconstrained, and 0 for any parameter that is fixed or MASKED for the duration of the optimization.

control

An optional list of control settings.

Further arguments to be passed to fn.

Value

A list with components:

par

The best set of parameters found.

value

The value of the objective at the best set of parameters found.

counts

A two-element integer vector giving the number of calls to 'fn' and 'gr' respectively. This excludes those calls needed to compute the Hessian, if requested, and any calls to 'fn' to compute a finite-difference approximation to the gradient.

convergence

An integer code. '0' indicates successful convergence. '1' indicates that the function evaluation count 'maxfeval' was reached.

message

A character string giving any additional information returned by the optimizer, or 'NULL'.

Details

Functions fn must return a numeric value.

The control argument is a list.

maxfeval

A limit on the number of function evaluations used in the search.

trace

Set 0 (default) for no output, >0 for trace output (larger values imply more output).

eps

Tolerance used to calculate numerical gradients. Default is 1.0E-7. See source code for hjn for details of application.

dowarn

= TRUE if we want warnings generated by optimx. Default is TRUE.

tol

Tolerance used in testing the size of the pattern search step.

Note that the control maximize should NOT be used.

References

Nash JC (1979). Compact Numerical Methods for Computers: Linear Algebra and Function Minimisation. Adam Hilger, Bristol. Second Edition, 1990, Bristol: Institute of Physics Publications.

See Also

optim

Examples

Run this code
# NOT RUN {
#####################
## Rosenbrock Banana function
fr <- function(x) {
    x1 <- x[1]
    x2 <- x[2]
    100 * (x2 - x1 * x1)^2 + (1 - x1)^2
}

ansrosenbrock0 <- hjn(fn=fr, par=c(1,2), control=list(maxfeval=2000, trace=0))
print(ansrosenbrock0) # use print to allow copy to separate file that 

#    can be called using source()
#####################
genrose.f<- function(x, gs=NULL){ # objective function
## One generalization of the Rosenbrock banana valley function (n parameters)
	n <- length(x)
        if(is.null(gs)) { gs=100.0 }
	fval<-1.0 + sum (gs*(x[1:(n-1)]^2 - x[2:n])^2 + (x[2:n] - 1)^2)
        return(fval)
}

xx<-rep(pi,10)
lower<-NULL
upper<-NULL
bdmsk<-NULL

cat("timings B vs U\n")
lo<-rep(-100,10)
up<-rep(100,10)
bdmsk<-rep(1,10)
tb<-system.time(ab<-hjn(xx,genrose.f, lower=lo, upper=up,
          bdmsk=bdmsk, control=list(trace=0, maxfeval=2000)))[1]
tu<-system.time(au<-hjn(xx,genrose.f, control=list(maxfeval=2000, trace=0)))[1]
cat("times U=",tu,"   B=",tb,"\n")
cat("solution hjnu\n")
print(au)
cat("solution hjnb\n")
print(ab)
cat("diff fu-fb=",au$value-ab$value,"\n")
cat("max abs parameter diff = ", max(abs(au$par-ab$par)),"\n")

######### One dimension test
sqtst<-function(xx) {
   res<-sum((xx-2)*(xx-2))
}

nn<-1
startx<-rep(0,nn)
onepar<-hjn(startx,sqtst,control=list(trace=1)) 
print(onepar)
# }

Run the code above in your browser using DataCamp Workspace