package jsat.math.optimization;
import jsat.linear.DenseVector;
import jsat.linear.Vec;
import jsat.math.Function;
import static java.lang.Math.*;
import java.util.concurrent.ExecutorService;
import jsat.math.FunctionVec;
/**
* The Rosenbrock function is a function with at least one minima with the value zero. It is often used as a benchmark for optimization problems.
* The minima is the vector of all ones. Once N %gt; 3, more then one minima can occur.
*
* @author Edward Raff
*/
public class RosenbrockFunction implements Function
{
private static final long serialVersionUID = -5573482950045304948L;
@Override
public double f(Vec x, boolean parallel)
{
int N = x.length();
double f = 0.0;
for(int i = 1; i < N; i++)
{
double x_p = x.get(i-1);
double xi = x.get(i);
f += pow(1.0-x_p, 2)+100.0*pow(xi-x_p*x_p, 2);
}
return f;
}
/**
* Returns the gradient of the Rosenbrock function
* @return the gradient of the Rosenbrock function
*/
public FunctionVec getDerivative()
{
return GRADIENT;
}
/**
* The gradient of the Rosenbrock function
*/
public static final FunctionVec GRADIENT = new FunctionVec()
{
@Override
public Vec f(Vec x, Vec drv, boolean parallel)
{
int N = x.length();
if (drv == null)
drv = x.clone();
drv.zeroOut();
drv.set(0, -400 * x.get(0) * (x.get(1) - pow(x.get(0), 2)) - 2 * (1 - x.get(0)));
for (int i = 1; i < N - 1; i++)
{
double x_p = x.get(i - 1);
double x_i = x.get(i);
double x_n = x.get(i + 1);
drv.set(i, 200 * (x_i - x_p * x_p) - 400 * x_i * (x_n - x_i * x_i) - 2 * (1 - x_i));
}
drv.set(N - 1, 200 * (x.get(N - 1) - pow(x.get(N - 2), 2)));
return drv;
}
};
}
Ads help maintain this website.