ExponentiatedGradientMinimizer.java

package edu.berkeley.nlp.math;

import edu.berkeley.nlp.util.Logger;

public class ExponentiatedGradientMinimizer implements GradientMinimizer {
	private final static double EPS = 1e-10;
	
	private final Normalizer normalizer;
	private final int maxIterations;
	private final double stepSizeMultiplier;
	
	public ExponentiatedGradientMinimizer(Normalizer normalizer, int maxIterations, double stepSizeMultiplier) {
		this.normalizer = normalizer;
		this.maxIterations = maxIterations;
		this.stepSizeMultiplier = stepSizeMultiplier;
	}
	
	public double[] minimize(DifferentiableFunction function, double[] initial, double tolerance, boolean project){
		return null;
	}
	
	public double[] minimize(DifferentiableFunction function, double[] initial, double tolerance) {
		double[] guess = DoubleArrays.clone(initial);
		for (int iteration=0; iteration<maxIterations; iteration++) {
			double[] derivatives = function.derivativeAt(guess);
			double value = function.valueAt(guess);
			double[] direction = DoubleArrays.multiply(derivatives, -stepSizeMultiplier);
			double[] scale = DoubleArrays.exponentiate(direction);
			double[] newGuess = normalizer.normalize(DoubleArrays.pointwiseMultiply(guess, scale));
			double newValue = function.valueAt(newGuess);
			Logger.i().logs(String.format("Exponentiated gradient: iteration %d (max %d) completed with value %f", iteration + 1, maxIterations, newValue));
//			if (converged(value, newValue, tolerance)) {
//				return newGuess;
//			}
			guess = newGuess;
		}
		return guess;
	}

	private boolean converged(double value, double nextValue, double tolerance) {
		if (value == nextValue)
			return true;
		double valueChange = Math.abs(nextValue - value);
		double valueAverage = Math.abs(nextValue + value + EPS) / 2.0;
		if (valueChange / valueAverage < tolerance)
			return true;
		return false;
	}
}