/*
 * Copyright (c) 2012-2013, Peter Abeles. All Rights Reserved.
 *
 * This file is part of DDogleg (http://ddogleg.org).
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *   http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package org.ddogleg.optimization;

import org.ddogleg.optimization.funcs.*;
import org.ddogleg.optimization.functions.FunctionNtoMxN;
import org.ddogleg.optimization.functions.FunctionNtoN;
import org.ddogleg.optimization.functions.FunctionNtoS;
import org.ddogleg.optimization.impl.NumericalGradientForward;
import org.ddogleg.optimization.wrap.LsToNonLinear;
import org.ddogleg.optimization.wrap.LsToNonLinearDeriv;

/**
 * @author Peter Abeles
 */
public abstract class UnconstrainedMinimizationEvaluator {

	boolean verbose = true;
	boolean printSummary = true;
	int maxIteration = 1000;

	protected UnconstrainedMinimizationEvaluator(boolean verbose, boolean printSummary) {
		this.verbose = verbose;
		this.printSummary = printSummary;
	}

	/**
	 * Creates a line search algorithm
	 *
	 * @return Line search algorithm
	 */
	protected abstract UnconstrainedMinimization createSearch();

	/**
	 * Run the line search algorithm on the two inputs and compute statistics
	 *
	 * @param func Function being searched
	 * @param deriv Derivative being searched
	 * @param initial Initial point
	 * @return statics
	 */
	private NonlinearResults performTest( FunctionNtoS func , FunctionNtoN deriv ,
										  double initial[] , double optimal[] , double minimValue)
	{
		if( deriv == null ) {
			deriv = new NumericalGradientForward(func);
		}

		CallCounterNtoS f = new CallCounterNtoS(func);
		CallCounterNtoN d = new CallCounterNtoN(deriv);

		UnconstrainedMinimization alg = createSearch();
		alg.setFunction(f,d,minimValue);

		double initialValue = func.process(initial);

		alg.initialize(initial,1e-12,1e-12);
		int iter;
		for( iter = 0; iter < maxIteration && !alg.iterate() ; iter++ ){
			printError(optimal, alg);
		}
		printError(optimal, alg);
		if( verbose )
			System.out.println("*** total iterations = "+iter);
		double found[] = alg.getParameters();

		double finalValue = func.process(found);

		if( printSummary ) {
			System.out.printf("value{ init %4.1e final = %6.2e} count f = %2d d = %2d\n",
					initialValue, finalValue, f.count, d.count);
		}

		NonlinearResults ret = new NonlinearResults();
		ret.numFunction = f.count;
		ret.numGradient = d.count;
		ret.f = finalValue;
		ret.x = found;

		return ret;
	}

	private void printError(double[] optimal, UnconstrainedMinimization alg) {
		if( optimal != null ) {
			double x[] = alg.getParameters();
			double n = 0;
			for( int j = 0; j < x.length; j++ ) {
				double dx = x[j]-optimal[j];
				n += dx*dx;
			}
			if( verbose )
				System.out.println("||x(k)-x(*)|| = "+Math.sqrt(n));
		}
	}

	private NonlinearResults performTest( EvalFuncLeastSquares func ) {
		FunctionNtoS nl = new LsToNonLinear(func.getFunction());
		double[] initial = func.getInitial();

		FunctionNtoMxN jacobian = func.getJacobian();
		FunctionNtoN gradient = jacobian == null ? null : new LsToNonLinearDeriv(func.getFunction(),jacobian);

		if( verbose && func.getOptimal() != null )
			System.out.println("optimal = "+nl.process(func.getOptimal()));

		return performTest(nl,gradient,initial,func.getOptimal(),0);
	}

	private NonlinearResults performTest( EvalFuncMinimization func ) {
		double[] initial = func.getInitial();
		
		FunctionNtoS nl = func.getFunction();

		if( verbose && func.getOptimal() != null )
			System.out.println("optimal = "+nl.process(func.getOptimal()));

		return performTest(nl,func.getGradient(),initial,func.getOptimal(),func.getMinimum());
	}

	public NonlinearResults helicalValley() {
		return performTest(new EvalFuncHelicalValley());
	}

	public NonlinearResults rosenbrock() {
		return performTest(new EvalFuncRosenbrock());
	}

	public NonlinearResults rosenbrockMod( double lambda ) {
		return performTest(new EvalFuncRosenbrockMod(lambda));
	}

	public NonlinearResults dodcfg() {
		return performTest(new EvalFuncDodcfg(50,50,8.0e-3));
	}

	public NonlinearResults variably() {
		return performTest(new EvalFuncVariablyDimensioned(10));
	}

	public NonlinearResults trigonometric() {
		return performTest(new EvalFuncTrigonometric(10));
	}

	public NonlinearResults badlyScaledBrown() {
		return performTest(new EvalFuncBadlyScaledBrown());
	}

	public NonlinearResults powell() {
		return performTest(new EvalFuncPowell());
	}
}