📄 population.java
字号:
package com.dbhunter.ga;
import java.util.Vector;
import java.util.Random;
import javax.swing.ProgressMonitor;
public class Population {
private int populationNum;
private int selectionID;
private int crossID;
private int genovariationID;
private int iterateNum;
private int showStyle;
private double pc;
private double pv;
private Function ef;
private int bestGene;
private Gene sumGene;
private Gene avgGene;
private Gene[] genes;
private StringBuffer sbInitGenes;
/**
*
* @param populationNum 种群规模
* @param iterateNum 迭代次数
* @param pc 交叉概率
* @param pv 变异概率
* @param ef 评估函数
*/
public Population(int populationNumv,int iterateNumv,int selectionIDv,int crossIDv,int genovariationIDv,double pc,double pv,Function ef,int showStyle){
this.populationNum = populationNumv;
this.selectionID = selectionIDv;
this.crossID = crossIDv;
this.genovariationID = genovariationIDv;
this.iterateNum = iterateNumv;
this.pc = pc;
this.pv = pv;
this.ef = ef;
this.showStyle = showStyle;
this.sumGene = new Gene();
this.avgGene = new Gene();
this.sbInitGenes = new StringBuffer();
/**
* 初始化种群
* 一定要注意把BinaryGene bc =new BinaryGene(parameterNum,codeLen) 放入循环内部
*/
int codeLen= ef.getCodeLen();
int parameterNum = ef.getParameterNum();
this.genes = new Gene[populationNum];
for(int i=0;i<populationNum;i++){
Code bc =new Code(ef.getCodeType(),parameterNum,codeLen);
genes[i] = new Gene();
genes[i].setGeneID(i);
Vector[] vecGene = bc.getGene();
genes[i].setVecGene(vecGene);
genes[i].setEvaluateValue(ef.fun(vecGene));
}
genes = calculateRoulette(genes);
/**
* 下面调用选择，交叉，变异
*/
sbInitGenes = ef.getSbLastGenes(this.showStyle,genes);
final ProgressMonitor pm = new ProgressMonitor(null,"程序正在计算......","当前运行：0%",0,iterateNum);
pm.setMillisToPopup(0);
pm.setMillisToDecideToPopup(0);
for(int i=0;i<iterateNum;i++){
pm.setProgress(i);
pm.setNote("当前运行："+ef.doubleToString(2,(i*100.00)/iterateNum)+" %");
genes = selection(selectionID,genes);
genes = cross(crossID,genes);
genes = genovariation(genovariationID,genes);
genes = calculateRoulette(genes);
}
pm.close();
}
/**
*
* author dbhunter
* @param targetGene 未添加赌轮值得基因，该基因只有编码和适应度函数值
* @return 根据该基因的特征 生成下代可以选择的基因
*/
public Gene[] calculateRoulette(Gene[] targetGene){
double sumEvaluateValue =0;
for(int i=0;i<populationNum;i++){
/**
* 初始化为0
*/
targetGene[i].setSelectProbability(0);
targetGene[i].setExpectValue(0);
targetGene[i].setRouletteNum(0);
sumEvaluateValue = sumEvaluateValue +targetGene[i].getEvaluateValue();
}
this.sumGene.setEvaluateValue(sumEvaluateValue);
this.avgGene.setEvaluateValue(sumEvaluateValue/populationNum);
this.sumGene.setSelectProbability((double)1);
this.avgGene.setSelectProbability((double)1/populationNum);
this.sumGene.setExpectValue(populationNum);
this.avgGene.setExpectValue((double)1);
double minev = this.avgGene.getExpectValue();//最小期望值,初始化为平均期望值
for(int i=0;i<populationNum;i++){ //求选择概率和适应函数期望值
double sp = targetGene[i].getEvaluateValue()/this.sumGene.getEvaluateValue();
double ev = targetGene[i].getEvaluateValue()/this.avgGene.getEvaluateValue();
targetGene[i].setSelectProbability(sp);
targetGene[i].setExpectValue(ev);
if(sp<=minev){ //取出最小期望值基因
minev = sp;
this.bestGene = i;
}
}
/**
* 使用赌轮法选择基因
*/
int counter = 0 ;
for(int i=0;i<this.populationNum;i++){
if((int)targetGene[i].getExpectValue()==0){
targetGene[i].setRouletteNum(targetGene[i].getRouletteNum()+1);
counter++;
}
}
int index=1; //标识检查到小数点后第index为即可选择够基因
while(counter<this.getPopulationNum()){
for(int i=0;i<10;i++){
for(int j=0;j<this.populationNum;j++){
if(ef.getNum(targetGene[j].getExpectValue(),index)==i){
if(counter>=this.populationNum){
break;
}
else{
targetGene[j].setRouletteNum(targetGene[j].getRouletteNum()+1);
counter++;
}
}
}
}
index++;
}
return targetGene;
}
/**
* 2008.4.12 换用新的选择算法selection 降低空间复杂度
public Gene[] selectGene(Gene[] targetGene){
Gene[] newGenes = new Gene[this.populationNum];
int k = 0;
for(int i=0;i<this.populationNum;i++){
for(int j=0;j<targetGene[i].getRouletteNum();j++){
newGenes[k] = new Gene();
newGenes[k].setGeneID(k);
Vector[] vecGene = new Vector[ef.getParameterNum()];
for(int k=0;k<ef.getParameterNum();k++){
vecGene[k] = new Vector();
vecGene[k].addAll(targetGene[i].getVecGene()[k]);
}
newGenes[k].setVecGene(vecGene);
newGenes[k].setEvaluateValue(ef.fun(targetGene[i].getVecGene()));
k++;
}
}
return newGenes;
}
**/
public Gene[] selection(int selectionID,Gene[] targetGene){
switch(selectionID){
case 1:
int pointer0 = 0;
for(int i=0;i<this.populationNum;i++){
if (targetGene[i].getRouletteNum()>1){
for(int j=pointer0;j<this.populationNum;j++){
if (targetGene[j].getRouletteNum()==0){
pointer0 = j+1; //记录下一次循环的开始位置
/**
* 此处一定要重新开辟向量空间，不能使用targetGene[j].setVecGene(vecGenetargetGene[i].getVecGene()[k]);
* 那只是向量地址指向
*/
Vector[] vecGene = new Vector[ef.getParameterNum()];
for(int k=0;k<ef.getParameterNum();k++){
vecGene[k] = new Vector();
vecGene[k].addAll(targetGene[i].getVecGene()[k]);
}
targetGene[j].setVecGene(vecGene);
targetGene[j].setEvaluateValue(targetGene[i].getEvaluateValue());
targetGene[j].setRouletteNum(targetGene[j].getRouletteNum()+1);
targetGene[i].setRouletteNum(targetGene[i].getRouletteNum()-1);
break;
}
}
}
}
return targetGene;
default:
return targetGene;
}
}
/**
*
* author dbhunter
* @param crossID 交叉的方法
* @param targetGene 目标基因组
* @return 返回交叉后的基因
*/
public Gene[] cross(int crossID,Gene[] targetGene){
switch(crossID){
case 1:
/**
* 单点交叉
*/
for(int i=0;i<this.populationNum;i++){
Random rd = new Random();
double randomPc = rd.nextDouble();
if(randomPc<=this.pc){
int randoma,randomb,randomPosition;
int codeLen= this.ef.getCodeLen();
randoma = rd.nextInt(this.populationNum); //随机基因a
randomb = rd.nextInt(this.populationNum); //随机基因b
for(int j=0;j<ef.getParameterNum();j++){
randomPosition = rd.nextInt(codeLen); //交叉随机位
for(int k=0;k<randomPosition;k++){
int temp;
temp = ef.stringToInt(targetGene[randoma].getVecGene()[j].elementAt(k).toString());
targetGene[randoma].getVecGene()[j].setElementAt(targetGene[randomb].getVecGene()[j].elementAt(k),k);
targetGene[randomb].getVecGene()[j].setElementAt(temp, k);
}
}
}
}
/**
* 交叉后重新计算适应度值
* 2008.4.11
*/
for(int i=0;i<this.populationNum;i++){
targetGene[i].setEvaluateValue(ef.fun(targetGene[i].getVecGene()));
}
return targetGene;
case 2:
/**
* 使用佳点集
*/
for(int i=0;i<this.populationNum;i++){
Random rd = new Random();
double randomPc = rd.nextDouble();
if(randomPc<=this.pc){
int randoma,randomb;
randoma = rd.nextInt(this.populationNum); //随机基因a
randomb = rd.nextInt(this.populationNum); //随机基因b
for(int j=0;j<ef.getParameterNum();j++){
Vector vc = ef.getDifferentPartPosition(targetGene[randoma].getVecGene()[j], targetGene[randomb].getVecGene()[j]);
for(int k=0;k<vc.size();k++){
/**
* 此处用到下列函数
* getCodeLen 获得最大的维数
* minPrimeNum 取得该维数满足 p>=2*s+3的最小素数
* stringToInt 将vector中的string对象值转化为int类型
* goodPointFun 佳点集函数 r(k,p) = 2*cos(2*pi*k/p)
* getDecimalPart 取得该函数对应第i个基因的第k维对应的佳点集函数值
* codeToBinary() 将该函数值转化为二进制数
* 注意此处 佳点集的要求 第i个个体 1<i<N 1<K<s p>=2*s+3,s=vc.size()+1
*/
targetGene[randoma].getVecGene()[j].setElementAt(ef.codeToBinary(ef.getDecimalPart((randoma+1)*ef.goodPointFun(ef.stringToInt(vc.elementAt(k).toString())+1,ef.minPrimeNum(2*(vc.size()+1)+3)))),ef.stringToInt(vc.elementAt(k).toString()));
targetGene[randomb].getVecGene()[j].setElementAt(ef.codeToBinary(ef.getDecimalPart((randomb+1)*ef.goodPointFun(ef.stringToInt(vc.elementAt(k).toString())+1,ef.minPrimeNum(2*(vc.size()+1)+3)))),ef.stringToInt(vc.elementAt(k).toString()));
}
}
}
}
/**
* 交叉后重新计算适应度值
* 2008.4.11
*/
for(int i=0;i<this.populationNum;i++){
targetGene[i].setEvaluateValue(ef.fun(targetGene[i].getVecGene()));
}
return targetGene;
default:
return targetGene;
}
}
/**
*
* author dbhunter
* @param genovariationID
* @param targetGene
* @return
*/
public Gene[] genovariation(int genovariationID,Gene[] targetGene){
switch(genovariationID){
case 1:
for(int i=0;i<this.populationNum;i++){
for(int j=0;j<ef.getParameterNum();j++){
for(int k=0;k<ef.getCodeLen();k++){
Random rd = new Random();
double random = rd.nextDouble();
if(random< this.pv){
if(ef.stringToInt(targetGene[i].getVecGene()[j].elementAt(k).toString())==0){
targetGene[i].getVecGene()[j].setElementAt(1, k);
}
else{
targetGene[i].getVecGene()[j].setElementAt(0, k);
}
}
}
}
}
/**
* 变异后重新计算适应度值
* 2008.4.11
*/
for(int i=0;i<this.populationNum;i++){
targetGene[i].setEvaluateValue(ef.fun(targetGene[i].getVecGene()));
}
return targetGene;
case 2:
default:
return targetGene;
}
}
/**
*
* author dbhunter
* @param precision 指定字符串的精度
* @param d double类型的值
* @return
*/
public Gene[] getGenes() {
return genes;
}
public void setGenes(Gene[] genes) {
this.genes = genes;
}
public Gene getAvgGene() {
return avgGene;
}
public void setAvgGene(Gene avgGene) {
this.avgGene = avgGene;
}
public int getBestGene() {
return bestGene;
}
public void setBestGene(int bestGene) {
this.bestGene = bestGene;
}
public int getPopulationNum() {
return populationNum;
}
public void setPopulationNum(int populationNum) {
this.populationNum = populationNum;
}
public Gene getSumGene() {
return sumGene;
}
public void setSumGene(Gene sumGene) {
this.sumGene = sumGene;
}
public int getCrossID() {
return crossID;
}
public void setCrossID(int crossID) {
this.crossID = crossID;
}
public Function getEf() {
return ef;
}
public void setEf(Function ef) {
this.ef = ef;
}
public int getGenovariationID() {
return genovariationID;
}
public void setGenovariationID(int genovariationID) {
this.genovariationID = genovariationID;
}
public double getPc() {
return pc;
}
public void setPc(double pc) {
this.pc = pc;
}
public double getPv() {
return pv;
}
public void setPv(double pv) {
this.pv = pv;
}
public int getSelectionID() {
return selectionID;
}
public void setSelectionID(int selectionID) {
this.selectionID = selectionID;
}
public int getIterateNum() {
return iterateNum;
}
public void setIterateNum(int iterateNum) {
this.iterateNum = iterateNum;
}
public StringBuffer getSbInitGenes() {
return sbInitGenes;
}
public void setSbInitGenes(StringBuffer sbInitGenes) {
this.sbInitGenes = sbInitGenes;
}
}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -