到目前为止,我已经开发了一个类,可以使用树状图表示魔方(最好的方法?)每种颜色都映射到键 0 - 53。 键始终保持不变,并由立方体的 2D 视图表示。我做了一个 rotate90 方法,该方法将采用一种颜色并将颜色面(中间立方体永远不会移动)顺时针旋转 90 度。所以我的问题是,我如何实现一个好的搜索(也许是 A*?)让这个多维数据集朝着正确的方向发展到它的 goalState(我也在代码中将其初始化为全局,以便我可以将其与当前状态进行比较)。任何正确方向的反馈或提示都会很棒。谢谢!
这是没有 rotate 90 方法的代码和一个只检查文件文本是否正常的 checkInput 方法。输入只是一个表示 2D 立方体的文本文件,如下所示。现在的解决方法只是我胡闹,并没有真正到达任何地方。
GGW
RRG
RRG
OWWGGOYYR
OGOYYYRBR
YYYRBGRWW
BOY
BOB
BOB
OGO
WWB
WWB
package rubik;
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.util.NavigableMap;
import java.util.TreeMap;
public class RubikCube {
final static String FILE_NAME = "rubikTest.txt";
public TreeMap<Integer, Character> goalState;
//swaps two colors given a cube
public static void swapColorValue(int color1, int color2, TreeMap<Integer, Character> rubik){
char tempColor = rubik.get(color1);
rubik.put(color1,rubik.get(color2));
rubik.put(color2, tempColor);
}
//initialize the goal map
public void goalInit(TreeMap<Integer, Character> rubik){
//initialize the goal state map each rubik.get(magicNumber) is the static center cubie
TreeMap<Integer, Character> goalMap = new TreeMap<Integer,Character>();
for(int i =0;i<=8;i++){
goalMap.put(i,rubik.get(4));
}
//the left side
goalMap.put(9, rubik.get(19));
goalMap.put(10, rubik.get(19));
goalMap.put(11, rubik.get(19));
goalMap.put(18, rubik.get(19));
goalMap.put(19, rubik.get(19));
goalMap.put(20, rubik.get(19));
goalMap.put(27, rubik.get(19));
goalMap.put(28, rubik.get(19));
goalMap.put(29, rubik.get(19));
//the middle
goalMap.put(12, rubik.get(22));
goalMap.put(13, rubik.get(22));
goalMap.put(14, rubik.get(22));
goalMap.put(21, rubik.get(22));
goalMap.put(22, rubik.get(22));
goalMap.put(23, rubik.get(22));
goalMap.put(30, rubik.get(22));
goalMap.put(31, rubik.get(22));
goalMap.put(32, rubik.get(22));
//right side
goalMap.put(15, rubik.get(25));
goalMap.put(16, rubik.get(25));
goalMap.put(17, rubik.get(25));
goalMap.put(24, rubik.get(25));
goalMap.put(25, rubik.get(25));
goalMap.put(26, rubik.get(25));
goalMap.put(33, rubik.get(25));
goalMap.put(34, rubik.get(25));
goalMap.put(35, rubik.get(25));
//bottom
for(int i = 36;i<=44;i++){
goalMap.put(i, rubik.get(40));
}
//back
for(int i = 45;i<=53;i++){
goalMap.put(i, rubik.get(49));
}
//give it to the global variable
goalState = (TreeMap<Integer, Character>) goalMap.clone();
}
//Maps a Integer key to a color given a file.
public static NavigableMap<Integer, Character> setup(String file) throws IOException{
TreeMap<Integer, Character> rubik = new TreeMap<Integer, Character>();
BufferedReader br = new BufferedReader(new FileReader(file));
try {
StringBuilder sb = new StringBuilder();
String line = br.readLine();
//add each line (without white spaces) to the stringbuilder
while (line != null) {
sb.append(line.trim());
line = br.readLine();
}
//convert the stringbuilder(which has all the input from the file) to a char array
char [] colors = sb.toString().toCharArray();
//put the key,color into the Treemap.
for(int i =0; i < colors.length;i++){
rubik.put(i, colors[i]);
}
} finally {
br.close();
}
//type Tree map
return rubik;
}
public int solve(TreeMap<Integer, Character> rubik){
int j = 1;
int check = 0;
int redMatches=0;
char [] colors = {'r','g','y','b','o','w'};
for(int i = 0; i < 100;i++ ){
if(j==6) j = 1;
redMatches = 0;
rotate90(colors[j],rubik);
if(rubik.get(check)==goalState.get(check)){
System.out.print("rubik match at: "+i+" with: "+rubik.get(i)+"match at goalState: "+i+" with: "+goalState.get(i));
redMatches++;
}
j++;
}
return redMatches;
}
public static void main (String[] args){
try {
//check if file input is good
//System.out.print(new RubikCube().checkInput(FILE_NAME));
//Map the rubik cube(key,Color)
RubikCube cubeInst = new RubikCube();
//make sure to set up the mapping before calling rotate and goalInit
TreeMap<Integer, Character> cube = (TreeMap<Integer, Character>) (setup(FILE_NAME));
cubeInst.goalInit(cube);
//System.out.print(cubeInst.goalState);
//rotate90('y',cube);
//rotate90('y',cube);
//System.out.print(cube);
System.out.print(cubeInst.solve(cube));
} catch (IOException e) {
e.printStackTrace();
}
}
}
我从来没有写过代码来解决魔方,但如果你使用像 A* 这样的启发式图遍历算法,我可以想到一些好的启发式方法来判断你离目标有多远:
-
放错地方的小隔间数量。
-
每个小隔间离目标状态有多远的总和,可能使用曼哈顿距离来判断它有多远。
问题是我不认为你可以为每个小隔间状态使用一种颜色。 您可能需要从某个任意角度将多维数据集表示为数据结构位置。 所以例如"左上角"是目标状态下的左上角部分(这也是左侧和背面的角隔间...... 然后你就会知道你的"左上角"小隔间是否不合适。