Use of the shuffle algorithm in Java
- 2020-04-01 01:39:19
- OfStack
Fisher Yates shuffle basic ideas (Knuth shuffle) :
To shuffle an array a of n elements (indices 0.. N - 1) :
For I from n − 1 downto 1 do
J please random integer with 0 Or less j Or less I
Exchange a [j] and a [I]
The JDK source code is as follows:
public static void shuffle(List<?> list) {
shuffle(list, new Random());
}
@SuppressWarnings("unchecked")
public static void shuffle(List<?> list, Random random) {
if (!(list instanceof RandomAccess)) {
Object[] array = list.toArray();
for (int i = array.length - 1; i > 0; i--) {
int index = random.nextInt(i + 1);
if (index < 0) {
index = -index;
}
Object temp = array[i];
array[i] = array[index];
array[index] = temp;
}
int i = 0;
ListIterator<Object> it = (ListIterator<Object>) list
.listIterator();
while (it.hasNext()) {
it.next();
it.set(array[i++]);
}
} else {
List<Object> rawList = (List<Object>) list;
for (int i = rawList.size() - 1; i > 0; i--) {
int index = random.nextInt(i + 1);
if (index < 0) {
index = -index;
}
rawList.set(index, rawList.set(i, rawList.get(index)));
}
}
}
To test the code, multiple containers are used to ensure the same initialization for each case:
public class javaShuffle {
public static int temp = 0;
public static long start;
public static long end;
public static void main(final String args[]) {
Object changeTemp;
List<Integer> numList = new ArrayList<Integer>();
List<Integer> firstList = new ArrayList<Integer>();
List<Integer> secondList = new ArrayList<Integer>();
List<Integer> thirdList = new ArrayList<Integer>();
List<Integer> fourthList = new ArrayList<Integer>();
for (int i = 1; i <= 100000; i++) {
numList.add(i);
firstList.add(i);
secondList.add(i);
thirdList.add(i);
fourthList.add(i);
}
// first shuffle,use changeTemp
getStartTime();
int randInt = 0;
for (int i = 0, length = firstList.size(); i < length; i++) {
randInt = getRandom(i, firstList.size());
changeTemp = firstList.get(i);
firstList.set(i, firstList.get(randInt));
firstList.set(randInt, javaShuffle.temp);
}
getEndTime("first shuffle run time ");
// second shuffle,exchange list
getStartTime();
for (int i = 0, length = secondList.size(); i < length; i++) {
randInt = getRandom(i, secondList.size());
secondList.set(i, secondList.set(randInt, secondList.get(i)));
}
getEndTime("second shuffle run time");
// third shuffle, change generate random int
getStartTime();
Object[] tempArray = thirdList.toArray();
Random rand = new Random();
int j = 0;
for (int i = tempArray.length - 1; i > 0; i--) {
j = rand.nextInt(i + 1);
thirdList.set(i, thirdList.set(j, thirdList.get(i)));
}
getEndTime("third shuffle run time ");
// fourth shuffle, simulate java shuffle
getStartTime();
Random random = new Random();
if (!(fourthList instanceof RandomAccess)) {
Object[] array = fourthList.toArray();
for (int i = array.length - 1; i > 0; i--) {
int index = random.nextInt(i + 1);
if (index < 0) {
index = -index;
}
Object temp = array[i];
array[i] = array[index];
array[index] = temp;
}
int i = 0;
ListIterator<Integer> it = (ListIterator<Integer>) fourthList.listIterator();
while (it.hasNext()) {
it.next();
it.set((Integer) array[i++]);
}
} else {
List<Integer> rawList = (List<Integer>) fourthList;
for (int i = rawList.size() - 1; i > 0; i--) {
int index = random.nextInt(i + 1);
if (index < 0) {
index = -index;
}
rawList.set(index, rawList.set(i, rawList.get(index)));
}
}
getEndTime("fourth shuffle run time");
// java shuffle
getStartTime();
Collections.shuffle(numList);
getEndTime("java shuffle run time ");
}
public static void swap(int a, int b) {
javaShuffle.temp = a;
a = b;
b = javaShuffle.temp;
}
public static int getRandom(final int low, final int high) {
return (int) (Math.random() * (high - low) + low);
}
public static void getStartTime() {
javaShuffle.start = System.nanoTime();
}
public static void getEndTime(final String s) {
javaShuffle.end = System.nanoTime();
System.out.println(s + ": " + (javaShuffle.end - javaShuffle.start) + "ns");
}
}
If the value is small, for example 100000 Level, the output is approximately:
first shuffle run time : 85029499ns
second shuffle run time: 80909474ns
third shuffle run time : 71543926ns
fourth shuffle run time: 76520595ns
java shuffle run time : 61027643ns
first shuffle run time : 82326239ns
second shuffle run time: 78575611ns
third shuffle run time : 95009632ns
fourth shuffle run time: 105946897ns
java shuffle run time : 90849302ns
first shuffle run time : 84539840ns
second shuffle run time: 85965575ns
third shuffle run time : 101814998ns
fourth shuffle run time: 113309672ns
java shuffle run time : 35089693ns
first shuffle run time : 87679863ns
second shuffle run time: 79991814ns
third shuffle run time : 73720515ns
fourth shuffle run time: 78353061ns
java shuffle run time : 64146465ns
first shuffle run time : 84314386ns
second shuffle run time: 80074803ns
third shuffle run time : 74001283ns
fourth shuffle run time: 79931321ns
java shuffle run time : 86427540ns
first shuffle run time : 84315523ns
second shuffle run time: 81468386ns
third shuffle run time : 75052284ns
fourth shuffle run time: 79461407ns
java shuffle run time : 66607729ns
The results may vary from run to run, but the basic Java native shuffle is the fastest, followed by the third way. The first method takes the most time.
If it's 10000000, it looks like this:
First shuffle run time: 2115703288ns
Second shuffle run time: 3114045871ns
Third shuffle run time: 4664426798ns
Fourth shuffle run time: 2962686695ns
Java shuffle run time First shuffle run time: 2165398466ns
Second shuffle run time: 3129558913ns
Third shuffle run time: 4147859664ns
Fourth shuffle run time: 2911849942ns
Java shuffle run time First shuffle run time: 2227462247ns
Second shuffle run time: 3279548770ns
Third shuffle run time: 4704344954ns
Fourth shuffle run time: 294 635980ns
Java shuffle run time First shuffle run time: 2200158789ns
Second shuffle run time: 3172666791ns
Third shuffle run time: 4715631517ns
Fourth shuffle run time: 2950817535ns
Java shuffle run time First shuffle run time: 2201124449ns
Second shuffle run time: 3203823874ns
Third shuffle run time: 4179926278ns
Fourth shuffle run time: 2913690411ns
Java shuffle run time First shuffle run time: 2163053190ns
Second shuffle run time: 3073889926ns
Third shuffle run time: 4493831518ns
The fourth shuffle runs time: 2852713887ns
Java shuffle run time : 3773602415 ns
As you can see, the first method is the fastest and the fourth is the slowest. Java's native shuffle isn't ideal either.
When it comes to big data processing, if using Java libraries is inefficient, consider other options.