国产探花免费观看_亚洲丰满少妇自慰呻吟_97日韩有码在线_资源在线日韩欧美_一区二区精品毛片,辰东完美世界有声小说,欢乐颂第一季,yy玄幻小说排行榜完本

首頁 > 編程 > Java > 正文

Java 位圖法排序的使用方法

2019-11-26 16:11:50
字體:
供稿:網(wǎng)友

java JDK里面容器類的排序算法使用的主要是插入排序和歸并排序,可能不同版本的實現(xiàn)有所不同,關(guān)鍵代碼如下:

復(fù)制代碼 代碼如下:

/**
     * Performs a sort on the section of the array between the given indices
     * using a mergesort with exponential search algorithm (in which the merge
     * is performed by exponential search). n*log(n) performance is guaranteed
     * and in the average case it will be faster then any mergesort in which the
     * merge is performed by linear search.
     *
     * @param in -
     *            the array for sorting.
     * @param out -
     *            the result, sorted array.
     * @param start
     *            the start index
     * @param end
     *            the end index + 1
     */
    @SuppressWarnings("unchecked")
    private static void mergeSort(Object[] in, Object[] out, int start,
            int end) {
        int len = end - start;
        // use insertion sort for small arrays
        if (len <= SIMPLE_LENGTH) {
            for (int i = start + 1; i < end; i++) {
                Comparable<Object> current = (Comparable<Object>) out[i];
                Object prev = out[i - 1];
                if (current.compareTo(prev) < 0) {
                    int j = i;
                    do {
                        out[j--] = prev;
                    } while (j > start
                            && current.compareTo(prev = out[j - 1]) < 0);
                    out[j] = current;
                }
            }
            return;
        }
        int med = (end + start) >>> 1;
        mergeSort(out, in, start, med);
        mergeSort(out, in, med, end);

        // merging

        // if arrays are already sorted - no merge
        if (((Comparable<Object>) in[med - 1]).compareTo(in[med]) <= 0) {
            System.arraycopy(in, start, out, start, len);
            return;
        }
        int r = med, i = start;

        // use merging with exponential search
        do {
            Comparable<Object> fromVal = (Comparable<Object>) in[start];
            Comparable<Object> rVal = (Comparable<Object>) in[r];
            if (fromVal.compareTo(rVal) <= 0) {
                int l_1 = find(in, rVal, -1, start + 1, med - 1);
                int toCopy = l_1 - start + 1;
                System.arraycopy(in, start, out, i, toCopy);
                i += toCopy;
                out[i++] = rVal;
                r++;
                start = l_1 + 1;
            } else {
                int r_1 = find(in, fromVal, 0, r + 1, end - 1);
                int toCopy = r_1 - r + 1;
                System.arraycopy(in, r, out, i, toCopy);
                i += toCopy;
                out[i++] = fromVal;
                start++;
                r = r_1 + 1;
            }
        } while ((end - r) > 0 && (med - start) > 0);

        // copy rest of array
        if ((end - r) <= 0) {
            System.arraycopy(in, start, out, i, med - start);
        } else {
            System.arraycopy(in, r, out, i, end - r);
        }
    }


看到編程珠璣上有一個很有趣的排序算法-位圖法其思想是用1位來表示[0~n-1]中的整數(shù)是否存在。1表示存在,0表示不存在。即將正整數(shù)映射到bit集合中,每一個bit代表其映射的正整數(shù)是否存在。

比如{1,2,3,5,8,13}使用下列集合表示:

  0 1 1 1 0 1 0 0 1 0 0 0 0 1 0 0 0 0 0 0

偽代碼如下:

for (i  in  [0~n-1])  bit[i] = 0;
for(i  in [0~n-1])
  if (i in input file)     
    bit[i] = 1

for(i  in [0~n-1])
  if(bit[i] == 1) 
    output i

用java 代碼嘗試下,效率果然不錯:

復(fù)制代碼 代碼如下:

public class javaUniqueSort {
    public static int[] temp = new int[1000001];
    public static List<Integer> tempList = new ArrayList<Integer>();
    public static int count;

    public static void main(final String[] args) {
        List<Integer> firstNum = new ArrayList<Integer>();
        List<Integer> secondNum = new ArrayList<Integer>();

        for (int i = 1; i <= 1000000; i++) {
            firstNum.add(i);
            secondNum.add(i);
        }

        Collections.shuffle(firstNum);
        Collections.shuffle(secondNum);

        getStartTime();
        Collections.sort(firstNum);
        getEndTime("java sort run time  ");

        getStartTime();
        secondNum = uniqueSort(secondNum);
        getEndTime("uniqueSort run time ");

    }

    public static List<Integer> uniqueSort(final List<Integer> uniqueList) {
        javaUniqueSort.tempList.clear();
        for (int i = 0; i < javaUniqueSort.temp.length; i++) {
            javaUniqueSort.temp[i] = 0;
        }
        for (int i = 0; i < uniqueList.size(); i++) {
            javaUniqueSort.temp[uniqueList.get(i)] = 1;
        }
        for (int i = 0; i < javaUniqueSort.temp.length; i++) {
            if (javaUniqueSort.temp[i] == 1) {
                javaUniqueSort.tempList.add(i);
            }
        }

        return javaUniqueSort.tempList;
    }

    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");
    }
}

運行時間:

java sort run time  : 1257737334ns
uniqueSort run time : 170228290ns
java sort run time  : 1202749828ns
uniqueSort run time : 169327770ns


如果有重復(fù)數(shù)據(jù),可以修改下:
復(fù)制代碼 代碼如下:

public class javaDuplicateSort {
    public static List<Integer> tempList = new ArrayList<Integer>();
    public static int count;

    public static void main(final String[] args) {
        Random random = new Random();
        List<Integer> firstNum = new ArrayList<Integer>();
        List<Integer> secondNum = new ArrayList<Integer>();

        for (int i = 1; i <= 100000; i++) {
            firstNum.add(i);
            secondNum.add(i);
            firstNum.add(random.nextInt(i + 1));
            secondNum.add(random.nextInt(i + 1));
        }
        Collections.shuffle(firstNum);
        Collections.shuffle(secondNum);

        getStartTime();
        Collections.sort(firstNum);
        getEndTime("java sort run time  ");

        getStartTime();
        secondNum = uniqueSort(secondNum);
        getEndTime("uniqueSort run time ");

    }

    public static List<Integer> uniqueSort(final List<Integer> uniqueList) {
        javaDuplicateSort.tempList.clear();
        int[] temp = new int[200002];
        for (int i = 0; i < temp.length; i++) {
            temp[i] = 0;
        }
        for (int i = 0; i < uniqueList.size(); i++) {
            temp[uniqueList.get(i)]++;
        }
        for (int i = 0; i < temp.length; i++) {
            for (int j = temp[i]; j > 0; j--) {
                javaDuplicateSort.tempList.add(i);
            }
        }

        return javaDuplicateSort.tempList;
    }

    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");
    }
}


這種算法還是有很明顯的局限性的,比如說要知道數(shù)據(jù)中最大的數(shù)值,更重要的是數(shù)據(jù)的疏密程度,比如說最大值為1000000而要數(shù)組大小只有100,那么效率會下降的非常明顯。。。。。但是,使用位圖法進行排序,確實讓人眼前一亮。位圖法通常是用來存儲數(shù)據(jù),判斷某個數(shù)據(jù)存不存在或者判斷數(shù)組是否存在重復(fù) 。

發(fā)表評論 共有條評論
用戶名: 密碼:
驗證碼: 匿名發(fā)表
主站蜘蛛池模板: 泸溪县| 融水| 祁东县| 莱阳市| 利津县| 金门县| 林芝县| 金山区| 湄潭县| 北安市| 博白县| 察隅县| 红桥区| 卢氏县| 犍为县| 贵定县| 望谟县| 师宗县| 蓬安县| 普陀区| 辽中县| 临猗县| 偏关县| 广汉市| 衡南县| 托克逊县| 米林县| 阿荣旗| 永德县| 平乡县| 九寨沟县| 徐水县| 沙河市| 伊川县| 兴和县| 丽江市| 安阳县| 安图县| 时尚| 什邡市| 东港市|