算法——序列

引言

本文整理了常见的序列相关算法，方便以后查阅。更多相关文章和其他文章均收录于贝贝猫的文章目录。

最大子序列

package bbm.sequence;

/**
 * 基于分治思想的最大子序列，递归地将序列分割成两半，然后比较左半段，右半段，和中间跨越的段，最终取最大的作为结果
 *
 * 时间复杂度: O(n*log(n))
 * 空间复杂度: O(log(n)) 栈
 *
 * @author bbm
 */
public class MaxSubSequence {
    public int maxSubArray(int[] nums) {
        if (nums == null) {
            return 0;
        }
        if (nums.length == 1) {
            return nums[0];
        }
        return findMaxArray(nums, 0, nums.length - 1);
    }

    private int findMaxArray(int[] nums, int low, int max) {
        if (low == max) {
            return nums[low];
        }
        int mid = (low + max) / 2;
        int leftMax = findMaxArray(nums, low, mid);
        int rightMax = findMaxArray(nums, mid + 1, max);
        int midMax = findCrossArray(nums, low, mid, max);
        return Math.max(Math.max(leftMax, rightMax), midMax);
    }

    private int findCrossArray(int[] nums, int low, int mid, int max) {
        int leftMax = Integer.MIN_VALUE;
        int sum = 0;
        for (int x = mid; x >= low; x--) {
            sum += nums[x];
            if (sum > leftMax) {
                leftMax = sum;
            }
        }
        sum = 0;
        int rightMax = Integer.MIN_VALUE;
        for (int x = mid + 1; x <= max; x++) {
            sum += nums[x];
            if (sum > rightMax) {
                rightMax = sum;
            }
        }
        return leftMax + rightMax;
    }
}

测试代码：

package bbm.sequence;

import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.Test;

/**
 * @author bbm
 */
public class MaxSubSequenceTest {
    @Test
    public void testMaxSubSeq() {
        MaxSubSequence maxSubSequence = new MaxSubSequence();
        int result = maxSubSequence.maxSubArray(new int[] {-2, -1, 1 ,1, 4, -3, -4, 3});
        Assertions.assertEquals(6, result);
    }
}

顺序统计量

随机顺序统计量

package bbm.order;

import java.util.HashSet;
import java.util.Set;

/**
 * 顺序统计量，即获取某一数组中第 N 大的数(n=3)，这并不需要对整个数组进行排序，本例中采用了随机快速排序的思想
 * 但是只会对分割出来的两部分中的一个进行递归处理
 *
 * 平均时间复杂度: O(n)
 * 最差时间复杂度: O(n ^ 2)
 *
 * @author bbm
 */
public class RandomOrderStatistic implements OrderStatistic{

    private static final int n = 3;

    @Override
    public int getNLargestNumber(int[] nums) {
        Set<Integer> set = new HashSet<>();
        for (int i = 0; i < nums.length; i++) {
            set.add(nums[i]);
        }
        nums = new int[set.size()];
        int index = 0;
        for (int num : set) {
            nums[index++] = num;
        }
        int smallestN;
        if (nums.length - n + 1 > 0) {
            smallestN = nums.length - n + 1;
        } else {
            smallestN = nums.length;
        }
        return select(nums, 0, nums.length - 1, smallestN);
    }

    private int select(int[] nums, int p, int q, int n) {
        if (p == q) {
            return nums[p];
        }
        int r = partition(nums, p, q);
        int k = r - p + 1;
        if (n == k) {
            return nums[r];
        }
        if (n < k) {
            return select(nums, p, r - 1, n);
        } else {
            return select(nums, r + 1, q, n - k);
        }
    }

    private static int partition(int[] nums, int p, int r) {
        int pivot = nums[r];
        int leftEnd = p - 1;
        for (int rightEnd = p; rightEnd < r; rightEnd++) {
            if (nums[rightEnd] < pivot) {
                leftEnd++;
                int temp = nums[leftEnd];
                nums[leftEnd] = nums[rightEnd];
                nums[rightEnd] = temp;
            }
        }
        leftEnd++;
        int temp = nums[leftEnd];
        nums[leftEnd] = nums[r];
        nums[r] = temp;
        return leftEnd;
    }
}

稳定顺序统计量

package bbm.order;

import java.util.HashSet;
import java.util.Set;

/**
 * 在 {@link RandomOrderStatistic} 中最差时间复杂度到了 O(n^2) 这是快排中也会出现的问题，即便进行了随机化，也是可能出现该问题，
 * 所以在本例中，我们不再选用最后一个数作为主元素，而是我们指定一个特定的元素，该元素需要保持划分的平衡性，这样才能保证时间复杂度收敛与 O(n)
 *
 * 具体的做法是：
 * 将n个元素划分为n/5组
 * 寻找每组的中位数
 * 找出的中位数的中位数x, 这个中位数的性质是 有 1/4 的数小于它并且有 1/4 的数大于它，所以选用它来做主元素比较平衡
 * 使用x作为主元素执行快排的 PARTITION，计算得出 x 为第 k 小的元素
 * 如果目标值 i==k，返回x；如果i<k，在低区调用PARTITION找出第i小的元素；如果i>k，在高区调用PARTITION查找第i-k小的元素
 *
 * 时间复杂度: O(n)
 *
 * @author bbm
 */
public class StableOrderStatistic implements OrderStatistic{

    private static final int n = 3;

    @Override
    public int getNLargestNumber(int[] nums) {
        Set<Integer> set = new HashSet<>();
        for (int i = 0; i < nums.length; i++) {
            set.add(nums[i]);
        }
        nums = new int[set.size()];
        int index = 0;
        for (int num : set) {
            nums[index++] = num;
        }
        int smallestN;
        if (nums.length - n + 1 > 0) {
            smallestN = nums.length - n + 1;
        } else {
            smallestN = nums.length;
        }
        return select(nums, 0, nums.length - 1, smallestN, findMidNumber(nums));
    }

    private int select(int[] nums, int start, int end, int smallestN, int pivotIndex) {
        if (start == end) {
            return nums[start];
        }
        int rightIndex = partition(nums, start, end, pivotIndex);
        int order = rightIndex - start + 1;
        if (smallestN == order) {
            return nums[rightIndex];
        }
        if (smallestN < order) {
            return select(nums, start, rightIndex - 1, smallestN, rightIndex - 1);
        } else {
            return select(nums, rightIndex + 1, end, smallestN - order, rightIndex + 1);
        }
    }

    private int findMidNumber(int[] nums) {
        int size = (int) Math.ceil((double) nums.length / 5);
        int[][] groupedNums = new int[size][5];
        int[] indexes = new int[size];
        for (int i = 0; i < nums.length; i++) {
            int slot = i % groupedNums.length;
            groupedNums[slot][indexes[slot]] = nums[i];
            int tmp = indexes[slot];
            for (int j = indexes[slot] - 1; j >= 0; j--) {
                if (tmp < groupedNums[slot][j]) {
                    groupedNums[slot][j + 1] = groupedNums[slot][j];
                    groupedNums[slot][j] = tmp;
                }
            }
            indexes[slot] += 1;
        }
        int[] midNumbers = new int[size];
        int index = 0;
        for (int i = 0; i < groupedNums.length; i++) {
            if (indexes[i] == 5) {
                midNumbers[index++] = groupedNums[i][3];
            } else {
                midNumbers[index++] = groupedNums[i][indexes[i] / 2];
            }
        }
        int finalMid = select(midNumbers, 0, midNumbers.length - 1, size / 2, 0);
        int finalMidIndex = 0;
        for (int i = 0; i < nums.length; i++) {
            if (nums[i] == finalMid) {
                finalMidIndex = i;
            }
        }
        return finalMidIndex;
    }

    private static int partition(int[] nums, int p, int r, int position) {
        int temp = nums[position];
        nums[position] = nums[r];
        nums[r] = temp;
        int pivot = nums[r];
        int leftEnd = p - 1;
        for (int rightEnd = p; rightEnd < r; rightEnd++) {
            if (nums[rightEnd] < pivot) {
                leftEnd++;
                temp = nums[leftEnd];
                nums[leftEnd] = nums[rightEnd];
                nums[rightEnd] = temp;
            }
        }
        leftEnd++;
        temp = nums[leftEnd];
        nums[leftEnd] = nums[r];
        nums[r] = temp;
        return leftEnd;
    }
}

测试

package bbm.order;

import com.google.common.primitives.Ints;
import java.util.Arrays;
import java.util.Collections;
import java.util.Random;
import java.util.stream.Collectors;
import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.Test;

/**
 * @author bbm
 */
class RandomOrderStatisticTest {
    @Test
    public void testGetNSmallestNumber() {
        doTest(new RandomOrderStatistic());
        doTest(new StableOrderStatistic());
    }

    private void doTest(OrderStatistic orderStatistic) {
        int dataSize = 10000;
        int[] data = new int[dataSize];
        Random rand = new Random(System.currentTimeMillis());
        for (int i = 0; i < dataSize; i++) {
            data[i] = rand.nextInt(dataSize);
        }
        Arrays.sort(data);
        Collections.reverse(Ints.asList(data));
        int rightResult = Ints.asList(data).stream().distinct().collect(Collectors.toList()).get(2);
        int result = orderStatistic.getNLargestNumber(data);
        Assertions.assertEquals(rightResult, result);
    }
}

参考内容

[1] 《算法导论》