如何使用 LinkedHashMap 获取子图？

Question

目前，我使用TreeMap 存储一些x 和y 坐标，但与ArrayList 或HashMap 相比，迭代速度非常慢。我使用它是因为我需要 subMap() 方法，因此即使不存在确切的 X 值（键），我也可以获得确定范围内的 X 值。

LinkedHashMap 的速度与HashMap 几乎相同，我可以按插入顺序迭代键（我需要插入顺序或按比较器的顺序，因为它在 TreeMap 中完成）但我没有 @ 987654327@ 方法。在 TreeMap 中，我可以非常快速地生成子图。

是否有任何数据结构或某种方法可以比 TreeMap 更快地存储有序值（通过插入顺序或比较器），即使确切值不在地图中，也可以在某个范围内获取子图？我的意思是，也许我想要 2 到 25 之间的值，但 2 不存在，最接近的是 3，因此它将返回一个从 3 到 25 的子图。或者以某种方式将此功能添加到 LinkedHashMap？

Answer 1

听起来您需要一个 TreeMap，它的迭代速度并不比 LinkedHashMap 慢多少，并且可以满足您的需求。由于 HashMap 是无序的，所以 subMap 没有意义。

Answer 2

今天我终于得到了我的问题的答案。经过几次测试HashMap、LinkedHashMap 和TreeMap 比ArrayList 慢得多，我想用它们来创建subMaps()。所以我创建了一个扩展ArrayList 的新类，它给了我非常好的性能，并且在this 答案的帮助下，我创建了通过值而不是索引获取子列表的快速方法。这是完整的课程：

/**
 * The purpose of this class is to be a faster replacement to a {@link java.util.TreeMap} with
 *  the ability to get sublist containing a range of x values. ArrayList access time is O(1) while
 *  {@link java.util.TreeMap} is O(log(n)). When large data is handled the impact on performance is
 *  noticeable.
 */
public class XYDataset extends ArrayList<PointValue> {

    private final float COMPARISON_THRESHOLD = 0.01f;

    final Comparator<PointValue> comparator = new Comparator<PointValue>() {
        @Override
        public int compare(PointValue lhs, PointValue rhs) {
            if (Math.abs(lhs.getX() - rhs.getX()) < COMPARISON_THRESHOLD) return 0;
            return lhs.getX() < rhs.getX() ? -1 : 1;
        }
    };

    public XYDataset(int capacity) {
        super(capacity);
    }

    public XYDataset() {
    }

    public XYDataset(Collection<? extends PointValue> collection) {
        super(collection);
    }

    @Override
    public List<PointValue> subList(int start, int end) {
        return super.subList(start, end);
    }

    /**
     * Generate a sublist containing the range of x values passed
     * @param x1 lower x value
     * @param x2 upper x value
     * @return sublist containing x values from x1 to x2
     */
    public List<PointValue> subList(float x1, float x2){
        /**
         * Collections.binarySearch() returns the index of the search key, if it is contained in the list;
         *  otherwise it returns (-(insertion point) - 1).
         * The insertion point is defined as the point at which the key would be inserted into the list:
         *  the index of the first element greater than the key, or list.size() if all elements in the list
         *  are less than the specified key. Note that this guarantees that the return value will be >= 0 if
         *  and only if the key is found.
         */
        int n1 = Collections.binarySearch(this, new PointValue(x1, 0), comparator);
        int n2 = Collections.binarySearch(this, new PointValue(x2, 0), comparator);

        /**
         * Example, we assume the list is sorted. Based on (https://stackoverflow.com/questions/19198586/search-sorted-listlong-for-closest-and-less-than)
         *
         * long X = 500;
         * List<Long> foo = new Arraylist<>();
         * foo.add(450L);
         * foo.add(451L);
         * foo.add(499L);
         * foo.add(501L);
         * foo.add(550L);
         *
         * If we search for something that isn't in the list you can work backward from the return value
         *  to the index you want. If you search for 500 in your example list, the algorithm would return (-3 - 1) = -4.
         * Thus, you can add 1 to get back to the insertion point (-3), and then multiply by -1 and subtract 1 to get
         *  the index BEFORE the first element GREATER than the one you searched for, which will either be an index that
         *  meets your 2 criteria OR -1 if all elements in the list are greater than the one you searched for.
         */
        if(n1 < 0) n1 = -n1-1;
        if(n2 < 0) n2 = -n2-1;

        return this.subList(n1, n2);
    }
}

PointValue 只是一个包含 x 和 y 坐标的类。所以现在我只需调用 subList() 传递我想要的 x 坐标范围。在我的情况下，插入顺序也已排序，这对于使用 Collections.binarySearch()

很重要