加权随机算法随笔

开发一个加权随机算法：
核心主要有两点，一个是概率补偿，一个是即时更新权重占比

    public static <T> Set<T> weightRandomSelect(Map<T, Double> weightMap, int k) {
        double weightSum = 0;
        Set<T> result = new HashSet<>();
        for (Map.Entry<T, Double> entry : weightMap.entrySet()) {
            weightSum += entry.getValue();
        }
        int size = weightMap.size();
        if (size == 0) {
            return result;
        }
        double averageWeight;
        Iterator<Map.Entry<T, Double>> iterator = weightMap.entrySet().iterator();
        int cursor = 0;
        while (iterator.hasNext()) {
            averageWeight = weightSum / size;
            if (averageWeight == 0) {
                result.addAll(randomSelect(iterator, cursor, weightMap.size(), k));
                break;
            }
            Map.Entry<T, Double> entry = iterator.next();
            cursor++;
            if (k >= size) {
                result.add(entry.getKey());
                k--;
                size--;
                weightSum -= entry.getValue();
                continue;
            }
            double value = entry.getValue();
            if (value <= 0) {
                size--;
                weightSum -= entry.getValue();
                continue;
            }
            double p = value / averageWeight;
            if (isSelectAndWeight(size, k, p)) {
                result.add(entry.getKey());
                k--;
            }
            size--;
            weightSum -= entry.getValue();
        }
        return result;
    }

    public static <T> Set<T> randomSelect(Iterator<Map.Entry<T, Double>> iterator, int traversedCount, int size, int k) {
        int count = size - traversedCount;
        Set<T> result = new HashSet<>();
        if (count <= 0) {
            return result;
        }
        while (iterator.hasNext()) {
            Map.Entry<T, Double> next = iterator.next();
            if (isSelect(count, k)) {
                result.add(next.getKey());
                k--;
            }
            count--;
        }
        return result;

    }

    private static boolean isSelect(int size, int needNum) {
        if (size <= needNum) {
            return true;
        }
        return RandomUtil.generateRandomNum(0, size) < needNum;
    }

    private static boolean isSelectAndWeight(int size, int needNum, double p) {
        if (size <= needNum) {
            return true;
        }
        return RandomUtil.generateRandomDouble(0, size) / p < needNum;
    }