递归java程序不返回任何值。在return语句之前放置print语句会打印无限值

Question

//inefficient recursive algorithm to find number of coins needed for a some amount
public class bitcoin {

    public int mincoins;

    public static void main(String[] args) {
        int[] coins = {1, 5, 10, 21, 25};
        int change = 63;
        int min = 0;
        min = MinCoins(coins, change);
        System.out.println(min);
    }

    public static int MinCoins(int[] coins, int change) {
        int mincoins = change;
        int mincoins1 = 10000;

        for (int j = 0; j <= coins.length - 1; j++) {
            if (coins[j] == change) //base case
                return 1;
        }

        //recursive call breaking into two sub problems
        for (int i = 1; i <= (change) / 2; i++) {
            mincoins1 = (MinCoins(coins, i) + MinCoins(coins, (change - i)));

            if (mincoins1 < change)
                mincoins = mincoins1;
        }

        return mincoins;
    }
}

Answer 1

正如您的评论所暗示的，这是一种效率低下的递归算法。但是通过稍微移动代码，至少不会永远循环。相反，它最终找到了一个解决方案（对于我所做的所有测试，它已经找到了最佳的硬币计数）。

您错过的部分是退出失败标准。在这个例子中：

If the coin sum is more than 'change', give up.
(For example, if you have already added 3 coins with value 25 (total 75), there are no
coins that can help you reach 63, so stop trying to find them.)

这个if语句（见下文）处理失败时退出：

if (change - coins[j] >= 0)

public static int MinCoins(int[] coins, int change) {
    int mindepth = 10000;
    for (int j = 0; j < coins.length; j++) {
        if (coins[j] == change) {
            return 1; // end case
        } else if (change - coins[j] >= 0) {
            int depth = 1 + MinCoins(coins, change - coins[j]);
            if (depth >= 0 && depth < mindepth) {
                mindepth = depth;
            }
        }
    }
    return mindepth;
}