背包问题之完全背包问题

问题描述：有n个物品，第 i 个物品的重量与价值分别为 $w[i]$ 与 $v[i]$。背包容量为 V，试问在每个物品有无限个（物品必须保持完整）的情况下，如何让背包装入的物品具有更大的价值总和。现有数据如下：

w = [2,3,4,5];
v = [3,4,5,6];
V = 10;

解题思路：令 $dp[i][j]$ 表示从编号 1~i 的物品中挑选任意数量的任意物品放入容量为 j 的背包中得到的最大价值，那么有 d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j − k ∗ w [ i ] ] + k ∗ v [ i ] ∣ 0 ≤ k & k ∗ w [ i ] ≤ j } dp[i][j]=max\{dp[i-1][j-k*w[i]]+k*v[i]|0 \le k\&k*w[i] \le j\} dp[i][j]=max{dp[i−1][j−k∗w[i]]+k∗v[i]∣0≤k&k∗w[i]≤j}。根据题意，在背包容量不变的情况下，若增加一件物品仍然无法改变背包内的物品，那么最大价值仍然为 $dp[i-1][j]$，即 $dp[i][j]=dp[i-1][j]$。因此有 d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , d p [ i − 1 ] [ j − k ∗ w [ i ] ] + k ∗ v [ i ] ∣ 0 ≤ k & k ∗ w [ i ] ≤ j } dp[i][j]=max\{dp[i-1][j],dp[i-1][j-k*w[i]]+k*v[i]|0 \le k \&k*w[i] \le j\} dp[i][j]=max{dp[i−1][j],dp[i−1][j−k∗w[i]]+k∗v[i]∣0≤k&k∗w[i]≤j}。

    public int knapsackProblem(int[] w, int[] v, int cap) {
        int[][] dp = new int[w.length + 1][cap + 1];
        for (int i = 1; i <= w.length; i++) {
            for (int j = cap; j > 0; j--) {
                dp[i][j] = dp[i - 1][j];
                for (int k = 0; k * w[i - 1] <= j; k++) {
                    dp[i][j] = Math.max(dp[i][j], dp[i - 1][j - k * w[i - 1]] + k * v[i - 1]);
                }
            }
        }
        return dp[w.length][cap];
    }

由以上代码可知，时间复杂度为 $O(n{V}^2)$。

时间优化：

d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , d p [ i − 1 ] [ j − k ∗ w [ i ] ] + k ∗ v [ i ] ∣ 0 ≤ k & k ∗ w [ i ] ≤ j } dp[i][j]=max\{dp[i-1][j],dp[i-1][j-k*w[i]]+k*v[i]|0 \le k\&k*w[i] \le j\} dp[i][j]=max{dp[i−1][j],dp[i−1][j−k∗w[i]]+k∗v[i]∣0≤k&k∗w[i]≤j}

d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , m a x { d p [ i − 1 ] [ ( j − w [ i ] ) − ( k − 1 ) ∗ w [ i ] ] + ( k − 1 ) ∗ v [ i ] ∣ 0 ≤ ( k − 1 ) ∗ w [ i ] ≤ j } + v [ i ] } dp[i][j]=max\{dp[i-1][j], max\{dp[i-1][(j-w[i])-(k-1)*w[i]]+(k-1)*v[i]|0 \le (k-1)*w[i] \le j\} + v[i]\} dp[i][j]=max{dp[i−1][j],max{dp[i−1][(j−w[i])−(k−1)∗w[i]]+(k−1)∗v[i]∣0≤(k−1)∗w[i]≤j}+v[i]}

d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , m a x { d p [ i − 1 ] [ ( j − w [ i ] ) − k ∗ w [ i ] ] + k ∗ v [ i ] ∣ 0 ≤ k & ( k + 1 ) ∗ w [ i ] ≤ j } + v [ i ] } dp[i][j]=max\{dp[i-1][j], max\{dp[i-1][(j-w[i])-k*w[i]]+k*v[i]|0 \le k \& (k+1)*w[i] \le j\}+v[i]\} dp[i][j]=max{dp[i−1][j],max{dp[i−1][(j−w[i])−k∗w[i]]+k∗v[i]∣0≤k&(k+1)∗w[i]≤j}+v[i]}

纵观上述表达式的第二项，有

d p [ i ] [ j − w [ i ] ] = m a x { d p [ i − 1 ] [ ( j − w [ i ] ) − k ∗ w [ i ] ] + k ∗ v [ i ] ∣ 0 ≤ k & ( k + 1 ) ∗ v [ i ] ≤ j } dp[i][j-w[i]]=max\{dp[i-1][(j-w[i])-k*w[i]]+k*v[i]|0 \le k \& (k+1)*v[i] \le j\} dp[i][j−w[i]]=max{dp[i−1][(j−w[i])−k∗w[i]]+k∗v[i]∣0≤k&(k+1)∗v[i]≤j}

进而 d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , d p [ i ] [ j − w [ i ] ] + v [ i ] } dp[i][j]=max\{dp[i-1][j], dp[i][j-w[i]]+v[i]\} dp[i][j]=max{dp[i−1][j],dp[i][j−w[i]]+v[i]}

空间优化：

d p [ i ] [ j ] = m a x { d p [ i − 1 ] [ j ] , d p [ i ] [ j − w [ i ] ] + v [ i ] } dp[i][j]=max\{dp[i-1][j], dp[i][j-w[i]]+v[i]\} dp[i][j]=max{dp[i−1][j],dp[i][j−w[i]]+v[i]} => d p [ j ] = m a x { d p [ j ] , d p [ j − w [ i ] ] + v [ i ] } , j = 1 , 2 , . . . c a p dp[j]=max\{dp[j], dp[j-w[i]]+v[i]\}, j=1,2,...cap dp[j]=max{dp[j],dp[j−w[i]]+v[i]},j=1,2,...cap

    public int knapsackProblem(int[] w, int[] v, int cap) {
        int[] dp = new int[cap + 1];
        for (int i = 0; i < w.length; i++) {
            for (int j = 1; j <= cap; j++) {
                if (w[i] <= j) {
                    dp[j] = Math.max(dp[j], dp[j - w[i]] + v[i]);
                }
            }
        }
        return dp[cap];
    }

时间复杂度为 $O(nV)$。

---

上一篇：背包问题之0-1背包问题
下一篇：背包问题之多重背包问题