UVA - 323 Jury Compromise——01背包

第二遍做

因为问题有后效性， 所以当前状态不能由前面的状态转移而来，应该用这个状态更新后面的状态

#include <cstdio>
#include <cstring>
#include <iostream>
#include <algorithm>
#include <vector>
using namespace std;
const int maxn = 205;
int flag = 0, n, m, d[maxn], p[maxn], sum[maxn], sub[maxn], dp[25][805];
vector<int> path[25][800];
int main() {
    while (~scanf("%d %d", &n, &m) && (n+m)) {
        for (int i = 1; i <= n; i++) {
            scanf("%d %d", &d[i], &p[i]);
            sum[i] = d[i] + p[i];
            sub[i] = d[i] - p[i];
        }
        int t = m * 20, v = m * 40;
        for (int i = 0; i <= m; i++) {
            for (int j = 0; j <= v; j++) {
                dp[i][j] = -1;
                path[i][j].clear();
            }
        }
        dp[0][t] = 0;
        for (int k = 1; k <= n; k++) {
            for (int i = m; i >= 0; i--) {
                for (int j = 0; j <= v; j++) {
                    if (dp[i][j] < 0 || j + sub[k] < 0 || j + sub[k] > v) continue;
                    if (dp[i + 1][j + sub[k]] < dp[i][j] + sum[k]) {
                        dp[i + 1][j + sub[k]] = dp[i][j] + sum[k];
                        path[i + 1][j + sub[k]] = path[i][j];
                        path[i + 1][j + sub[k]].push_back(k);
                    }
                }
            }
        }
        int pos = 0;
        while (dp[m][t+pos] < 0 && dp[m][t-pos] < 0) pos++;
        int state = (dp[m][t+pos] > dp[m][t-pos]) ? t+pos : t-pos;
        int sumd = 0, sump = 0;
        for (int i = 0; i < path[m][state].size(); i++) {
            int index = path[m][state][i];
            sumd += d[index], sump += p[index];
        }
        printf("Jury #%d\nBest jury has value %d for prosecution and value %d for defence:\n", ++flag, sumd, sump);
        for (int i = 0; i < path[m][state].size(); i++) {
            printf(" %d", path[m][state][i]);
        }
        printf("\n\n");
    }
    return 0;
}