### zinoviev's blog

By zinoviev, history, 4 weeks ago, translation, ,

Hello everyone. I am stuck with upsolving problem B from SEERC-2019. Problem statement

Even after reading editorials and realizing that my solution seems correct, I keep getting WA#7 on codeforces. I am doing dp, like described in editorial, and sorting all quest before by $x_i$. I am using $dp[j][k]$ array and $tmp[j][k]$ array on every iteration because I need to use every quest at most one time. So I am doing optimization on $tmp$ array, and then copy its contents back to $dp$. Then answer should be at $dp[s_1][s_2]$.

Ill be very glad if someone more experienced than me could help me find the test, where my solution fails or maybe just give me some hints.

My code:

struct q {
ll e1, e2, t1, t2;
};

bool cmp(q q1, q q2) {
return q1.e1 < q2.e1;
}

ll tmp[501][501];
ll dp[501][501];
ll n, s1, s2;

#define chkmin(x, y) (x) == -1 ? (x) = (y) : (x) = min((x), (y))

void solve() {
cin >> n >> s1 >> s2;
vector<q> v(n);
for (int i = 0; i < n; i++)
cin >> v[i].e1 >> v[i].t1 >> v[i].e2 >> v[i].t2;
sort(all(v), cmp);
for (int j = 0; j <= s1; j++)
for (int k = 0; k <= s2; k++)
dp[j][k] = j == 0 && k == 0 ? 0 : -1;

for (int i = 0; i < n; i++) {
for (int j = 0; j <= s1; j++)
for (int k = 0; k <= s2; k++)
tmp[j][k] = dp[j][k];

for (int j = 0; j <= s1; j++)
for (int k = 0; k <= s2; k++) {
if (dp[j][k] == -1)
continue;
if (j != s1) {
if (j + v[i].e1 <= s1)
chkmin(tmp[j + v[i].e1][k], dp[j][k] + v[i].t1);
else
chkmin(tmp[s1][min(k + (j + v[i].e1 - s1), s2)], dp[j][k] + v[i].t1);
}
chkmin(tmp[j][min(k + v[i].e2, s2)], dp[j][k] + v[i].t2);
}

for (int j = 0; j <= s1; j++)
for (int k = 0; k <= s2; k++)
dp[j][k] = tmp[j][k];
}
cout << dp[s1][s2] << endl;
}
`

Thank you!

UPD: Fixed bug with $j = s_1$, now it is WA#7 UPD: It was very stupid mistake in comparator. I got AC, code above is correct.

• 0

 » 4 weeks ago, # |   0 Auto comment: topic has been updated by zinoviev (previous revision, new revision, compare).
 » 4 weeks ago, # |   +3 If $j = s_1$, you shouldn't make a transition in your solution (just guessing).
•  » » 4 weeks ago, # ^ |   0 Thank you, I thinked about it and got it. My previous solution tried to use quest overflow even if you reached $s_1$ experience. I fixed it and now I have WA#7. It is better tho
•  » » » 4 weeks ago, # ^ |   +3 Are you sure that your comparator is correct? Shouldn't it be $q_1.e_1 < q_2.e_1$?
•  » » » » 4 weeks ago, # ^ |   0 Oh my.. It was very stupid and unexpected. Thank you, I managed to got AC. Thank you again
 » 4 weeks ago, # |   0 Auto comment: topic has been updated by zinoviev (previous revision, new revision, compare).