"Hope it doesn't become unrated" comments incoming

gl to anyone participating!!!

Hope it doesn't become unrated

Hope, I reach Expert :(

I'm at school when this contest start, so I will truant not be able to participate sadly.

one of the worst timing for me

i will be in school when it starts

Conducting contest on unusual time is better suited if contest is on weekend.

I’m very sorry that I won’t be able to participate in my first Div. 1 round after becoming Master because I have to take a flight.

As a PKUer, I sincerely hope that this contest hosted by Tsinghua University will be a great success.

As you know, I’ve always been a huge fan of Tsinghua University.

GL&HF to everyone!

why the unusual time :(

Another day time contest for chinese codeforcers,don't be unrated please

I actually hope some more events will be held at 14:05UTC+8.

22:35UTC+8 is not too Asia friendly.

Excited for Codeforces Round 1097

Big thanks to the organizers and problem setters for this amazing contest!

Ready to test my skills, learn new tricks, and hopefully keep my code bug-free. See you all in the round!

this timing is very unusual..i suggest unusual timing is on weekend is good not week day..also i am not excited tbh.

i won't be able to participate in this contest due to school :((

Bad timing.

btw, Tsinghua round again :(

Oh, Wednesday, I have to go to school :(

bro its at 2 am what

The perfect time. I want all the rounds to be in this time :))

plz no problem leak thx

Codeforces said; unusual timing+weekday+rated

we said: sleep is optional T_T.

Good Luck everyone...

The Zhili Cup offline contest is held in Beijing exactly one day before APIO.

As an APIO 2026 contestant, I originally planned to go to Beijing for APIO and take part in the offline Zhili Cup at the same time.

However, the organizers recently revoked the eligibility of all middle school students to participate in the Zhili Cup, so I’ve ended up with an extra Div.1 contest before APIO instead :)

edit:

Have lost my chance to participate offline

As a tester, wish you all perform at your best and enjoy the contest.

idk...

I hope to become purple this round pls.

qp

May 6 is my only day(from May 1 to May 12) when I must go to school!!! (which means I can't participate in this contest)

Will be at college at the given time. cant skip because exams are goin on :(

What a Chinese Round but not Chinese-friendly time!

The perfect time. I want all the rounds to be in this time :))

I hope it goes well.

Hope it won't be unrated

The one and only KAN for his invaluable support in hosting the round

Bro rlly has the nuts to disrespect a "Headquarters" member

1250 D and 2250 E in Div.2, interesting:)

Hopefully i don't get cooked this time!

May green elo find me please :)

As a tester, good luck to cute MaxBlazeIceInk and all other participants!

As a tester, I wish everyone satisfactory results!

Is it rated?

looking at the score distribution, ig i can do 3 to 4 problems fs

I hope to become specialist again :)

Is it still rated?

Small font for stories should become an industry standard!

Fortunately, I don't have school today.

dirtforces

Implementationforces.

Shorter statements -> Harder problems?

Fuck problem Div.2D,I optimized my code's time complexity it hit memory after that I optimized time and it hit time. I've tried ordered_multiset,segtree with compression,Dynamic segtree but none of them passed

Check out my submissions

Why is the TL for problem div1-C so strict?

Only manage to come up $$$O(N^3)$$$ sol to D:(

this round was total bs, every time tsinghua sets the contest either the problems are way too hard or some unrated mess happens, this time the anti-ai statements got completely messed up in codeforces better's translation (like that hidden html text got translated) and that's why i got 6 extra wrong submissions on a and b, plus problem d was insane, so disappointing and frustrating :(

div2-D is supposed to be fenwick tree counting + compression right? My solution is like $$$O(N^2 * log(4e6))$$$. Should it be some kind of squeezing like global declaration to avoid overhead on allocating big vector?

Ahh, again BIT >> ordered_set. Why I read once about ordered_set....

wish time limits where less strict, got screwed by constant optimization on both B and C

in div2D, my (n^2) * log(n^2) sol using pbds ordered-set TLEd (6seconds),
very sad.

rewrote with merge sort

Great round, thanks to authors!

speedforces

How to do problem E? :(

guessforces

Submission

The above submission passed all pretests, but outputs the incorrect answer for the testcase generated by:

  cout << 1 << endl;
  cout << 2000 << endl;
  for (int i = 2000; i >= 1; i--) {
    cout << i << " ";
  }
  cout<<endl;
  for (int i = 1; i <= 2000; i++) {
    cout << i << " ";
  }
  cout << endl;

Obtained answer: -36131418 Correct answer: 962112935

sorry for the wrong words. :(

Ok I don't know what is codeforces' policy on this but i resubmitted my TLE code for C after the contest and it got accepted. Is it possible to get a rejudge? This would greatly impact my rating.

During contest: https://mirror.codeforces.com/contest/2223/submission/373691965

After contest: https://mirror.codeforces.com/contest/2223/submission/373695945

The time limit for D1C is incredibly tight. My $$$O(n \log \log V + q \log^2 n)$$$ solution failed to pass (373689679), yet an optimized brute-force approach managed to get through (373668849). :(

I literally failed solving C(Div 1). Pure DFS works fantastic instead went with the binary lifting and other stuff damn ! I should have solved C. My bad !

Problem Div1 C. Zhily and Signpost technically violates input format. Problem statement says "The second line contains ...", whereas in test 2 apparently every test case has two empty lines between its "first" and "second" lines. I'm not suggesting any particular action, but SSerxhs, I'd like to ask whether this is considered acceptable on codeforces and hence something a participant should be prepared for.

I found a very strange issue in problem C(div 1.).

In my original submission, I defined the function like this:

long long lcm(long long a, long long b) {
    if (a == -1) return -1;
    long long g = gcd(a, b);
    if (a / g > INF / b) return -1;
    return a / g * b;
}

and called it in the DFS:

long long nm = lcm(m, sz);

This version got TLE on test 18.

However, after only renaming the function from lcm to LCM / get_lcm, without changing the algorithm or any other logic:

long long LCM(long long a, long long b) {
    if (a == -1) return -1;
    long long g = gcd(a, b);
    if (a / g > INF / b) return -1;
    return a / g * b;
}

the code passed.

I guess this may be related to the standard library std::lcm, because I used:

#include <bits/stdc++.h>
using namespace std;

So the name lcm may cause worse overload resolution / inlining decisions, especially since this function is called many times in a tight loop.

Has anyone else encountered this? Why does naming the function lcm cause TLE, while LCM gets accepted?

Update: fallleaves01 found the real reason.

It was not just a naming / inlining issue. The problem was overload resolution.

My own function was:

long long lcm(long long a, long long b) {
    if (a == -1) return -1;
    long long g = gcd(a, b);
    if (a / g > INF / b) return -1;
    return a / g * b;
}

But I called it like this:

int sz = adj[x].size();
long long nm = lcm(m, sz);

So the argument types were:

lcm(long long, int)

Because I had:

#include <bits/stdc++.h>
using namespace std;

std::lcm was also visible. Then overload resolution preferred std::lcm<long long, int> over my lcm(long long, long long), because std::lcm matches both arguments directly, while my function needs to convert the second argument from int to long long.

As a result, my overflow-limited lcm function was not called. Instead, std::lcm was called, so my intended return -1 behavior on overflow was lost. This caused the pruning logic to fail and eventually led to TLE.

Renaming the function to LCM / get_lcm, or casting sz to long long, fixes it:

long long nm = get_lcm(m, sz);
// or
long long nm = lcm(m, (long long)sz);
// or
long long nm = ::lcm(m, sz);

So the real bug was caused by the combination of using namespace std, a function named lcm, and passing sz as an int.

D<B<C（

I spent the entire contest thinking about Problem B. I really wanted to move on to the later problems, but I just kept hoping I could get Problem B accepted. In the end, I only managed to pass Problem A.