Блог пользователя zap2727

Автор zap2727, история, 2 часа назад, По-английски

Why I think the problems are easier after Sept. begins? Or just my skill grew?

In Codeforces Round 1124 (Div.2), I might found it to easy the problems A-D; In near 2 Atcode Beginner Contest, I solve five problems, one prob. more than summer, and my spring ABCs also got five.

Who can tell me please?

Полный текст и комментарии »

  • Проголосовать: нравится
  • 0
  • Проголосовать: не нравится

Автор zap2727, история, 2 месяца назад, По-английски

I'm very happy that i got -9 contrib, please give me more downvote so i can got enough anger to destroy the world!!!

If you give me an upvote, please comment and i will be so angry that give your every blog an upvote!!!

Полный текст и комментарии »

  • Проголосовать: нравится
  • +8
  • Проголосовать: не нравится

Автор zap2727, история, 2 месяца назад, По-английски

For Chinese Version, click this

Special thanks to Codeforces for the evaluation resources!

To avoid any suspicion of cheating, I used problems I have already ACed for testing.

I don't think Doubao can pass the second round (

Each AI gets one revival chance (unless it does nothing or cannot be used for free)

Round 1

Let's start with an easy *800 problem!

CF2248A

It's a super simple greedy problem, both players obviously delete from the very front.

AI/Agent Model Status No.
*chatgpt Accepted 1
*claude Accepted 2
*Grok X. No Response. 3 ‌Out
*deepseek Accepted 4
*perplexity X. Wrong Answer on test 1 5 ‌Restart
Gemini 3.5 Flash-Lite Accepted 6
Qwen 3.7 Accepted 7
Doubao Accepted 8
Baidu Wenxin Yiyan Accepted 9
DeepSeek V4 Pro Accepted 10
DeepSeek R1 X. None. 11 ‌Restart
Moonshot X. The same as Kimi. It will appear as kimi. ‌Out
Kimi K2.6 Accepted 12
Kimi K3 X. Busy. You can't use it anytime 13 ‌Out
Kimi K3 Cluster Need Money 14 ‌Out
iFlytek Spark Accepted 15
AstronClaw Compile error for just copy, Accepted after edit code 16

Analysis: Perplexity gave a perfectly logical explanation, but the submitted code ended up wrong. DeepSeek R1 crashed mid-thinking and gave no answer. AstronClaw had an invalid output format.

Round 2

Let's move on to a trickier *800 problem.

CF2250A

Extremely tricky! Let's see if the AI is careful enough. I myself got it wrong many times.

AI/Agent Model Status No. Revival Allowed
*chatgpt Wrong Answer on test 1 1 ‌Restart Yes
*claude Wrong answer on test 2 2 ‌Restart Yes
*deepseek Compile Error 4 ‌Restart Yes
*perplexity Wrong Answer on test 1 5 ‌Out No
Gemini 3.5 Flash-Lite Wrong Answer on test 1 6 ‌Restart Yes
Qwen 3.7 Accepted! 7 Yes
Doubao Wrong Answer on test 1 8 ‌Restart Yes
Baidu Wenxin Yiyan Accepted! 9 Yes
DeepSeek V4 Pro Wrong Answer on test 1 10 ‌Restart Yes
DeepSeek R1 * Accepted! 11 No
Kimi K2.6 Wrong Answer on test 1 12 ‌Restart Yes
iFlytek Spark Accepted! 15 Yes
AstronClaw Cannot_Use 16 ‌Out Yes

Analysis: So far domestic Chinese AIs seem to take the lead. DeepSeek R1 works extremely well, but it has to be used online otherwise it crashes. Qwen and iFlytek Spark perform nicely, while other AIs fail to notice the details and require carefully crafted extra prompts.

Sorry it's a "Chinese-style" problem with tons of tiny details, but it's a great test anyway — everyone can go try it out themselves!

Round 3

*900 Math Problem CF2238B

AI/Agent Model Status No. Revival Allowed
*chatgpt Wrong Answer on test 1 1 ‌Out No
*claude Accepted 2 No
*deepseek Time limit exceeded on test 2 4 ‌Out No
Gemini 3.5 Flash-Lite Accepted 6 No
Qwen 3.7 Accepted 7 Yes
Doubao Wrong Answer on test 1 8 ‌Out No
Baidu Wenxin Yiyan Accepted 9 Yes
DeepSeek V4 Pro Wrong Answer on test 1 10 ‌Out No
DeepSeek R1 Accepted 11 No
Kimi K2.6 Accepted 12 No
iFlytek Spark X. None. 15 ‌Restart Yes

Now the top 8 top 7 are selected, the finalists list:

No. Model Name
2 Claude
6 Gemini 3.5 Flash-Lite
7 Qwen 3.7
9 Baidu Wenxin 5.1
11 DeepSeek R1
12 Kimi K2.6
15 iFlytek Spark

Now the Challenge Round

Challenge Problem: CF689D

An RMQ + Binary Search problem. *2100

AI/Agent Model Status No.
claude Time Limit Exceeded on test 7 2
Gemini 3.5 Flash-Lite Wrong Answer on test 1 6
Qwen 3.7 Wrong Answer on test 4 7
Baidu Wenxin 5.1 Compile Error 9
DeepSeek R1 X. None. Unsolved. 11
Kimi K2.6 Accepted! Got a Restart Chance in Knockout Round! 12
iFlytek Spark X. Over 15 minutes no result 15

Полный текст и комментарии »

  • Проголосовать: нравится
  • +1
  • Проголосовать: не нравится

Автор zap2727, история, 2 месяца назад, По-английски

For Chinese Version, click this

Special thanks to Codeforces for the evaluation resources!

To avoid any suspicion of cheating, I used problems I have already ACed for testing.

I don't think Doubao can pass the second round (

Each AI gets one revival chance (unless it does nothing or cannot be used for free)

Round 1

Let's start with an easy *800 problem!

CF2248A

It's a super simple greedy problem, both players obviously delete from the very front.

AI/Agent Model Status No.
*chatgpt Accepted 1
*claude Accepted 2
*Grok X. No Response. 3 ‌Out
*deepseek Accepted 4
*perplexity X. Wrong Answer on test 1 5 ‌Restart
Gemini 3.5 Flash-Lite Accepted 6
Qwen 3.7 Accepted 7
Doubao Accepted 8
Baidu Wenxin Yiyan Accepted 9
DeepSeek V4 Pro Accepted 10
DeepSeek R1 X. None. 11 ‌Restart
Moonshot X. The same as Kimi. It will appear as kimi. ‌Out
Kimi K2.6 Accepted 12
Kimi K3 X. Busy. You can't use it anytime 13 ‌Out
Kimi K3 Cluster Need Money 14 ‌Out
iFlytek Spark Accepted 15
AstronClaw Compile error for just copy, Accepted after edit code 16

Analysis: Perplexity gave a perfectly logical explanation, but the submitted code ended up wrong. DeepSeek R1 crashed mid-thinking and gave no answer. AstronClaw had an invalid output format.

Round 2

Let's move on to a trickier *800 problem.

CF2250A

Extremely tricky! Let's see if the AI is careful enough. I myself got it wrong many times.

AI/Agent Model Status No. Revival Allowed
*chatgpt Wrong Answer on test 1 1 ‌Restart Yes
*claude Wrong answer on test 2 2 ‌Restart Yes
*deepseek Compile Error 4 ‌Restart Yes
*perplexity Wrong Answer on test 1 5 ‌Out No
Gemini 3.5 Flash-Lite Wrong Answer on test 1 6 ‌Restart Yes
Qwen 3.7 Accepted! 7 Yes
Doubao Wrong Answer on test 1 8 ‌Restart Yes
Baidu Wenxin Yiyan Accepted! 9 Yes
DeepSeek V4 Pro Wrong Answer on test 1 10 ‌Restart Yes
DeepSeek R1 * Accepted! 11 No
Kimi K2.6 Wrong Answer on test 1 12 ‌Restart Yes
iFlytek Spark Accepted! 15 Yes
AstronClaw Cannot_Use 16 ‌Out Yes

Analysis: So far domestic Chinese AIs seem to take the lead. DeepSeek R1 works extremely well, but it has to be used online otherwise it crashes. Qwen and iFlytek Spark perform nicely, while other AIs fail to notice the details and require carefully crafted extra prompts.

Sorry it's a "Chinese-style" problem with tons of tiny details, but it's a great test anyway — everyone can go try it out themselves!

Round 3

*900 Math Problem CF2238B

AI/Agent Model Status No. Revival Allowed
*chatgpt Wrong Answer on test 1 1 ‌Out No
*claude Accepted 2 No
*deepseek Time limit exceeded on test 2 4 ‌Out No
Gemini 3.5 Flash-Lite Accepted 6 No
Qwen 3.7 Accepted 7 Yes
Doubao Wrong Answer on test 1 8 ‌Out No
Baidu Wenxin Yiyan Accepted 9 Yes
DeepSeek V4 Pro Wrong Answer on test 1 10 ‌Out No
DeepSeek R1 Accepted 11 No
Kimi K2.6 Accepted 12 No
iFlytek Spark X. None. 15 ‌Restart Yes

Now the top 8 top 7 are selected, the finalists list:

No. Model Name
2 Claude
6 Gemini 3.5 Flash-Lite
7 Qwen 3.7
9 Baidu Wenxin 5.1
11 DeepSeek R1
12 Kimi K2.6
15 iFlytek Spark

Now the Challenge Round

Challenge Problem: CF689D

An RMQ + Binary Search problem. *2100

AI/Agent Model Status No.
claude Time Limit Exceeded on test 7 2
Gemini 3.5 Flash-Lite Wrong Answer on test 1 6
Qwen 3.7 Wrong Answer on test 4 7
Baidu Wenxin 5.1 Compile Error 9
DeepSeek R1 X. None. Unsolved. 11
Kimi K2.6 Accepted! Got a Restart Chance in Knockout Round! 12
iFlytek Spark X. Over 15 minutes no result 15

Полный текст и комментарии »

  • Проголосовать: нравится
  • -15
  • Проголосовать: не нравится

Автор zap2727, история, 2 месяца назад, По-английски

For Chinese version, click here

This competition will feature 17 free AI models competing against each other, with the top three strongest AIs ultimately selected.

Competition Format

  • Qualifier Round: Select 8 AIs from the total of 17. Starting with Codeforces problems rated *800, the difficulty increases progressively to eliminate participants until no more than 8 AIs remain. The remaining eliminated AIs can participate in a challenge round; if they successfully solve a problem rated *2100, they earn one resurrection chance.
  • Knockout Round: After the qualifiers, the 8 AIs engage in one-on-one PKs with progressively increasing difficulty. The first AI to provide an incorrect answer is eliminated. If an eliminated AI has a resurrection chance, it can consume one to advance to the next round alongside the winner.
  • It is evident that the knockout round will consist of at least 3 rounds and at most 4 rounds.

Participating AIs

AI/Agent Large Model Made in China Link
*chatgpt No https://chatai.org/chat
*claude No https://chatai.org/claude/chat
*Grok No https://chatai.org/grok/chat
*deepseek Yes https://chatai.org/deepseek/chat
*perplexity No https://chatai.org/perplexity/chat
Gemini 3.5 Flash-Lite No https://gemini.google.com/app
Qwen Yes https://www.qianwen.com/?ch=tongyi_redirect
Doubao Yes https://www.doubao.com/chat/
Baidu Wenxin Yiyan Yes https://wenxin.baidu.com/?extParams=%7B%22enter_type%22%3A%22home_operate%22%7D
DeepSeek V4 Pro Yes Baidu Version / Official
DeepSeek R1 Yes Baidu Version / Official
Moonshot Yes https://www.moonshot.ai/
Kimi K2.6 Yes https://www.kimi.com/
Kimi K3 Yes https://www.kimi.com/agent?chat_enter_method=change_model
Kimi K3 Cluster Yes https://www.kimi.com/agent-swarm?chat_enter_method=change_model
Spark Desk (From Iflytek) Yes https://www.kimi.com/agent-swarm?chat_enter_method=change_model
AstronClaw Yes https://agent.xfyun.cn/astron-claw/chat

P.S. For first five models marked *, I used https://chatai.org/ and can't know which models they are. Sorry. I will research out it.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +7
  • Проголосовать: не нравится

Автор zap2727, история, 2 месяца назад, По-английски

Problem link

We find an odd nodes tree has only one centroid. So if $$$n$$$ is odd, we don't change it. If $$$n$$$ is even, we delete an leaf and make it odd. Then we insert the deleted leaf to the centroid. And sorry for I met many bugs. Hope you can't meet them. Becareful.

1st Wrong Answer Okay, we need max to be min.

2nd Wrong Answer No, we just calculated the deleted leaf in!!!

3rd Wrong Answer It's not finished though. The nodes after delete is $$$n-1$$$ not $$$n$$$.

AC link

#include <bits/stdc++.h>
#define N 100005
using namespace std;
int T,n;
vector<int> g[N];
int s[N],mx[N],C,Leaf;
int leaf(int x,int fa){//O(depth) find a leaf
    if(g[x].size()==1&&fa)return x;//Not the root so fa!=0
    for(int y:g[x]){
        if(y==fa)continue;
        return leaf(y,x);
    }
}
void dfs(int x,int fa){//Calculate all subtree node numbers
    s[x]=1;mx[x]=0;//Init
    for(int y:g[x]){
        if(y==fa||y==Leaf)continue;
        dfs(y,x);//Calculate the subtree
        mx[x]=max(mx[x],s[y]);
        s[x]+=s[y];
    }
    mx[x]=max(mx[x],n-1-s[x]);//Caution: As x is the root, then the origin father is follow x's subtree. So we have another subtree with its father and uncles.
    //     (1)
    //     / \
    //   (x) (3)
    //    |
    //   (2)
    //Like up graph and at X we have subtree (2) and also (1)-(3).
}
void solve(){
    cin>>n;
    for(int i=1;i<=n;i++)g[i].clear();//clear the graph
    for(int i=1,u,v;i<n;i++){//input the graph
        cin>>u>>v;
        g[u].push_back(v),g[v].push_back(u);
    }
    if(n&1){//N is odd and the tree can only have one centroid. Just randomly delete an edge and add it back.
        cout<<1<<" "<<g[1][0]<<"\n";
        cout<<1<<" "<<g[1][0]<<"\n";
        return;
    }
    //Now N is even. We need delete an edge and connect it with our Subtree's centroid.
    Leaf=leaf(1,0);
    cout<<Leaf<<" "<<g[Leaf][0]<<"\n";
    C=0;mx[C]=0x3f3f3f3f;
    dfs(1,0);
    //Now we find the centroid.
    for(int i=1;i<=n;i++){
        if(i==Leaf)continue;
        if(mx[i]<mx[C]){
            //cout<<i<<" is better:"<<mx[i]<<'\n';
            C=i;
        }
    }
    //Now centroid is C.
    cout<<Leaf<<" "<<C<<'\n';
}
int main(){
    ios::sync_with_stdio(0),cin.tie(0),cout.tie(0);//Make input faster
    cin>>T;while(T--)solve();
    return 0;
}

Полный текст и комментарии »

  • Проголосовать: нравится
  • +2
  • Проголосовать: не нравится