The round is not finished yet, please don't discuss difficulties.

Codejam Round 1B (Div. 1)

A problem is one of the worst problems I have ever seen.

Still I am not surprised to see my rank drop by more than 500 places in a matter of few minutes... it's like always...

learnt 2 things from today's contest-
1. practicing for 1 year has only given me useless ratings, I am still at the same level(0 problems completely solved in Codejam 1B 2020 and 2021)
2. what's the point in making problems if coders can solve them? — logic behind making the problemset

edit-learnt a 3rd thing today
the only things that worked more efficiently than coder's brains were telegram and whatsapp
1 2 3 4 5

Yeah 1A had easier problem 1 and previous 1B also had easier problem 1 this time its too hard

Why didn't they write "In case there are multiple correct answers, you may output any of them" to the output-section of problem 1??

I spent an hour thinking there must be only one correct solution, and trying to figure out why I do not have the same output as sample 2, before finally noticing that the fact that you may output any solution WAS HIDDEN IN THE SAMPLE EXPLANATION?

Bruteforcing my way through B2, and still can't explain how did it pass. Btw A is the wackest problem ever.

I need more practice :/

How to get editorials? If anyone got solutions please explain

Me after reading A

spent an hour to figure out why my solution for C gived RE while using local tester provides WA.

even caught RE in this solution

ii(t,n,b,p)
while(t--){
    for(int i=0;i<n*b;++i){
        ii(x)
        assert(x<10); //if(x>9) while(1); gives WA
        cout<<i%n+1<<endl;
    }
}

ii(final)

still don't know what the problem

geometry sucks so I skip the first question(with rotation) and solve the other 2 questions instead.

For B we need to find a suitable bound for the answer. If it is too large than may TLE. By brute force we can know its 402.

For C, If there is a '9' then fill the top block. For second top digit if there is a >='8' then fill it. If there is no place for the top block then we need to fill the second digit with a lower value. I set it to >='7'. In my local environment it has some chance (I guess >20%?) so I Submit a few time and pass the second test case.

A was the worst. Wasted about an hour on C, and all ended up in TLE. In the end, solved B but it was too late. Missed by 77 ranks, bad day :(

Promotion is based on relative difficulty i.e. how well you do against other competitors. As long as the round offers differentiation (the problems are of varying difficulty, and the scoreboard isn't just a speedforces), it is fine since promotion is relative.

Complaining about things like problem quality is fine. Complaining about things like difficulty doesn't make sense. There were 1500 participants that solved more problems (or more quickly) than you. That wouldn't really have changed if the problems were easier, assuming differentiation still exists.

If this contest was on codeforces, it deserves something like 500 downvotes for its announcement blog.

A is totally boring code writing. B is just brute force, I even tried to ans = 10000 and still passed.

Problem A is quite possibly the worst CP problem ever.

That was some really low-quality round (perhaps one of the worst GCJ I've ever seen), in my opinion.

• P1 was just cumbersome modular arithmetic, with products of 64 bit numbers and a lot of eye-rolling;
• P2 wouldve been a nice problem, but for the fact that $$$O(T \cdot MAX^2)$$$ boring solution passes without a problem; not to mention, probably a lot of people implemented it without proving that it works, just YOLO;
• P3 was just atrocious.

I hope R1C/R2/R3 will be more decent.

My code in problem C has edit distance 0 to get AC. I just need to submit it again to get it, so frustrated...

I'm only candidate master and I managed to solve the first two problems and qualify with rank 646. They seemed impossible at first but I eventually found nice solutions for both of them:

Broken Clock

https://www.youtube.com/watch?v=21ADzj9dzz4

Subtransmutation

https://www.youtube.com/watch?v=p0YtxvbuRjI

I will admit that Broken Clock is harder than Subtransmutation, especially since ticks, and the hands movements aren't intuitive.

Why it is so hard? Are they making this for grandmasters only?

I dunno, It seems I got successfully promoted to the next stage

As you may notice I'm not exactly a grandmaster...

But yes, it is as always with such events, a lot of reading and coding. Much less fun than from codeforces rounds

The results are in preview right now and the ranks are updating. My rank got increased by almost 500. This means that 500 people got caught in plagiarism check or is there something else I am missing out? TIA.

Funny thing

B has score 31
A has score 30

Unofficial cutoff is 31, so
31 is enough to pass to the next round (depending on timing though)
30 is not enough

Though I don't find this round hard, I don't find this round good. And you don't need to be grandmaster to solve 100: I'm candidate master and was able to solve everything and placed 104th with some strange feeling that I've 15 more minutes till the end of round and nothing to do. I struggled a lot with A, but both B and C have seemed very easy for me. If somehow anybody is interested in my solving of this round without any commentary: my screencast

Are they making this for grandmasters only?

Personally I don't find the rounds hard. P1 was simply modulo arithmetic (N==s+t mod 360x12x10^9 or sth then subtract and everything is uniquely solved, rmb permutation) and P3 was greedy 9 which I guessed by taking a nap in-contest

After reading problems(Specially A), I was checking the announcement blog for downvote.

It was really bad.

I am not enough qualified to talk here. but why binary search failed in second test of B. I am binary searching over the answer for both the parities seperately. and my isitpossible function looks like this.

bool isitpossible(ll A,ll B, ll mid , vector<ll> v, ll n) {
    vector<ll> v1(mid+1);
    v1[mid]=1;
    while(1) {
        vector<ll> v22=v,v33=v1;
        ll flag=0;
        for(ll i=1;i<=mid;i++) {
            if(v[i]>0) {
                    ll z=min(v1[i],v[i]);
                    v1[i]-=z;
                    v[i]-=z;
                    
            }
            if(v1[i]) {
                if(i>=B) {
                    v1[i-B]+=v1[i];
                }
                if(i>=A) {
                    v1[i-A]+=v1[i];
                    v1[i]=0;
                }
            }
        }
        for(ll i=1;i<=mid;i++) {
            if(v1[i]>0 ) flag=1;
        }
        if(flag==0) break;
        flag=0;
        for(ll i=1;i<=n;i++) {
            if(v[i]>0) flag=1;
        }
        if(flag==0) return 1;
        ll z1=v33 ==v1;
        ll z2= v22== v;
        if(z1 && z2) break;
    }
    ll  flag=0;
        for(ll i=1;i<=n;i++) {
            if(v[i]>0) flag=1;
        }
        if(flag==0) return 1;
        return 0;
}

For anyone interested, here's a link to my screencast: https://youtu.be/zvV4sjS0fBU

I tried to upsolve A in python and faced this...

let's find the clock rotation angle for current permutation. I obtained the following formula (so we can test each permutation in O(1), idk why it's not mentioned in editorial):

x = (A * 12 - B) * inv % mod

where mod = 360 * 12 * 10 ** 10 and inv is an inverse of 11 for this mod. Naturally, I decided to precompute inv using the familiar code:

inv = pow(11, mod - 2, mod)

and guess what, this line gives an incorrect answer, apparently because of integer overflow. In python. Can someone clarify this? I would've freaked out if it was an actual contest...