AtCoder try not to make multiple problems with the same theme just bigger constraints challenge: IMPOSSIBLE

Why set the precision of F to 10^-9, even though its solution is real number binary search? It is quite annoying to deal with precision issue.

Edit: the official solution is not real number binary search.

does solution of $$$F$$$ contain some elements of geometric right angle triangle calculation?Yes or No?

HELP — Can anybody help with this solution of D,it fails 3 testcases for some reason.

For problem G. I got this dp

for(int i=1;i<=n-2;i++){
    for(int j=i;j>=1;j--){
        dp[j] += dp[j-1] * i;
    }
}

Which is the coefficients of this polynomial $$$(x+1)(x+2)(x+3)...(x+n)$$$ which is equal the stirling numbers of first kind, but failed to get this is $$$O(n)$$$ or $$$O(n log n)$$$, only thought about an $$$O(n log^2 n)$$$ fft solution but didn't even code it.

I wasted 40min because I wrote 'continue' as 'break'.

:(

Why doesn't binary search work for F? Is it because the required precision is not achievable under the given constraints for the iterations of the binary search?

Can anyone help as to why my submission for F is failing on 1 test case and I'm unable to decipher what is wrong with it. my submission

For problem F, why do i set the upper bound of binary search 2e18 WA but 1LL<<60 accept??? code 2e18 code 1LL<<60

How to solve problem F?

How can I solve problem F? I use a binary search to find out the answer, but it got wa. I think my precision of my answer has been hacked, but I can't fix it. So can anyone help me?

Sorry for my poor English.

Here is my code.

D>>F>E

The difficulty of F is centred on the consideration of the boundary case, and the precision.

Please don't set such high precision requirements! Is there a difference between 1e-6 and 1e-9? A lot of people find the right solution but can't break the limits of floating-point accuracy. Such a high precision requirement is difficult to meet and does not improve the "mental difficulty" of the problem, so it is meaningless. （Sorry for the bad translation software.）

I think there are some major issues in Problem $$$F$$$. Kindly check this submission.

We can observe the following:

1. If the line segment touches the the top of building $$$A$$$ then the top of building $$$B$$$, $$$B$$$ is considered not visible for the initial condition ($$$H = 0$$$), and considered visible elsewhere (notice the 2nd arg in the 2 invocations of the function '$$$can$$$', changing that arg in any of the invocations gives wrong answer).
2. Changing the binary search upper bound to other limit like $$$1e18L$$$ or $$$2e18L$$$ gives wrong answer. It seems working only with the written limit (which is equivalent to $$$1LL$$$ << $$$60$$$) and some values around it, as if the test cases were hand-crafted on such limit, which is incorrect testing approach.

Oh this is a fun contest,but I think D is kinda complicated so that I waste much time on it

Why am I getting wrong answer if I use binary search in E problem as if we can see from H height, then we can see from all heights greater than that?

Why problem F could be solved with a monotone stack?