I'd gladly exchange my skill of solving easy problems with your stupidity

lol xxD

same here, in today's global contest i spent 18 minutes to solve A , thats give me emotional damage.

I would like to understand your thought process ?

were you stuck in finding the core-logic ?

Or were you stuck with implementation ?

In my case, I had found logic in the first 30 minutes while solving D. But I had to think a lot while implementation, to avoid getting into large numbers. Because, there would have been integer overflows. I had to split each number into two parts. 1) odd number, 2) power of 2's in that number. This took me more than 40 minutes to do it fully. .

I've taken part in Codeforces Global Round 27 and spent too much time solving 2035C, so I solved 2035D instead.

I think it's not uncommon to find that a problem's point value or number of solves doesn't always correspond to how difficult the problem is for you? At least you solved D in the end. :)

For example, I thought that problem A from today's ARC was harder than all the others, and that F from today's global round was comparable or easier than D.

I find D today to be crazy hard, here is what my thought process is:

Firstly, no immediate properties comes to mind. So, start by the next line of generic thinking:

Start from an optimal configuration, (decent amount of thinking ), figure out a condition for every pair

then, ( decent amount of thinking later), lift this condition for the global configuration

then,(even more thinking later), find out that a stack can indeed maintain the optimal global configuration and is correct to do so.

Perhaps, the idea is it is far easier if you just guess a solution and verify that it works. I am also curious how can anyone figure this out considerably easier.

in contrast, E and F are basically trivially: they fall to the first statement — an immediate property did come to my mind. SQRT for E, monotone by +=2n for F.

TLDR: I think 2035D is crazy hard and I used the maximum amount of thinking for it, not unlike problems in regions of >=2700.

Same :(

It seems like a common theme for me in contests to approach easy problems wrong often and lose a ton of time or even get completely stuck.

In my last contest I solved C slower than F1&F2...

I think its just that A-D requires an almost different skillset compared to anything E and beyond

So if you kept practicing the harder problems naturally you will get a bit rusty on those

Btw, how to prove this greedy, which is heavily used by my solution?

How to solve the "simple" problem. Assume number $$$a_i$$$ is even. Then we can before start divide it by two and remember one operation: choose number $$$j \ge i$$$ and multiply $$$a_j$$$ by two. Let's apply all those divisions, and finally we will have an array of only odd numbers and a set of suffixes, where we choose an element and multiply it by two.

Now to maximize the final sum use the following greedy: go from right to left (e.g. from small suffixes to big) and apply multiplication to the largest currently number.

It sounds for me quite correct, but I can't prove it.

i am wondering the same thing :(

Bro suffering from success.

I'm actually wondering about problem E from Global Round. I tried different ways to bruteforce it, but it got either TL or WA. Then I just additionally bruteforced around $$$\sqrt{(z*y/x)}$$$, which is an optimal point for $$$f(u) = x * u + y * z / u$$$, where u — is the number of upgrades used. As you can see this function is simplified and doesn't consider the restriction given by $$$k$$$. Surprisingly it still worked, and I don't know why. Maybe the optimal point doesn't shift that much, or maybe it's just hard to make strong tests in this problem.