foreword
very lazy.
Too lazy.
Not just lazy to write questions. . I don't even bother to write a travelogue. .
Take the time to fill the hole. . However, I didn't write the NOI/APIO/CTSC travel notes last year. .
Before the provincial election
The board seems to be finished early in the morning, and there is nothing to do.
It is also unwilling to open new questions, and it is even more impossible to learn new things. It is only by playing GTA and CR every day that I can maintain my life like this.
Still a little panic, what kind of pot will HAOI make? What if you are retired?
Usually my exam strategy is very conservative, so I decided to adopt a more conservative one this time.
Except for the questions that are easy to write at a glance, other questions only write about violence, regardless of the correct solution.
Forget this thing. . Getting in is the most important thing. Just like last year, I joined the team after just playing violence in HA. This year, I just decided to keep the violence points steady, and I don't think about the captain. What if the correct solution is written and it is the same as the violence score?
exam day, morning
Last year's NOI taught me what "mindset comes first", so I decided to adjust my mentality according to my own method this time. As long as my mentality doesn't explode, it's possible to make a comeback in the last minute.
The test started at 8:30 in the morning, Wo deliberately arrived two minutes late, and then entered the venue calmly.
Parents and coaches waited in a hurry. Coach kept calling me and I didn't hear that. . It is estimated that they are going crazy qwq.
After all, the provincial election time is very tight, only 3.5 hours, and there is no time to think about any correct solutions. . . It's like NOIp, and the question is so difficult, how can it be possible to write a correct solution?
Look at the question first, and count the violence points.
T1
Um? Backpack? NTT?
Obviously not. . .
The exponential enumeration seems to have 40 points, let's take a look first.
It feels a bit like number theory, something related to gcd, right?
Do you want to expand Pei Shu's theorem? Take the gcd of all numbers? (Guessing the conclusion)
(In the process of vigorously proving...)
Forget it. . That's right for two numbers, so that should be fine, right?
40pts to hand
T2
This question is very similar to a CF question? Do you just construct a spanning tree and choose a path to reverse?
False, obviously not. (Brain ing)
Let's fight violence qaq.
Consider 70 points, list an xor equation for each point, and use bitset to solve the equation system directly
. Ask again after each point is deleted?
Set up a divide and conquer horn, I feel that I can't pass it, but 60 points is still no problem qwq. (brain pumping)
60pts received
T3
Why is this string of questions so unworkable?
Put a SAM first, then? Then? Then what to do qaq?
Forget it or 30 points of violence. . Squaring LCP is also very easy to write, so there is no need to write SA.
But this data range is so insane. . Can low-complexity violence be enough?
If you have time, write SA, but there should be no time for T2 with such a large amount of divide and conquer code. .
30pts is in hand . In
this way, there are 40+60+30=130pts. It feels good. The captain is out of play. It should be more than enough to join the team, right?
Start writing.
T1
After writing 40 points of violence, and after a large sample, the conclusion should be fine.
Think about the score behind? Can it be DP?
How about DP for the problem of selecting subsets? There must be a clever way of doing it. (brain pumping)
2 hours left
T2
2h should be enough to finish and adjust the line segment tree divide and conquer + linear basis, right?
Otherwise, how to write this question. . My idea has nothing to do with it except this one?
Be tough. . .
(After 30 minutes)
It seems that it is really the same as the CF question. . Is it possible to reverse the path?
What was I just thinking about. . . How fat four ah. . .
Then this question is not BFS gone? linear?
Oh, there are still many inquiries, then continue to divide and conquer + merge and check the set, and there is still a log.
There are a lot of things to maintain in this collection. . It's only 1.5h now. . Even if you finish writing it in 1h, you won't necessarily call it out?
In this way, the square violence of 70pts is still very easy to write, and n times of BFS is enough.
After a large sample, 70pts is available.
T3
There are about 20 minutes left, so let's not write SA. There are so many details about this method?
Finished writing squared violence quickly. Get 30pts.
Then extended for 30min. . .
If I had known about the 30min extension, I would have written the correct solution for T2!
With 40 minutes left, search for other scores.
T1
Oh it does seem to be able to DP. . . What happened to me just now?
However, if DP is to press the current gcd into the state, the complexity is O(np), which is 60 points.
You can take out all the factors of p, but writing the map requires one more log, which is a bit complicated, and the time is not enough.
Finally, there is a free 10 points for p=998244353.
In this way, the T1 has 70pts, and it is no problem to take a shot.
T2
Only 5 minutes left. . In a panic, I went to shoot T2.
I wrote an index violence first, and then a data generator.
Generators are a bit cumbersome to write. . Because there can't be a double edge self-looping or something. . But this has no effect on my program, right?
The exam has just ended. . Shoot now!
wa.
? ? ? I've passed the big example, huh?
Shoot again!
TLE, an infinite loop.
? ? ? ? ? ? What's the matter? ? ? ?
Taking a look at the data, it seems that there is a point where there is a heavy edge and self-looping, and I hang up. But isn't my program able to handle self-loops with multiple edges?
never mind. . 70 becomes 0. . . Out of the exam room. .
Exam day, noon
In this way, it is only 70+0+30=100pts, which is quite dangerous, right?
It's good to stop the violence in the afternoon.
imone said that he can do the first two questions. T3 wrote the low-complexity violence of SA. I don't know how many points can be deceived.
"At least 60?" I sucked at him.
"Then you have 260 in the morning, tql!" I sucked him again.
My mind explodes when I eat. . The brain is running fast, thinking about how the T2 hangs up.
I finally figured it out later. . My program does get errors when dealing with multiple-edge self-loops. .
And according to the output of my morning, the error is in line with my expectations, it should be no problem.
Even so, imone still slaps me with 90 points?
Mike comforted me and told me to come back in the afternoon, and although I knew it wasn't possible, I agreed.
I heard someone say that T1 is a Möbius function. . Seems to make sense, doesn't it? Why didn't I think about it when I thought about factors?
Exam day, afternoon
Just like last year, the exam has not yet started, and the morning stuff has not been deleted, so I will take the morning T2 first.
(I also took the morning T2 qwq after I arrived at the exam room last year)
No problem, don't worry.
Take the exam and watch the violence.
T1
This question means. . this range. . .
What else is there to do besides square the DP?
And it seems that apart from squaring the DP, there is nothing to do to get points, right?
How to write the cube? This data range is funny, right?
In addition to 100 points is 20 points, and the violence of these 20 points is so difficult to write, of course, write 100 points.
Although I'm not very good at counting, I just have to be tough.
Just take a look at T2 horn qwq before hard.
T2
yql told me a similar question. This kind of question has nothing to do with w[k]. It must be calculated for each k first and then multiplied by w[k].
So I enumerate k first, and then how to combine it?
How to combine it, it's so difficult, QAQ?
Is this mod a bit weird? Is it NTT modulus?
Sure enough. . It's either a smoke bomb or really polynomial.
Generate function? really?
My mind is so messed up, no, goodbye.
The strategy is basically determined, first go to Menggang T1, and then come to T2 to get a little violence score.
Immediately panicked. . If this T1 just doesn't come out, it will basically explode in the afternoon.
But few people are good at counting, right? I guess everyone has a little bit of a point.
T1
Pushed the formula for a long time, without any progress.
It was supposed to end at 16:30, but I remembered it as ending at 16:00, a panicked batch.
Don't dare to guess the formula, it is very likely that you will FST after guessing the sample.
I just hope that after finishing this question, you can leave me 0.5h to write T2 violence.
After working for a long time, I came up with a formula, but I couldn't pass a small sample.
After a wave of calculations, I found that I really couldn't pass it, so I started to adjust the coefficients against the sample.
Just changing a wave of coefficients will pass a large sample. . .
Immediately, I felt that Dongdong was very real. When he told me about "51nod Alignment and Merger Machine", he said that he just changed a wave of coefficients by himself. It turns out that this kind of operation really exists.
The number of questions passed the big example and took the FST, and it must have been AC.
Looking at the watch, I found that it was only 50 minutes before the start of the game, which was very refreshing. Originally, this problem was expected to be completed in 2 hours.
It's like when I played CF Simulator, I suddenly passed a certain question 23333.
T2
Go ahead, first enumerate k, then what?
It happens that there are k colors, can it be tolerated in a broad sense?
(Writing an inclusive-exclusion formula)
Does that mean there is no such thing as an NTT?
The board said it! Generalized tolerance set NTT! Said many times!
But I didn't understand it once. QAQ?
Sorry for not listening well. . .
(bitter.jpg)
and then slowly began to push the formula of accepting and rejecting the change to NTT.
After pushing for a long time, I came up with one. Is this thing right? It looks quite reliable.
There is still 1h left at this time. . If this thing is written and written, it will definitely not be able to be adjusted. .
First of all, my tolerance formula is not necessarily correct. . Secondly, my NTT formula is not necessarily correct. . Secondly, my NTT may be wrong. . .
Decided to write the square tolerance first, regardless of the NTT part.
It is quite conscientious to find that the direct square tolerance is 50 points.
After writing, I really can't pass the small sample, continue to calculate by hand, and then adjust the coefficient against the sample.
After adjusting the coefficients, it has passed a large sample, which is cool.
There are only 15 minutes at this time. After backing up this code, I start to code NTT like crazy.
The NTT part feels that it is not difficult to write, that is, to write the formula in the code, but the boundary problem is still worth considering, although in fact there is no problem at all. . .
Then extended 30min, refreshing.
After writing, I couldn't pass the sample, and found that NTT was written and hung up. After NTT was corrected, it was OK.
It's already 16:45 at this time. Fortunately, it has been extended for half an hour. AK is hot~
Then I started to worry about the boundary problem, such as n=0, m=0, s=0, etc. I tried a lot and considered a lot, but it didn't feel like much. question.
I created some more extreme data, and I can run it.
Then the afternoon is over.
exam day, after
imone will not be T2 in the afternoon, as expected.
When Cydiater's jungler just finished the exam, he knew that I was AK and he also beat me. I didn't expect him to be AK too, strong.
There was almost no A question in the afternoon, and the scores were basically 20/30/100. It seems that few people can count the number of questions. . .
Then there are evaluations, and all kinds of pots and so on. . I wrote it on Zhihu.
Click here https://www.zhihu.com/question/271131295
end
After reading the list, I think there are quite a lot of HAOI masters this time. . No more chickens pecking at each other. . Complete violence will not be able to enter the provincial team. .
Primy retired. . very sad. . But nothing to say. . He probably won't buy D. .
Many people have retired. . . After all, there are only so many people in the provincial team. . It is impossible for everyone to be happy. .