Блог пользователя maomao90

Автор maomao90, 7 недель назад, По-английски

Disclaimer: This is a rant about Meta Hacker Cup and may not contain any useful information.

Meta Hacker Cup is one of the biggest annual programming competitions, but it has the strangest submission format, unlike any other online judge. Why does it have to be so different? Let’s take a look at how Meta Hacker Cup 2024 Round 1 went for maomao90

The time now is 1:00 AM in Singapore. The contest starts, and maomao90 begins solving the problems.

The time now is 1:41 AM. maomao90 has solved problems A, B, and C without much trouble and starts working on problem D. After quickly coming up with a theoretical solution, maomao90 begins coding.

The time now is 2:05 AM. The code is ready and passes the sample tests. maomao90 proceeds to validate the solution.

The time now is 2:06 AM. Validation passes, and the input zip file is downloaded.

The time now is 2:07 AM. maomao90 runs the code on the final test.

Image showing assertion failed

Oh no! How did the code pass validation tests but fail the final test with an assertion error? Panicking, maomao90 scrambles to debug the code.

The time now is 2:11 AM. Five minutes have passed since downloading the zip file. maomao90 fails to debug his code and is no longer allowed to submit problem D. maomao90 wasted 30 minutes of his time and is left frustrated and in tears.

Problem 1: Why is the validation test so weak?

Is the validation test intentionally weak, or is it a mistake by the problem setter?

If it’s intentional, what's the goal? To make participants suffer? Brute-force algorithms often pass validation easily but take far longer than five minutes for the final test. Why is that?

Problem 2: Why are participants allowed only a single 5-minute attempt?

Almost every other online judge allows multiple submissions when your solution is incorrect. Why does Meta Hacker Cup limit participants to just one try?

One possible reason is that if someone's code takes more than 5 minutes to run, they can wait until their code finishes running before making a second attempt and AC the problem even though their solution took much longer than 5 minutes to finish running. However, there's an easy solution to this:

  • Instead of one input file, create three strong input files, each worth $\frac{1}{3}$​ of the total points.
  • Allow participants to download each input file individually, with a 5-minute submission window for each file.
  • This way, if a participant fails to submit for the first input file, they can still debug and submit for the second and third, potentially earning $$$\frac{2}{3}$$$ of the total points.
  • This approach would also strengthen the final test with three times more input data.

The time now is 2:12 AM. After a brief crying session, maomao90 starts on problem E.

The time now is 3:41 AM. maomao90 validates the solution for problem E but lacks confidence after the disaster with problem D.

The time now is 3:44 AM. After a final check, maomao90 downloads the zip file and runs the code for the final test.

The time now is 3:45 AM. maomao90 submits the output for problem E. There’s nothing else to do now, as problem D can’t be submitted. maomao90 is tired and wants to sleep, but at the same time, maomao90 wants to know whether his final output is correct. Unfortunately, the final verdict will only released after the contest...

Problem 3: Why is the final verdict delayed until after the contest?

Is it to reduce server load by judging only after the contest ends? The server doesn’t even need to compile or run code~--- it only has to compare two text files. Is that really too much for the server during the contest?

If the final verdict were provided immediately, along with the solution proposed in Problem 2, the contest experience would be far more pleasant. Yet, after 14 years, there’s still no improvement in the grading system. Why is that? Even Codeforces is experimenting with pretest=system test to prevent "Fail System Test" issues.

The time now is 4:00 AM. The contest finally ends, and maomao90 can check if he solved problem E correctly. Thankfully, it was accepted and he celebrates.

The time now is 4:01 AM. Looking at the leaderboard, maomao90 sees the number of WAs on problem D.

image

So many red crosses! maomao90 laughs, realizing many others faced the same weak validation issues on problem D.

Problem 4: Why doesn’t Meta Hacker Cup follow other online judges and run the code for us?

The ultimate solution to all these problems is simple: adopt the standard system used by most online judges, where participants submit their code, and the platform compiles and runs it. Why hasn’t Meta Hacker Cup implemented this?

Codeforces held its first round in 2010, using the current code submission system, and Meta Hacker Cup started in 2011. Why did Meta Hacker Cup opt for this convoluted system of downloading password-encoded zip files instead of following the code submission system that Codeforces uses?

Please upvote this blog if you faced similar issues or agree with the solutions mentioned. Hopefully, Meta will consider these suggestions and improve the system in the future. :(

Полный текст и комментарии »

  • Проголосовать: нравится
  • +510
  • Проголосовать: не нравится

Автор maomao90, история, 4 месяца назад, По-английски

Introduction

Remember my previous blog? I mentioned that I am planning to write a blog about static top tree. AtCoder already has a pretty good explanation of the static top tree, so I was thinking how I can do better. Then, VisualAlgo came into mind. Why not I make an interactive website that showcases the static top tree? Well, that is exactly what I did!

Website

Here is the website that is hosted using GitHub pages. The public GitHub repository of my code is here.

The website is created using the React framework. This is my first time using React, so the code is quite messy. Feel free to let me know how I can improve the code in the comments, or even submit pull requests on GitHub!

I will be making use of my website to create diagrams for my Static Top Tree blog that will be coming up in the near future. Let me know what else you would like to see from the visualisation or how I can make the visualisation clearer and better. Feel free to suggest any additional features as well.

Credits

I am not an artistic person, so most of the UI is inspired by other websites. The majority of my UI is inspired by VisualAlgo and CS academy graph editor.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +136
  • Проголосовать: не нравится

Автор maomao90, история, 5 месяцев назад, По-английски

Introduction

I recently came across this problem which required an interesting trick to compute $$$a_i \otimes a_{i + 1} \otimes \ldots \otimes a_{i + w - 1}$$$ for all $$$1 \le i \le n - w + 1$$$ in $$$O(n)$$$ time and space. I found the trick very interesting, so I decided to write a short blog about it.

Problem

Given an array $$$a$$$ of size $$$n$$$ and an integer $$$w$$$. You are required to compute the value of $$$a_i \otimes a_{i + 1} \otimes \ldots \otimes a_{i + w - 1}$$$ for all $$$1 \le i \le n - w + 1$$$ in $$$O(n)$$$ time and space. $$$\otimes$$$ is a binary operation that is associative ($$$(x \otimes y) \otimes z = x \otimes (y \otimes x)$$$).

Solution

Split array $$$a$$$ into blocks of size $$$w$$$. In each block, we calculate applying the operator on each prefix and on each suffix. Then, any range of width $$$w$$$ can be formed by combining the suffix of one block with the prefix of another block.

The implementation is very easy as well. Let $$$p_i = a_{\left\lfloor\frac{i - 1}{w}\right\rfloor\cdot w + 1} \otimes \ldots \otimes a_i$$$ and $$$s_i = a_i \otimes \ldots \otimes a_{\left\lceil\frac{i}{w}\right\rceil\cdot w}$$$ for all $$$1 \le i \le n$$$. Then, $$$a_i \otimes a_{i + 1} \otimes \ldots \otimes a_{i + w - 1} = s_i \otimes p_{i + w - 1}$$$.

Extension

If we try to generalize this solution to work for queries of arbitrary width, we will realize that it becomes a disjoint sparse table. In disjoint sparse table, the queries can have arbitrary width, so we need to split into blocks of width $$$2^k$$$ for all $$$1 \le k \le \log_2 n$$$. For our application, since we only have queries of a fixed width, we only need one layer, and hence we can obtain a solution in linear time and space.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +178
  • Проголосовать: не нравится

Автор maomao90, 5 месяцев назад, По-английски

It's about time... 6 months have passed since Hello 2024 and the halving is taking place. It is about time I retire from my top 1 contribution spot :(

I want to take this opportunity to thank the community for being so generous with your upvotes for my Hello 2024 round. Thank you for letting me have this achievement of reaching the top contributor spot that not many people will get to experience. Regarding this, I hope that people will be more generous with their upvotes for other rounds as well. A big part of why Hello 2024 received so many upvotes is that Goodbye 2023 was not a great round. If a round went smoothly with no serious issues, why not upvote the round? This will encourage the problem-setters and motivate them to create more rounds in the future.

In return for the community's support, I am considering writing some educational blogs. Let me know what type of blog you want to see from me in the comments. I am considering writing a tutorial about Static Top Tree since there are some functionality that the atcoder editorial does not explain, and there has been quite a few problems that can be solved using Static Top Tree recently. Let me know what you think in the comments. Thanks!

Полный текст и комментарии »

  • Проголосовать: нравится
  • +635
  • Проголосовать: не нравится

Автор maomao90, история, 8 месяцев назад, По-английски

Introduction

Many communication problems involve sending information from one function to another by sending $$$0$$$s and $$$1$$$s (binary) or some number smaller than $$$b$$$ (base $$$b$$$). In these problems, we often need to change the information that we want to send from one base to another.

This can be particularly tricky when the information that we want to send is a sequence of numbers. A common way to do so is to send each number one by one using $$$\lceil \log_b (k) \rceil$$$ digits where $$$b$$$ is the base we can send information in and the elements of the sequence are between $$$0$$$ and $$$k - 1$$$. The problem with this is that if we are sending $$$l$$$ numbers, we are possibly wasting some digits as $$$l\lceil \log_b (k) \rceil \ge \lceil l\log_b (k) \rceil$$$.

In this blog, I will be demonstrating a way to encode $$$n$$$ sequences $$$a_0, a_1, \ldots, a_{n - 1}$$$, each sequence $$$a_i$$$ consisting of $$$l_i$$$ non-negative integers $$$0 \le a_{i, 0}, a_{i, 1}, \ldots, a_{i, l_i - 1} < k_i$$$. The sequences will be encoded into a single sequence $$$x$$$ of length $$$m$$$ consisting of non-negative integers $$$0 \le x_0, x_1, \ldots, x_{m - 1} < b$$$, where $$$m = \lceil\sum_{i=0}^{n - 1} l_i\log_b k_i\rceil$$$. Conveniently, the inverse operation to decode sequence $$$x$$$ back into $$$n$$$ sequences $$$a_0, a_1, \ldots, a_{n - 1}$$$ follows a very similar structure which I will show below.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +138
  • Проголосовать: не нравится

Автор maomao90, 10 месяцев назад, По-английски

I guess I can farm even more contribution since I am top 1 contributor now 🤡 . Maybe some of you might have some questions about my CP journey or Hello 2024 so feel free to ask here. No guarantees that I will answer every question because I will be serving in the Singapore Police Force soon (all male Singaporeans are required to serve two years of military service ://), which also explains the comments here. Thanks everyone for your support on Hello 2024 and helping me to climb from ~130 contribution to almost 170 contribution :).

Полный текст и комментарии »

  • Проголосовать: нравится
  • +519
  • Проголосовать: не нравится

Автор maomao90, история, 11 месяцев назад, По-английски

How did Hello 2024 announcement blog get 1000 upvotes before the contest even started 🤪 ? Is 4000 downvotes on Goodbye 2023 too much for codeforces to handle so upvotes on other blogs scale exponentially now 🤔 ? I mean... I don't mind but it's kind of funny that I am now one of the top 10 contributors because of Hello 2024 🤡 . Maybe this blog will get another 1000 upvotes 😁

Полный текст и комментарии »

  • Проголосовать: нравится
  • +340
  • Проголосовать: не нравится

Автор maomao90, история, 11 месяцев назад, По-английски

Hello Codeforces,

We are very glad to invite you to participate in Hello 2024, which will start on Jan/06/2024 17:35 (Moscow time). You will be given 8 problems and 2.5 hours to solve them. One of the problems will be divided into two subtasks. The round will be rated for everyone. There will be at most 2024 interactive problems, so please read the guide for interactive problems before the contest.

All the problems are written and prepared by me.

Spoiler

We would like to give our sincere thanks to:

The score distribution is $$$250 - 500 - 1000 - 1500 - 2250 - (1500 + 1500) - 4000 - 5000$$$.

Hope everyone will enjoy the round!

Congratulations to the winners!

  1. ecnerwala
  2. ksun48
  3. VivaciousAubergine
  4. gamegame
  5. cnnfls_csy
  6. maroonrk
  7. tourist
  8. Geothermal
  9. kmjp
  10. yosupo

Congratulations to the first solves as well!

UPD: Editorial

Полный текст и комментарии »

Анонс Hello 2024
  • Проголосовать: нравится
  • +2422
  • Проголосовать: не нравится

Автор maomao90, 11 месяцев назад, По-английски

1919A - Wallet Exchange

Author: maomao90

Hint 1
Solution
Code

1919B - Plus-Minus Split

Author: maomao90

Hint 1
Solution
Code

1919C - Grouping Increases

Author: maomao90

Hint 1
Solution 1
Hint 1
Hint 2
Hint 3
Solution 2
Code (Solution 1)
Code (Solution 2)
Bonus

1919D - 01 Tree

Author: maomao90

Hint 1
Hint 2
Solution
Code

1919E - Counting Prefixes

Author: maomao90

Hint 1
Hint 2
Solution
Code
Bonus

1919F1 - Wine Factory (Easy Version)

Author: maomao90

Hint 1
Solution 1
Solution 2
Code (Solution 1)

1919F2 - Wine Factory (Hard Version)

Author: maomao90

Hint 1
Hint 2
Hint 3
Solution
Code

1919G - Tree LGM

Author: maomao90

Hint 1
Hint 2
Hint 3
Hint 4
Hint 5
Solution
Code

1919H - Tree Diameter

Author: maomao90
Full solution: dario2994

Background
Hint 1
Hint 2
Hint 3
Hint 4
Hint 5
Hint 6
Solution
Code

Полный текст и комментарии »

Разбор задач Hello 2024
  • Проголосовать: нравится
  • +760
  • Проголосовать: не нравится

Автор maomao90, история, 12 месяцев назад, По-английски

Hello Codeforces,

We are very glad to invite you to participate in CodeTON Round 7 (Div. 1 + Div. 2, Rated, Prizes!), which will start on Nov/25/2023 17:50 (Moscow time). You will be given 8 problems and 2.5 hours to solve them. The round will be rated for everyone.

All the problems are written and prepared by lanhf, Mike4235, thenymphsofdelphi, xuanquang1999 and me.

We would like to give our sincere thanks to:

The score distribution is $$$500-1000-1500-2000-2250-2750-3250-(4000+1000)$$$.

Hope everyone can enjoy the round!

Congratulations to the winners!

  1. Radewoosh
  2. ksun48
  3. ecnerwala
  4. maroonrk
  5. jqdai0815
  6. hos.lyric
  7. zh0ukangyang
  8. 244mhq
  9. Vercingetorix
  10. BigYellowDuck

Congratulations to the first solves as well!

UPD1: The contest is delayed by 15 minutes due to prior issues with the registration system in order to make sure everyone is correctly registered. Please double-check that you are registered.

UPD2: Editorial

And here is the information from our title sponsor:

Hello, Codeforces!

We, the TON Foundation team, are pleased to support CodeTON Round 7.

The Open Network (TON) is a fully decentralized layer-1 blockchain designed to onboard billions of users to Web3.

Since July 2022, we have been supporting Codeforces as a title sponsor. This round is another way for us to contribute to the development of the community.

The winners of CodeTON Round 7 will receive valuable prizes.

The first 1,023 participants will receive prizes in TON cryptocurrency:

  • 1st place: 1,024 TON
  • 2–3 places: 512 TON each
  • 4–7 places: 256 TON each
  • 8–15 places: 128 TON each
  • 512–1,023 places: 2 TON each

We wish you good luck at CodeTON Round 7 and hope you enjoy the contest!

Полный текст и комментарии »

  • Проголосовать: нравится
  • +788
  • Проголосовать: не нравится

Автор maomao90, история, 14 месяцев назад, По-английски

It is not uncommon to have interactive tree problems where you are allowed to query some connected component of the tree and use the return value to determine whether the answer is in the connected component or outside the connected component (Link to example problem). The general approach for these kinds of problems is to always choose a connected component of size $$$\frac{n}{2}$$$. However, there are also problems where the allowed queries are more restricted, preventing $$$\frac{n}{2}$$$ halving from being possible. This blog covers one of those types of problems.

Definitions

  • Subtree: $$$S(r, u)$$$ contains the set of vertices in the subtree of vertex $$$u$$$ if the tree was rooted at vertex $$$r$$$.
  • Neighbour: $$$N(u)$$$ contains the set of vertices that are directly adjacent to vertex $$$u$$$.
  • Extended subtree: $$$ES(r, V) = \bigcup_{v\in V} S(r, v)\text{ if } v\in N(r)$$$. In other words, an extended subtree is a combination of the subtrees of a chosen set of vertices that are directly adjacent to the root.

Problem Structure

There is a hidden special vertex in a tree with $$$n$$$ vertices. Find the special vertex using at most $$$\lceil\log_{1.5}n\rceil$$$ of the following query:

  • Choose an extended subtree of the tree. The grader will return whether the special vertex is in the chosen extended subtree. More formally, choose any vertex $$$r$$$ and a subset of neighbours $$$V \subseteq N(r)$$$, then the grader will return whether the special vertex $$$x \in ES(r, V)$$$.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +119
  • Проголосовать: не нравится

Автор maomao90, история, 16 месяцев назад, По-английски

After 21 days and 56 submissions on IOI 2010 Maze, I finally broke the 100 point barrier and achieved a score of 100.081 / 100.

Score distribution

Полный текст и комментарии »

  • Проголосовать: нравится
  • +310
  • Проголосовать: не нравится

Автор maomao90, история, 19 месяцев назад, По-английски

Code golf challenge for 1817D - Toy Machine

In case you do not know what code golf is, the objective is to write the shortest code possible that solves the problem.

After losing 41 rating from Codeforces Round 869 (Div. 1) and almost becoming yellow again, I was depressed and decided to try to upsolve this problem. Surprisingly, I was able to discover a very simple pattern that allowed me to come up with a short and cute solution. Why do I always only solve problems after contest and not during contest 😭

Anyways, here is my code in python:

n,k=map(int,input().split())
print(["RDLU"*(n-k-2)+"LDLU"*n+"RDL","LDRU"*(k-1)+"L"][k<n/2])

Here is my original C++ code that is more readable 204002089.

The prize for coming up with an even shorter code is an ego boost. Bonus points if you come up with an even simpler solution that is different from the pattern that I discovered.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +52
  • Проголосовать: не нравится

Автор maomao90, история, 21 месяц назад, По-английски

Recently when I was doing Universal Cup Round 5, I got stuck on a tree problem A as I realized that my solution required way too much memory. However, after the contest, I realized that there was a way that I could reduce a lot of memory using HLD. So here I am with my idea...

Structure of Tree DP

Most tree DP problems follow the following structure.

struct S {
    // return value of DP
};
S init(int u) {
    // initialise the base state of dp[u]
}
S merge(S left, S right) {
    // returns the new dp state where old state is left and transition using right
}
S dp(int u, int p) {
    S res = init(u);
    for (int v : adj[u]) {
        if (v == p) continue;
        res = merge(res, dp(v, u));
    }
    return res;
}
int main() {
    dp(1, -1);
}

An example of a tree DP using this structure is maximum independent set (MIS).

Code

Suppose struct $$$S$$$ requires $$$|S|$$$ bytes and our tree has N vertices. Then this naive implementation of tree DP requires $$$O(N\cdot |S|)$$$ memory as res of the parent is stored in the recursion stack as we recurse down to the leaves. This is fine for many problems as most of the time, $$$|S| = O(1)$$$, however in the case of the above question, $$$|S| = 25^2\cdot 24$$$ bytes and $$$N = 10^5$$$, which will require around $$$1.5$$$ gigabytes of memory, which is too much to pass the memory limit of $$$256$$$ megabytes. Below, I will show a way to use only $$$O(N + |S|\log N)$$$ memory.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +270
  • Проголосовать: не нравится

Автор maomao90, история, 22 месяца назад, По-английски

For the past 6 months, I have been changing from red to yellow and back to red repeatedly. As a competitive programmer, I used my acute observation skills to try to figure out the problem. Immediately, I was able to see a trend. Whenever I am yellow, my rating increases. However, the moment I reach red, my rating will decrease.

My rating graph

Hence, can the color of GM be changed to yellow so that I can reach LGM? Thank you!

Полный текст и комментарии »

  • Проголосовать: нравится
  • +273
  • Проголосовать: не нравится

Автор maomao90, история, 23 месяца назад, По-английски

I was attempting to implement the solution to 1770G - Koxia and Bracket after reading the editorial recently but I kept getting TLE on test case 12. However, after changing the compiler from GNU C++17 to GNU C++17 (64), I got AC in 1185ms, which is very far off from the time limit of 5 seconds.

GNU C++17: 188211296

GNU C++17 (64): 188211331

I tried testing it on errorgorn solution in the editorial as well and there was the same problem.

GNU C++17: 188211673

GNU C++17 (64): 188211705

I thought that it might be because of the NTT implementation that we used (KACTL), however, I even tried on Radewoosh submission which uses a different NTT implementation, but still faces the same issue.

GNU C++17: 188203821

GNU C++20 (64): 187349301

I have seen some cases where 64-bit compiler runs faster than 32-bit before, but never to such a large extent. Does anyone know the reason why? Does NTT run a lot faster on 64-bit compiler? Or is it something about our implementation?

If anyone know the reason why, I will appreciate it very much if you could explain it to me down in the comments. I guess it is about time that I shift from GNU C++17 to GNU C++17 (64) 😢

Полный текст и комментарии »

  • Проголосовать: нравится
  • +35
  • Проголосовать: не нравится

Автор maomao90, история, 23 месяца назад, По-английски

I recently had to make slides to teach people about maximum flow and minimum cost maximum flow, so I thought it would be good if I share it with Codeforces as well. I think this will benefit everyone as the beginners can get a better understanding of the basics of maximum flow and minimum cost maximum flow in the first few slides and the more advanced people can look at the slides further on which explains how we can use minimum cost maximum flow to solve some greedy problems. Hope everyone will enjoy my slides!

Link to google slides

Please let me know in the comment if you see any mistakes or you want to suggest any improvements. Thank you!

Полный текст и комментарии »

  • Проголосовать: нравится
  • +190
  • Проголосовать: не нравится

Автор maomao90, история, 2 года назад, По-английски

Introduction

As mentioned in my previous blog, I will be writing a tutorial about slope trick. Since there are already many blogs that goes through the concept of slope trick, my blog will focus more on the intuition behind coming up with the slope trick algorithm.

Hence, if you do not know slope trick yet, I suggest that you read other slope trick blogs such as https://mirror.codeforces.com/blog/entry/47821 and https://mirror.codeforces.com/blog/entry/77298 before reading my blog. In the future explanation on the example problems, I will assume that the reader already knows the big idea behind slope trick but do not know how to motivate the solution.

Great thanks to errorgorn for proofreading and writing the section on convex convolution and merchant otter.

When to use slope trick?

Most of the time, slope trick can be used to optimise dp functions in the form of $$$dp_{i, j} = \min(dp_{i - 1, j - 1}, dp_{i - 1, j} + A_i)$$$ or dp functions containing costs with absolute values. In this kind of dp functions, the graph of the dp function where the x-axis is $$$j$$$ and y-axis is $$$dp_{i, j}$$$ changes predictably from $$$i$$$ to $$$i + 1$$$ which allows us to store the slope-changing points and move to $$$i + 1$$$ by inserting and deleting some slope-changing points.

Sometimes, slope trick can also be an alternative solution to a greedy question. The code will probably end up being the same as well, so sometimes slope trick can help you to find out the greedy solution instead. Personally, I find that slope trick is very helpful in this area as we do not have to proof the greedy since dp completely searches all possible states and is definitely correct.

Полный текст и комментарии »

  • Проголосовать: нравится
  • +211
  • Проголосовать: не нравится

Автор maomao90, история, 3 года назад, По-английски

Slope trick is one of my favorite algorithms, so I have been considering writing a blog about it. However, there are already a lot of resources about slope trick, so I am not sure whether anyone would benefit from yet another slope trick blog.

If I were to write a slope trick blog, it would focus more on the intuition behind how I solve slope trick problems together with multiple difficult example problems (not 713C - Sonya and Problem Wihtout a Legend) where I show my step by step thought process. If you would like to see this slope trick blog, please upvote this blog. I will write a slope trick blog if this blog receives more than 100 upvotes. Thanks for all your support 👍

EDIT: Wow, already more than 100 upvotes in such a short time. See you in my upcoming slope trick blog 😉

EDIT 2: The long awaited slope trick blog is here! Hope that you will enjoy it :)

Полный текст и комментарии »

  • Проголосовать: нравится
  • +395
  • Проголосовать: не нравится

Автор maomao90, 3 года назад, По-английски

Hope that everyone enjoyed the round. Feel free to ask questions in the comments if you do not understand any part of the editorial

1672A - Log Chopping
Author: errorgorn

Hints
Tutorial
Solution

1672B - I love AAAB
Author: errorgorn

Hints
Tutorial
Solution

1672C - Unequal Array
Author: maomao90

Hints
Tutorial
Solution

1672D - Cyclic Rotation
Author: errorgorn

Hints
Tutorial 1
Tutorial 2
Solution 1
Solution 2

1672E - notepad.exe
Author: errorgorn, oolimry

Hints
Tutorial
Solution

1672F1 - Array Shuffling and 1672F2 - Checker for Array Shuffling
Author: errorgorn

Hints
Tutorial
Solution for F1
Solution for F2

1672G - Cross Xor
Author: maomao90, errorgorn

Hints
Tutorial
Solution

1672H - Zigu Zagu
Author: maomao90, errorgorn

Hints
Tutorial
Solution

1672I - PermutationForces
Author: errorgorn

Hints
Tutorial
Solution

Полный текст и комментарии »

Разбор задач Codeforces Global Round 20
  • Проголосовать: нравится
  • +135
  • Проголосовать: не нравится

Автор maomao90, история, 3 года назад, По-английски

I decided to write a blog on this as I was doing a problem on our local judge and I decided to try to speed up my brute force code. However, it was quite difficult to find resources on SIMD vectorisation, so I decided to try to compile some of the resources I found to hopefully allow more people to learn to scam brute force solutions

Thanks to iLoveIOI and jamessngg for proofreading.

Introduction

SIMD stands for single instruction, multiple data. SIMD allows us to give vector instructions which will allow the code to run faster. Vector instructions are instructions that handle short (length 2-16) vectors of integers / floats / characters in a parallel way by making use of the extra bits of space to do operations simultaneously.

The most common form of vectorisation is making use of pragmas such as

#pragma GCC optimize("O3,unroll-loops")
#pragma GCC target("avx2,bmi,bmi2,lzcnt,popcnt")

This form of vectorisation is known as auto-vectorisation as the compiler vectorises the code for you. However, for more complicated examples, the compiler might be unable to detect what to vectorise automatically, so in this case, we have to vectorise our code manually by using SIMD vectorisation.

Syntax

The compilation of all the code given in the syntax section is given here

Code

To make use of SIMD, we have to add the following at the top of the code.

#include <nmmintrin.h>

#pragma GCC target("avx2")

Полный текст и комментарии »

  • Проголосовать: нравится
  • +277
  • Проголосовать: не нравится

Автор maomao90, история, 4 года назад, По-английски

Abridged statement

There are $$$N$$$ vertices in the graph where $$$N=2^n$$$ where $$$n$$$ is an integer. An array $$$A$$$ of size $$$M$$$ is given. An edge can be drawn from $$$i$$$ to $$$i\oplus x$$$ ($$$\oplus$$$ represents xor operation) if $$$x\notin A$$$. Print $$$N - 1$$$ edges such that the edges form a tree.

Statement

Issue

The intended solution uses xor basis / gaussian elimination. However, I found some submissions that uses completely different algorithms that ACs all the testcases.

In summary, the code iterates through all the $$$x\notin A$$$, and for each $$$x$$$, iterate through all the vertices $$$v$$$ from $$$0$$$ to $$$n - 1$$$. While $$$v$$$ and $$$v \oplus x$$$ are not connected, connect them and move on to vertex $$$v + 1$$$, otherwise, break. This algorithm runs in $$$O(n)$$$ as it will only connect edges $$$n - 1$$$ times and when it cannot connect edges, it breaks immediately. However, does anyone have a proof that it is correct? Will there be any case where breaking early results in the wrong answer? I tried creating a few test cases by hand and it seems to always generate the correct answer.

Code

In another submission, a similar idea was used, however, instead of breaking early, it iterated through all the vertices from $$$0$$$ to $$$n - 1$$$ as long as $$$0$$$ and $$$x$$$ are not connected. This clearly results in the correct answer as by looping through all the vertices from $$$0$$$ to $$$n - 1$$$, it will definitely result in at least one edge being created, so $$$n - 1$$$ edges will be created after all iterations. However, it looks as if the algorithm runs in $$$O(n^2)$$$. Why does it not TLE?

Code

Полный текст и комментарии »

  • Проголосовать: нравится
  • +14
  • Проголосовать: не нравится