Today we have released a new version of the CSES Problem Set:
CSES Problem Set is a collection of algorithmic programming problems that can be used to practice competitive programming. We have now added 100 new problems, and the total number of problems is 300. There are both easy and difficult new problems, and some of them cover advanced topics, such as treaps, suffix structures, and FFT.
In 2020, there were more than one million submissions to CSES, which is a new record. Thank you for your submissions and new test cases through hacking!
Our ultimate goal is to create a problem set of 1000 problems, so you can expect many new problems also in the future. If you have ideas on how to improve the problem set, you can discuss them here.
Thank you for the problems
Great. Make it like one stop Platform where any beginner can blindly follow.
Sample explanations please.
Yes, we are working on it
I copied the content of "Nested Ranges count" to chatGPT. Something like this may be automated for all problems.
It gave this nice explanation:
The input contains 4 ranges: [1,6], [2,4], [4,8], and [3,6].
The first line of the output gives the number of ranges that each range contains, in the order they were given in the input. Range 1 contains ranges 2 and 4, and no other ranges. Range 2 is contained in range 1, and no other ranges. Range 3 is contained in no other ranges. Range 4 contains range 3, and no other ranges. Therefore, the first line of the output is "2 0 0 0".
The second line of the output gives the number of ranges that contain each range, in the order they were given in the input. Range 1 is contained by no other ranges. Range 2 is contained by range 1, and no other ranges. Range 3 is contained by range 4, and no other ranges. Range 4 is contained by no other ranges. Therefore, the second line of the output is "0 1 0 1".
Please add tutorial option in the problem statement page if possible.
https://cses.fi/problemset/task/2413/ Counting Towers
I am missing a proper defintion on how to count distinct towers. Is a rotation or a reflection another tower?
Thanks, now the problem statement should be easier to understand.
yanire Yoav Noam13 almogwald inbarin Back to the grind?
Why does trying to hack Digit Queries give internal error?
Thanks, now it should be fixed
How to sort problems by difficulty level?
For anyone else, like me, curious for a list of just the new 100 problems to look through:
CSES — CSES Problem Set — Tasks
Introductory Problems
Sorting and Searching
Dynamic Programming
Graph Algorithms
Range Queries
Tree Algorithms
Mathematics
String Algorithms
Geometry
Advanced Techniques
Additional Problems
Nice good sir.
Thanks for doing this, I really appreciate your book and the problemset!
Couple of questions:
Is there any way to hide problem "tags" like DP, Graphs, etc? The only thing that stopped me from solving CSES is that you know if the problem will be DP or Graphs or Math in advance, thus making it less interesting.
Why 1000 problems? It feels great to solve everything, and this will be hard to do with 1000 problems, just from the time standpoint. 200 was perfect for this, 300 is not bad. I think it's possible to cover most essential topics/techniques, and still keep the problemset size small.
If you've started at the right time, after each expansion you only have to solve a 100 new problems :P
The idea is that the "tutorial" problems are divided into sections and then there are more difficult problems in the last section without hints.
With 300 problems the problem set is already quite comprehensive, but there are still many good educational problems that are missing. I agree with your point that it may take really long to solve them all in the future.
please add some decent questions of constructive algorithms to practice
best problemset for beginners and intermediates. thankyou!
I think it would be great if there are discuss section like leetcode.
Some problems are quite hard for beginners like me to figure out the solution all by myself.
You are doing a great Job. Thank you!
I always dreamt of being 1st on the CSES leaderboard, but I never imagined being 2nd simultaneously!
"There are only two hard things in computer science: cache invalidation and naming things."
orz pllk
Firstly, thank you for the helpful initiative!
I had a small clarification in this regard. The original post mentions the addition of problems based on some advanced concepts like treaps, suffix structures, and FFT. I just wanted to enquire as to if the book contains an explanation for these topics as well.
If not, can this please be done ?
Once again, thank you!
At the moment no, someday it will contain. Before that there are other sources that can be used to learn the topics.
Okay, thank you!
Official editorials of the problems may be a good idea.
Thank you for the amazing problem set!
I think it is the best website for someone who want to learn some new algorithm
Can you please add the sample explanations for the sample test case? I am finding some problems a little difficult to understand like Prefix sum queries.
Suggestion: add hacking leaderboards or some other way of crediting hackers.
Perhaps add next to a testcase the name of the person who added it? There are a few possibilities.
People will abuse it. Please take a look at educational round hacks.
Can anyone please help me out with the Counting Tilings problem?? I haven't been able to find any recurrence relation.Any hint??
I think it's bitmask dp, so $$$dp[i][mask]$$$ would be the answer for a $$$n \times i$$$ grid such that its $$$i$$$th column has positions covered at $$$mask$$$.
Can I ask what does the mask represent? And how do we transition?
EDIT: okay this might be wrong, let me get back.
till then here is a sub optimal way to solve the problem:
dp[i][mask] = number of ways to fill the cols from ith to mth given that cells of the ith col represented by mask are already filled.
final answer = dp[1][0] i.e. ways to fill from 1st till mth col given that no cell of 1st col is currently filled.
when trying to calculate dp[i][mask] try out all possible ways to fill the ith col and recur for i+1th till mth col.
the way you fill the ith col will decide the mask for the i+1 th col.
time complexity should be (2^N)*(2^N)*M
This solution will work. As if you precalculate the set of valid mask for each mask. There will at most 89 valid mask per normal mask and even less. so that one 2^n factor in the time complexity will be just 89(worst case), which is enough to get accepted.
Yes, made a video with proof for worst case time complexity of the solution: https://youtu.be/lPLhmuWMRag
Can we expect some sort of editorials and sample explainations for the existing problems in near future ?
Could someone explain problem Two Stacks Sorting? I can't understand its statement.
Consider having two stacks. Then we get numbers from left to right from input. We need to push each number onto one of the stacks.
Also we need to create a sorted list, by poping elements from the stacks. We can pop at any time, from any stack.
So, these rules imply that we never push a number onto a stack if there is a smaller number on top of that stack, because else we can never create the sorted list.
Hi pllk!
Since there are more problems now please consider adding subcategories. For example in math section you can add: number theory, combinatorics, probability, etc.
https://cses.fi/problemset/task/2137 Can somebody tell me about the solution of this problem? :(((
see https://cses.fi/book/book.pdf#page=111
awesome book btw
yes, i saw it. Thank u so much :))
https://cses.fi/problemset/task/2138 Please help me with this problems :((( i can't use dp in this dag
I think it would be lot more helpful if you guys will consider different time limits for different languages like it is in codechef and codeforce. Example — My linkedlist java solution for Josephus problem 1(https://cses.fi/problemset/task/2162) give tle while the same algo in cpp got accepted.
I don't think CF has different time limits for different languages.
And, at least IMO, language multipliers are a terrible idea, as described here.
I dont know much about cf but it is in cc. And bro sometime in cses my o(n) solution doesn't get ac, but the same logic in cpp works really fine. Though java is pretty fast then also this happens.
Choose the right tool for the right task (for cp, it's cpp).
Why do I need to log in to see statistics? https://cses.fi/problemset/stats/
it is what it is.. accept it or stop using cses
I have stopped. If needed I will just look at the problems and check whether I know the idea for solving it or not. Because I don't want to brainwash myslef because of TLE(which is not my fault)
We have had some problems with bots, so we have had to restrict some pages at the moment (for example, bots that fetch statistics too often).
Hi pllk
https://cses.fi/problemset/task/1194 Monsters
I spent some time trying to cover the following test case, because I erroneously interpreted the statement of the problem.
The expected answer for this test case is "NO" because both exits are farther away from
Athan they are fromM.But I thought
Acan escape becauseMcan't really commit to either direction untilAreaches all the way to the bottom, after which it's too late because thenAwill be closer to an exit thanM.I did not realize that
Ahas to plan the entire path before making the first step, and then commit to that path.Of course this is all my fault, because I misunderstood the problem statement, but in case there are more people like me, maybe it's worth stressing the fact that
Amust choose its entire path before making its first step.Thanks for the website btw, it's a great resource!
The problem states "Your plan has to work in any situation; even if the monsters know your path beforehand". I think that's clear enough.
Of course you're right.
My problem was that I misinterpreted the word plan.
A's plan is to walk down and then either dash left or right depending on whereMhappens to be. AndMcan't do anything against this plan.But of course
Amust plan its entire path beforehand.New tests have been added to both Fixed-Length Paths I & II hacking most solutions.
Are there any plans on 'hacked' notifications?
CSES problemset is just perfect, no tags, no editorials. It simply leaves us all by ourselves to find the solution which is what CP is all about. I can’t thank you enough.
when is it useful to do CSES problems? If I am rated 1150 is it useful to do CSES problems? Which sections?
Can anyone explain what the point of the question "Grundy's game" is?
As far as I know there is no legitimate solution that runs in the time limit--you either need to guess the answer and/or write a full solution locally.
Why not just lower the bounds of the problem--CSES is supposed to be educational.