### MrPaul_TUser's blog

By MrPaul_TUser, history, 2 years ago, 1593A - Elections

Idea: MikeMirzayanov

Tutorial
Solution

1593B - Make it Divisible by 25

Idea: MikeMirzayanov

Tutorial
Solution

1593C - Save More Mice

Idea: ITMO student team

Tutorial
Solution

1593D1 - All are Same

Idea: MikeMirzayanov

Tutorial
Solution

1593D2 - Half of Same

Idea: MikeMirzayanov

Tutorial
Solution

1593E - Gardener and Tree

Idea: MikeMirzayanov

Tutorial
Solution

1593F - Red-Black Number

Idea: MikeMirzayanov

Tutorial
Solution

1593G - Changing Brackets

Idea: nastya_ka

Tutorial
Solution Tutorial of Codeforces Round 748 (Div. 3) Comments (33)
| Write comment?
 » Can someone explain the solution to F pls
•  » » 2 years ago, # ^ | ← Rev. 4 →   F is a dynamic programming problem. You can use taken amount red digits count the rem of red number mod a the rem of black numer mod b to describe a state.The taken amount is phase of the dp. If you use enumeration method to find all states, the amount of states is $2^{40}$. But if you use dp to descibe all states, the amount of states is only $40^4$.Here is my code, written in Python3, but it get TLE. https://codeforces.com/contest/1593/submission/132439261Here is the same code rewritten by C++, it get AC. https://codeforces.com/contest/1593/submission/132442220I think the python code is more clearly than C++ code.
•  » » » Cool! Thanks
 » d2 is really intresting.
 » It seems that in the tutorial for A, the answer should be max(0, max(b,c) + 1 — a) (as in the solution code), instead of min(0, max(b,c) + 1 — a).
 » We don't need that extra array to recover answer string. We can store masks in DP and recover the answer from that. Submission with comments : 131983798
•  » » Thanks a lot! Cool explanation.
 » Your dp is so messy.(F) -__- Good problemset btw.
 »
•  » » Thank you very much, I was having a bug for a long time and I had no idea what it could be (눈‸눈)
•  » » » Feeling nice to hear that it helped.
 » 2 years ago, # | ← Rev. 3 →   For the E problem, will the standard graph bfs not work? I pushed all the leaves ( adjacency list size == 1 ) to a queue and assigned them a distance of 1 ( distance = number of times the operation has to be performed to cut off that node ). Then I performed a bfs assigning increasing distance. Answer is simply all the nodes whose distance is greater than k. This is failing though. Please if somebody could help me out. Thanks in advancehttps://codeforces.com/contest/1593/submission/132264294UPD : I changed the visited logic to an in-degree logic while preserving the bfs (AC : 132341078). Hope somebody finds this helpful. Thank you dedsec_29
•  » » Once you mark a node visited, your code will never update its distance. But this is incorrect. You have to mark a node visited only when it becomes a leaf. If you mark it before it becomes a leaf, d[i] will no longer denote iterations after which the node is no longer part of tree
•  » » » "If you mark it before it becomes a leaf, d[i] will no longer denote iterations after which the node is no longer part of tree."Can you justify this with a counter-example please? I tried a lot of manual cases (Sample cases passed too), every time the node gets marked as visited, it is correctly assigned the distance from the closest leaf. Say the distance of the node is 3 from the closest leaf and k=3, so that node will be cut off at the end of all k operations, right? Or am I missing something? Thanks again.
•  » » » » Take this test case for example:1 8 2 1 2 2 3 3 4 4 5 5 6 6 7 8 3 Your output: 2 Correct output: 3 Your code deletes node 3 as you mark the distance 2 for it, but it will be deleted in the 3rd iteration, so d should be 3, not 2
•  » » » » » Oh.. Yeah I found my mistake. Thank you so much.
•  » » » » » Thnks for your help
•  » » Thanks man
•  » » » You're welcome :)
 » 2 years ago, # | ← Rev. 2 →   Can Someone help with the time complexity of this code: https://codeforces.com/contest/1593/submission/132395142 for problem D2. I have used dp table for calculating the maximum gcd possible when I consider exactly n/2 numbers( because I have to make atleast half elements equal) dp[i][cnt][last] stores set of GCDs that are possible after considering cnt number of elements upto ith position in the array with index of previous selected element as last. It got submitted with 15ms runtime. According to me, the time complexity of this code is O(n^5 * (log(max(Ai)) + log(n))). I want to know whether I am wrong with the time complexity or test cases are not strong enough ! Reason behind asking this question Before I implemented this idea, I tried to calculate the time complexity of the algorithm, and as you can see from the above code, the three outer for loop will contribute n*n*n(n^3) times to the time complexity and the inner most for loop can iterate n*n times in the worst case( because n*n pairs possible with distinct differences). And with that there comes additional factors due to __gcd() function and inserting elements to the set. Hence total time complexity should be O(n^5 * (log(max(Ai)) + log(n)) ) After calculating the time complexity, I thought this should result in TLE, but still I implemented this to check the correctness of the idea and try to optimize it further if possible. After getting sample test case passed, I submitted it to check how it performs with all test cases. To my surprise it passed all tests within 15ms.
•  » » Can anyone explain the reason for this???
 » Did anyone solve F using meet in the middle idea ?
•  » » I'm trying to do it here but I need a way to combine the answer faster, I'm doing it in O(2^(n/2)*a*b) but it didn't pass.
 » My idea for E was to look at one diameter of the tree and take a middle point in that diameter. After that i root the tree in this middle point and after that it's easy but unfortunately i got memory limit exceeded. Is this idea works?
•  » » yes this idea works, be careful on how you are finding middle point(s) when diameter is even or odd. code
 » CAN F QUESTION DE DONE USING MEET INT THE MIDDLE?
•  » » Yes, but you need to write it with very low constant factor. You can check my submission here
 » 21 month(s) ago, # | ← Rev. 3 →   deleted
•  » » Because maximum difference is 1e6 — (-1e6) = 2 * 1e6; not 1e12
 » I found it in the code written in D1, and when I copied the code and confirmed it, it gave me: WA on test 1 After two years I discovered the error In the coming days, we must test the code before publishing it :)
•  » » When removing a "break", the code became valid, and the reason for this is that the written code must come in a smaller number, and this is a simpler code number. #include using namespace std; int main() { int t=0; cin>>t; while(t--){ bool s=1; int n,mn=1e9; cin>>n; int x[n]; for(int i=0;i>x[i]; if(i>0&&x[i]!=x[i-1])s=0;mn=min(mn,x[i]);} if(s)cout<<-1<
 » Can anycody give a counter case in which this approach fails for the problem E (link to problem)? I am not able to think about a counter case thanks Submission
•  » » Take a look at Ticket 17108 from CF Stress for a counter example.