Hello Codeforces!
On Oct/20/2022 17:35 (Moscow time) Educational Codeforces Round 138 (Rated for Div. 2) will start.
Series of Educational Rounds continue being held as Harbour.Space University initiative! You can read the details about the cooperation between Harbour.Space University and Codeforces in the blog post.
This round will be rated for the participants with rating lower than 2100. It will be held on extended ICPC rules. The penalty for each incorrect submission until the submission with a full solution is 10 minutes. After the end of the contest you will have 12 hours to hack any solution you want. You will have access to copy any solution and test it locally.
You will be given 6 or 7 problems and 2 hours to solve them.
The problems were invented and prepared by Adilbek adedalic Dalabaev, Vladimir vovuh Petrov, Ivan BledDest Androsov, Maksim Neon Mescheryakov and me. Also huge thanks to Mike MikeMirzayanov Mirzayanov for great systems Polygon and Codeforces.
Good luck to all the participants!
UPD: Editorial is out
Hope to cross 1100 <3
Best of luck bro
What F*ckhead created problem C ?
What is wrong with this problem?
Oh excuse me, absolutely nothing.
Its just IMPLEMENTATION IS TERRIBLE
If you don't have a specific idea, you gonna (-s-u-c-k-) Perform bad
Firstly, If you don't have a specific idea on any problem you cannot solve it
And secondly, bruteforce with simulation which is two nested cycles and five simple ifs is not that hard to implement
XD
it's more related for problem C
yup
When You Realize After The Contest You're Getting The WA Only Because Of a Typo
edit: and that was me today too I spent one hour because of a typo but gladly solved C in the end
Thanks awoo! Hope the problems will be as great as the last contest. And i hope to become a pupil again in this contest. Best of luck everyone!
same, but expert instead of pupil
I am hoping to become pupil for the first time.
congratulations
.
haha
I want to be the user with the lowest contribution in codeforces, can you help me?
good luck
Although the contest of codeforces is vwery good, some users are very unqualified.
Thanks to those of us who helped me.
Oh yeah by the way, awoo where's the list of testers?
upd: nevermind, realized previous edu contests didn't have testers either
This sleepless night, I'm going to hit my highest rating, you give me a good look.
We wish you will. Good Luck!
thank you
HAPPY VIETNAMESE WOMEN'S DAY – October 20 ❤️❤️❤️
Edu Rounds are my best ❤❤
Wait I think I have seen it somewhere :)
Hoping to cross 2050! Good luck!
We wish you will. Good Luck!
time to be specialist
Hoping this contest to be math free ...
()
You just used this in real life by making this meme
best of luck, guys! my second contest. hope to solve at least A again :)
I forgot my notebook and my school's computers don't have any IDEs nor compilers in them.
Gonna use an online compiler. Let's see how it goes
Thanks for the round, C is the stupidest problem I have ever seen.
You couldn't solve it doesn't mean the problem was bad.
It's not that I couldn't solve it, it's a trivial n^3 simulation, I just didn't feel like implementing it. I just don't think problem C should be just a straight forward implementation, instead it should be some smart idea.
Seeing the number of solves of each problem and my experience, I feel that it was a pretty balanced round with a good gradient of toughness of the problems. (Though, some users might have had a sour experience with problem C)
How to solve C?
I've just simulated it; iterating through from k = 100 to k = 0, if Alice wins then print out the k and return
Binarysearch, with greedily picking Alice the biggest less than or equal to k-i+1, and Bob picking lowest left value. Storing values in multiset.
To win a game with k turns, the array must have at least k '1', and at least 1 number that is smaller or equal to 2, 3, ..., k. This is because Bob can use a strategy to lock the final move out by adding to each '1' once. Therefore, Bob can lock at most (k-1) '1', so the array must have k '1' to compensate. From there, just search out for the solution!
C can also be solved in $$$O(n)$$$ time without using Binary Search. The important observation to make here is that during the $$$k^{th}$$$ stage Alice must remove an element $$$1$$$ from the array. Thus, the optimal strategy for Bob is to always remove $$$1$$$ (by adding $$$k-i+1$$$ to it) at every turn. If the count of $$$1$$$'s becomes zero before the $$$k^{th}$$$ stage is reached, Bob automatically wins.
Hence, we store the frequency of every element $$$a[i]$$$ in an array $$$cnt$$$. Note that the answer is zero iff $$$cnt[1] = 0$$$ else the answer will be $$$\geq 1$$$. Now, we iterate over every $$$k$$$ from $$$2$$$ to $$$n$$$, while maintaining a counter $$$cur$$$ which stores the number of elements $$$\geq 2$$$ and $$$\leq k$$$ which have not been deleted yet. In every round, Alice must delete a number $$$\leq k$$$, so if $$$cur > 0$$$ we decrement $$$cur$$$, otherwise we decrement $$$cnt[1]$$$. We then decrement $$$cnt[1]$$$ to account for Bob's move. Now, if at this point $$$cnt[1] \geq 1$$$ then Alice can always make her final move and win if she chooses to play $$$k$$$ rounds. So we update $$$ans$$$.
Submission: 177175786
UPD 1: Why is a move by Bob equivalent to removing an element from the array?
Claim: Any element modified by Bob in the $$$i^{th}$$$ round cannot be removed by Alice in any further round.
Proof: Let Bob modify the element $$$a[j]$$$ in the $$$i^{th}$$$ round. Therefore $$$a[j]$$$ changes to $$$a[j] + (k - i + 1)$$$. Note that in the following rounds, all the elements Alice removes are $$$\leq {k - i + 1}$$$.
However, from problem constraints,
Hence, the Claim is proved, and we can say, for the purposes of the solution, that the element has been "removed" from the array.
readforces
.
Binary search
bob can win iff he can destroy all 1 in the array while alice kept choosing the largest possible option, then simulate with two pointers
I was just curious looking the hacked solution of Unknown_VN_Coder by k12 and his submission to the same problem. And I think that it was a fake hack. I could be wrong though
that solution will shut itself after outputing 'non' if t==13 for whatever reason and got hacked just as he wishes like piles of other such hacks, idk why people kept doing this
Video Solution for Problem C and Problem D
Problem B : Justt think in a simple wayy O_O
Me : Bruhh lets do a simple graph and get 2x WR <(_ _)>
Isn't E a dp problem. Let $$$dp[i][j]$$$ means minimum number of extra cactus need to be added for right part of grid from column $$$j$$$ to $$$m$$$ if I put a cactus at $$$(i, j)$$$ .Why am I getting wrong answer.
Anyone help please, Submission: https://codeforces.net/contest/1749/submission/177209214
Yeah, I tried many examples for last 30 mins but couldn't really figure out the counter example. I wish I could see the test 2 now...
Many submission have got WA on same test case (563).
UPD: this was invalid testcase. Look at the comment just below it for a test case
I think your test case is invalid.You cannot have cactus on two adjacent cell. But i may have figured out where my solution is wrong.
My solution will give 1 but actual is 0
That is not even a valid placement of cacti. Probably the issue of most WA submissions was one-directional dp, while we needed bfs.
was it just me or C was really challenging?
I think C is just a C, as always
Actually C is bit easy in this round, but i wasnt able to solve b this round , C is prefixsum problem , in which there should be atleast "k 1's" , "k+1 numbers <=2" , "k+2 numbers <=3" , similarly till k , you can check this Problem_C_Solution
it doesn't need much analysis actually. you just brute force all k and simulate the game. optimal action for both players is just simple greedy
Can someone tell me what is wrong with my modular arithmetic please? https://codeforces.net/contest/1749/submission/177211216 I got the algorithm done with 30 minutes left, then spent the rest of the time trying all combinations and figuring out how my modular arithmetic is wrong...
The issue is the range of m. For example, you multiply by m/2. Use (m/2)%MOD instead.
Thanks for replying! I already tried (m/2)%MOD, but it is still wrong... https://codeforces.net/contest/1749/submission/177211216 There are probably a lot of things wrong in there TwT
you have another problem :
int n,m;
int ans = 0, minu = m;
int po[300005];
I already defined int to be long long so the int stuff should work out normally
No, you are still multiplying by m/2. You need to do minu *= (m/2)%MOD; minu %= MOD. In the end, you also need ans %= MOD before if(ans<0)
Sorry for posting the same solution :_: this is the with (m/2)%MOD but still wrong https://codeforces.net/contest/1749/submission/177215823 Do you have any links to learn modular arithmetic in CP so that these problems aren't happening again? Thank you!
Change n, m, ans, etc to long long, then I guess it will be fine. What about wiki page? https://en.wikipedia.org/wiki/Modular_arithmetic Or you can just read the intro section of any elementary number theory book.
Thanks for your help! Unfortunately, I still could not find out what went wrong, maybe finishing it tonight tho...
minu = m % MOD. But seriously, the diagnostics message is already pointing the exact problematic line in this case. It doesn't make sense that you can't debug this.
thanks so much for all the help! I cannot believe how did I miss that either, was the difference between -20 and +80 this contest lol
change this line :
int ans = 0, minu = m;
into this two lines :here is your code getting AC : 177225248
Oh my god why did I miss that... Thank you so much, forgot the initial part of modding the first
Can anyone please share their approach for Problem E?
It's a shortest path problem, you need to find a path of '#' connecting the left side of the grid to the right side. All cells on the path must have the same color if we colored the grid like a chess board. Also you can't have 2 '#' beside each other, so some cells are "banned" of ever becoming '#' (because of the initial placement of '#'). After coding these constraints, it becomes simple dijkstra or 0-1 bfs.
after getting ac in a i accidentally submitted c in 'a' instead of 'c' but it didn't pass the first sample test would i get wa in 'a' after system testing?
no
Why is #829 and #830 on same day with only 30 minutes gap?
Maybe this bad will be fixed
does anyone have a tricky test case for E??
Thanks a lot.
got it, thx
Alice and Bob bullied me a lot today...
I solved B after noticing that many coders have solved it lol. Actually what I was doing earlier in intuitive way is that selecting the element with least spell which can be transferred to neighbors and then checking that element is leftmost, rightmost or at the center so that accordingly I can add spells to the ans and therefore stucked with WA, and then I realized at the last 3 min of contest that simplest way is to take n-1 min spells and thus finally able to submit it. uffff
I keep getting 726722623 as the answer for the fourth test case in D. Did anyone else have this issue?
My approach: In a non-ambiguous array, the $$$i$$$-th element must be divisible by the product of all prime numbers $$$\leq i$$$. The number of such candidates is equal to $$$m$$$ divided by this product (integer division, no modulo). Once the product exceeds $$$m$$$, all arrays of this size are ambiguous. The number of ambiguous arrays of size $$$i$$$ = $$$m^i - $$$ the number of non-ambiguous arrays of size $$$i$$$.
Submission: 177210359
Doesn't it overflow?
Ahhhh, I see now! Since $$$m$$$ is so huge, I pretty much need to apply the MOD on almost every calculation involving $$$m$$$. Thanks!
I was able to AC now (but the contest is already over :( )
Why is this true?
In a non-ambiguous array, the ith element must be divisible by the product of all prime numbers <= i let say ith is 6 and element is 8. gcd(8,6) is 2 but 8 is not divisible by 3.
array here is 1 1 1 1 1 8
If there is an element 8 at index $$$\geq 3$$$, then you can keep picking the first element until 8 is at index 3. Then you can pick the 8 because $$$gcd(8, 3) = 1$$$, so this array is not ambiguous.
oh i see. so any thing at ith must have gcd of everything from 2 to i not 1 and hence divisible by all the primes. Why did you have overflow problem? Wouldn't that be at max 10-15 primes before the product is larger than 10^12
The overflow is because $$$m$$$ itself is as high as $$$10^{12}$$$. So even for $$$n = 2$$$, the number of non-ambiguous arrays is $$$(m * (m/2)) \% MOD$$$, but the intermediate $$$(m * m/2)$$$ already triggers overflow.
In fact, I used binary exponentiation to calculate $$$m^i$$$ (total number of arrays of size $$$i$$$), but the multiplication for odd $$$i$$$ overflows unless I apply the $$$MOD$$$ on the base $$$m$$$ itself.
Also you can check for such overflows by compiling with the flag
-fsanitize=signed-integer-overflow
. Helped me debug the exact same errorcan anyone tell me their approach for problem c? I had couple of approach but none worked.I thought one with binary search,tho it was taking o(n^2) overall.
It's optimal for Alice to always pick the largest valid number, while Bob always picks the smallest number. This means Bob will eliminate the smallest $$$k - 1$$$ numbers. Therefore, if Alice can win, she can win by picking only the $$$k$$$ numbers that are after the first $$$k - 1$$$ numbers in sorted order. In other words, she can pick the $$$k$$$-th number when her upper bound is 1, she can pick the $$$(k + 1)$$$-th number when her upper bound is 2, she can pick the $$$(k + 2)$$$-th number when her upper bound is 3, and so on.
We can binary search to find the largest $$$k$$$ that fulfills this condition. My submission: 177165675
With how small $$$n$$$ is, a simple linear check for each $$$k$$$ should work too (instead of binary searching).
How to solve E?
What is wrong in problem $$$D$$$ in this logic?
If $$$gcd(a[i],j)>1$$$ for all $$$2 \le j \le i$$$ and $$$2 \le i \le n$$$ then, the array would be unambiguous (obviously), also if $$$gcd(a[i],j)=1$$$ for some $$$2 \le j \le i$$$ the array would be ambiguous, because after $$$i-j$$$ moves by removing $$$1^{st}$$$ element, the $$$i^{th}$$$ index becomes $$$j^{th}$$$ index now, and now instead of removing the $$$1^{st}$$$ element in remaining array in $$$(i-j+1)^{th}$$$ move, I will simply remove the $$$j^{th}$$$ element.
But something is wrong as I can't seem to pass $$$4^{th}$$$ sample :(
In Line 16:
curr=(curr*m)%MOD;
can overflow, since $$$m$$$ is so freaking huge. You should apply the MOD on $$$m$$$ first, before multiplying.Changing it to
curr=(curr*(m % MOD))%MOD;
passes the fourth test case, and will probably be Accepted.Yes it passes now :( Suffering from overflow issues :(
How to solve D?
We know that one of the b would be an array of all 1. For something to be non-ambiguous, it must not have anything other than 1 in the b array.
My approach that ith element must be relative prime to all number from 2 to ith. Because if this is not the case, then at some point the gcd(ith, 2-i) != 1. It is just a matter of calculate how many number are possible that is less than m at ith.
How do you even do this though?
Exactly man! Now I too made the same observation, now just see that since the product of first $$$12$$$ primes (upto $$$37$$$) is greater than $$$1e12$$$ $$$(m < 1e12)$$$, so any array of size > $$$37$$$ must contain an element $$$a[i]$$$ which would not be a multiple of some prime $$$p \le i$$$. So, any array of size > $$$37$$$ must be ambigious
In an un-ambiguous array, the $$$i^{th}$$$ element must have $$$gcd>1$$$ with all $$$2\leq j\leq i$$$. This means it should be divisible by all primes $$$\leq i$$$*. The count of values $$$\leq m$$$ that are divisible by all primes $$$\leq i$$$ is $$$\lfloor \dfrac{m}{\prod_{j=1}^{j=i}{j\cdot isPrime(j)}}\rfloor$$$.
*Why the $$$i^{th}$$$ element should be divisible by all primes $$$\leq i$$$: In an un-ambiguous removal sequence, the $$$i^{th}$$$ element's position will decrease by $$$1$$$ after each removal. So, it will hit all the prime positions $$$\leq i$$$, and any composite position $$$\leq i$$$ is just a combination of primes $$$\leq i$$$.
How to proof the conclusion in the last part?Is it a theorem or something else?
Which conclusion exactly?
"The count of values ≤m that are divisible by all primes ≤i is ⌊m∏j=ij=1j⋅isPrime(j)⌋"
For any value to be divisible by all the primes $$$\leq i$$$ it has to be divisible by their product ($$$prod=\prod_{j=1}^{j=i}{j\cdot isPrime(j)}$$$). These values are: $$$prod, 2\cdot prod, 3\cdot prod, ..., k\cdot prod$$$, where $$$k$$$ is the maximum integer such that $$$k\cdot prod\leq m$$$. So, $$$k=\lfloor \dfrac{m}{prod}\rfloor$$$
There is nothing to prove bro.
The $$$u=v$$$ case of problem F is very similar to Sprinkler from JOI Spring Camp 2022. Dealing with $$$u \neq v$$$ is pretty easy.
Almost solveD...
Just avoid overflow issue and now get passed. Feel sad...
I have a two pointers solution for B here.
I'm not sure if it works 100% of the time as edu round pretests are usually weak. Can anyone try and hack it?
keep the largest b value greedy. ur right
Problem C can be solved in O(n*logn*logn) with binary search + multiset. Why were the constraints kept low for this problem? Even a O(n^3) solution will pass.
To throw you off??
I see many solutions of D calculating LCM directly. Idk how lcm is not overflowing. I was thinking of same approach but I thought lcm will overflow.
we will calculate it until lcm<=m (m<=1e12) .if lcm>m there will always be ambigous solution.
so there wont be overflow
Anyone else overcomplicate B?
According to me , in problem B the testcase given below should have answer 244 but the correct answer is 242. Can anyone please explain why?
Also when we remove an element do we have to add its strength to both of its neighbors or just one of them?
You add to both neighbors if it has two neighbors. Also, to get 242, u remove the left 3 then remove the right 3 and then remove middle.
remove the elements in this order 7 5 6 1 2 3 4
you will get 242
we should add its strength to both neighbors
The answer is the sum of all $$$a[i]$$$ and $$$b[i]$$$ minus the largest $$$b[i]$$$.
This is because you can always kill an enemy who is currently at either the left end or the right end, so the $$$b[i]$$$ gets added only to one enemy, and never to two enemies. The last enemy you kill will not have their $$$b[i]$$$ added, so the optimal choice is for the maximum $$$b[i]$$$ to be the last enemy to kill.
So one solution here is to kill the first three from left to right (each $$$b[i]$$$ gets added to only one enemy), then the last three from right to left (again, only added to one enemy), and then finally kill the remaining element (which has the maximum $$$b[i]$$$, but it is not added to anyone's health, due to being killed last).
My head hurts after getting WA 5 times in a row on C
after getting WA on test 3 *
no shit man I got WA on test 2 5 times
D is too easy for its position.
Nice A And B is also nice but not as good as A
Was A really a good problem? All you had to check was n == m.
Really good.
Yes, it's a good problem imo. It's not an arbitrarily contrived scenario, nor is it a trivial problem (it may not be immediately obvious as to why $$$n != m$$$ suffices). The simplicity of the solution only makes it better.
How to solve problem D I was only able to come across observation that For an array to be not ambiguous It's Ith element must not be divisible by 2,3,...i for all I >= 2 (1 based indexing) How to approach this question further I am goota stuck here New methods are also appreciated Thanks in advance
You are pretty close. Just think about is there any array with zero or one removal sequence?
At least two is equivalent to excluding zero and one.
Plese help me To become lowest contributor in codeforces history
Can someone explain D ??
Copied from my other post:
D. Counting Arrays
Let's fix a length $$$\ell$$$. There will be $$$m^\ell$$$ total arrays. Now let's count the number of arrays that are not ambiguous, then subtract it from the total. The array is not ambiguous if and only if, for every prime number $$$p$$$ and index $$$i \ge p$$$, $$$p \mid a_i$$$. Hence at each index, the value must be divisible by the product of all prime numbers less than or equal to the index.
So for each index $$$i$$$, maintain a prefix product of primes at most $$$i$$$, and the number of possible values you can put at that index is $$$\lfloor m/(\prod_{p \le i \text{ and } p \text{ is prime}}p)\rfloor$$$. Take the product of this for all $$$1 \le i \le \ell$$$ for the number of not ambiguous arrays of size $$$\ell$$$.
For the final answer, take the sum of the number of total arrays minus not ambiguous arrays of size $$$\ell$$$ over all $$$1 \le \ell \le n$$$.
Thank u
can you explain how you are calculating the
number of k from 1 ... m for an i such that , GCD(k, p: p | i) > 1
177174924 do this one is ok?
I got 66th place (and 18th for rated participants); pretty good!
My solutions (video):
A. Cowardly Rooks
Note that if $$$m < n$$$, one of the columns will be empty. You can simply keep a rook in the same row, and move it to the empty column. So the answer is "YES."
If $$$m = n$$$, every cell in the grid will be attacked, so there's no way to move a rook. The answer is "NO."
B. Death's Blessing
At least one value of $$$b_i$$$ will not be added to the final answer because the final monster you remove will not have any neighbors. Since you are minimizing the amount of time, you should choose the greatest $$$b_i$$$ to be this value. Each other value of $$$b_i$$$ will be added to the total time once, by deleting the monsters from the edges of the array. The answer is $$$\sum_{i=1}^n(a_i+b_i) - \max_{i=1}^n(b_i)$$$.
C. Number Game
Note that whenever it is Bob's turn, after he adds $$$k - i + 1$$$ to any element, it becomes unusable for Alice. Since the smallest elements in the array are the most available, Bob's best move is to choose the smallest element of the array during his turn. Since Alice has to remove the element 1 at the end, and Bob has $$$k - 1$$$ chances to remove elements (because Bob's last turn doesn't matter), there needs to be at least $$$k$$$ 1's in $$$a$$$. Now, Bob's best move is to remove $$$k - 1$$$ 1's, so all you need to do is check if $$$a$$$ has an additional element $$$\le 2$$$, $$$\le 3$$$, ..., and $$$\le k$$$. To find the maximum value of $$$k$$$, you can use binary search, or even linear search, since $$$n$$$ is small.
D. Counting Arrays
Let's fix a length $$$\ell$$$. There will be $$$m^\ell$$$ total arrays. Now let's count the number of arrays that are not ambiguous, then subtract it from the total. The array is not ambiguous if and only if, for every prime number $$$p$$$ and index $$$i \ge p$$$, $$$p \mid a_i$$$. Hence at each index, the value must be divisible by the product of all prime numbers less than or equal to the index.
So for each index $$$i$$$, maintain a prefix product of primes at most $$$i$$$, and the number of possible values you can put at that index is $$$\lfloor m/(\prod_{p \le i \text{ and } p \text{ is prime}}p)\rfloor$$$. Take the product of this for all $$$1 \le i \le \ell$$$ for the number of not ambiguous arrays of size $$$\ell$$$.
For the final answer, take the sum of the number of total arrays minus not ambiguous arrays of size $$$\ell$$$ over all $$$1 \le \ell \le n$$$.
E. Cactus Wall
This problem statement boils down to finding if there is a "cactus path" from the leftmost column to the rightmost column. Construct a graph, where each cell is a node. Delete all the nodes with cells that you can't place a cactus on initially.
Color the grid using a checkerboard pattern. A path using the minimum number of cacti will require adding cacti to only black cells, or only white cells. So connect an edge from $$$(r, c)$$$ to $$$(r \pm 1, c \pm 1)$$$. Each edge will have a weight of $$$0$$$ if there is already a cactus at the node the edge is pointing to. If not, the weight will be $$$1$$$.
Now, add all cells in the leftmost column to a double-ended queue. Make sure all the cells that initially have a cactus on it have a distance initialized to $$$0$$$, and all empty cells have distance a initialized to $$$1$$$. Also make sure that the leftmost cells with distance $$$0$$$ come before the leftmost cells with distance $$$1$$$.
Use Dijkstra's algorithm, with an array storing which node you came from, so that you can reconstruct the path. Once Dijkstra's is done, iterate through the distances of the rightmost cells, and choose the cell with the least distance. You can reconstruct your "cactus path" from there.
For Problem B, the constraint on a is (1 <= a <= 10^9) but in the pre test 3 Test case 13 one of the input is 2453494488, this is not expeted right 1749B - Death's Blessing 177245405. Can anyone confirm is this fine
That's a number you're outputting; it's not from the input.
No a and b are array inputs for this problem. If you see my attached submission i tried to print that input to verify if it's below 10^9.
You're modifying $$$a$$$ before printing it out though. I put assert statements in my submission to make sure no inputs go above $$$10^9$$$.
Got it I was so dumb ;) thanks much
I FUCK EDUCATIONAL ROUND. Educational Codeforces Round 138 [Rated for Div. 2] was the worst match I've ever seen.Fuck you....................................
Sorry for posting such lenghty code instead of a link.
Never post code like this way. You should use link of the code.
Waste almost 30mins to solve Problem D just because of OVERFLOW. XD
Nice !
Can anyone please tell what is be wrong with this solution 177255794? I have tried to make sure that overflow doesn't occur anywhere, but still for the large test cases, I am getting a different output.
Edit: nvm, found my mistake
I don't know why people still don't use templates for modints. This can help you avoid adding additional logic for overflow and makes code look a lot cleaner. And for taking inverse you can just put a simple division operand and it will handle everything. Code and usage is in spoiler
Problem E is easy if you solved this recent Indian ICPC problem.
.
Is there a way to see the editorial or others submissions for this contest?
When will system test start?
when will the editorial be updated ?
What are you say?Fuck you................................
In question number Everyone approach is Sum(health)+sum(strength)-max(strength) But it fails on test case 4 2 6 7 3 3 6 1 5 its correct ouput is 28 But its outout from this approach is 27
are you sure????
It is the correct solution.
Yes I am sure 2 6 7 3 3 6 1 5 You can check by doing dry run
2 6 7 3 3 6 1 5
+2
9 7 3 6 1 5
+3
9 12 6 1
+12
10 6
+10
=27
Why not 27? You sure what? Lmao
just wait for the editorial and try to understand why the solution is like that.
I understand Sir...We have take only boundaries values so that sum can be minimize
then why don't just simulate this solution in your testcase?
Ratings updated preliminary. Unfortunately Mike won't be in touch until Monday, so cheaters will be removed later.
Problem A does not require rook positions [Lol] [Lol]
Please provide editorials....asap
Where's the editorial? Please post it as soon as possible.
Thanks for the contest! I really enjoyed upsolving(I had school when it was held T_T).
I think c,d are interesting and I like solving e a lot(even though I completely bricked it for the first 1 hour i spent solving it :P)
EDIT: Please disregard the comment. This is due to an issue in the way I printed the test.
There is potentially an issue in the tests of https://codeforces.net/contest/1749/problem/E It seems like the case 17 of the test 2 violates the constraint of no side adjacent cacti in the grid. However, in the statement it states that the initial state satisfies the constraint. Following is the test case. Am I missing something?
Auto comment: topic has been updated by awoo (previous revision, new revision, compare).
Never give up :)
Grats to you for finding out that
upper_bound
is essentially the same withlower_bound
with a very small value added. You tried to fool the plagiarism test by swapping out one function and making the indents a big mess. Good try, but I think you failed.UPD (Yes, it had to be a separate comment): Thank you, comment OP, for trying to conceal your sins and making me look like a motherfucker. Thank you very much.