PikMike's blog

By PikMike, history, 5 weeks ago, translation, In English,

985A - Chess Placing

Tutorial
Solution (Vovuh)

985B - Switches and Lamps

Tutorial
Solution (Vovuh)

985C - Liebig's Barrels

Tutorial
Solution (adedalic)

985D - Sand Fortress

Tutorial
Solution (PikMike)

985E - Pencils and Boxes

Tutorial
Solution (PikMike)

985F - Isomorphic Strings

Tutorial
Solution (adedalic)

985G - Team Players

Tutorial
Solution (adedalic)
 
 
 
 
  • Vote: I like it  
  • +80
  • Vote: I do not like it  

»
5 weeks ago, # |
Rev. 2   Vote: I like it +21 Vote: I do not like it

Okay, formulas got better.

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

In problem D, for n=5,H=2, why is [2,3] wrong? I seem to be missing something here

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it +29 Vote: I do not like it

    The pillars are infinite. So your example is [2, 3, 0, 0, ...]. The difference between 3 and 0 is beyond 1.

    • »
      »
      »
      5 weeks ago, # ^ |
      Rev. 2   Vote: I like it 0 Vote: I do not like it

      Ahh damn I knew I was missing something simple. Thanks!

»
5 weeks ago, # |
  Vote: I like it +5 Vote: I do not like it

Problem D binary search on h the height becomes h,h-1,h-2...2,1,0 and if h>H make it becomes H and change the height on the right too. if H=5 There are two case: 1. [9,8,7,6,5,4,3,2,1,0][ 5,6,7,6,5,4,3,2,1,0] Sum=(9+1)*9/2-(4+2+0...) 2. [8,7,6,5,4,3,2,1,0][5,6,6,5,4,3,2,1,0] Sum=(8+1)*8/2-(3+1+0..) find the smallest h that sum>=n Note: 1+3+..n=[(n+1)/2]^2=tmp 2+4+..n=tmp+n east to calculate

»
5 weeks ago, # |
Rev. 2   Vote: I like it +5 Vote: I do not like it
»
5 weeks ago, # |
  Vote: I like it +1 Vote: I do not like it

Can you please explain this- "Otherwise, for each i from 1 to n let's take no more than k smallest staves from this segment in the i-th barrel, but in such way, that there are at least n - i staves left for next barrels."

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Every barrel will have atleast one stave from the range [1,rg) otherwise there will be two barrels whose volume difference will be greater than l.

    So for each barrel i from 1 to n, start picking up the staves (smallest first) from [1,rg) until you reach the required number of staves (k) or if there there are fewer than n-i staves left in the range [1,rg) because you still need to build n-i barrels and each of these needs atleast one stave from this range

    • »
      »
      »
      5 weeks ago, # ^ |
      Rev. 3   Vote: I like it 0 Vote: I do not like it

      i didn't get the logic for selecting n staves in the range[1,rg) to get the maximum sum of volume can we use this logic

      n, k, l = map(int, input().split()) a = sorted(list(map(int, input().split())))

      p = 0 while p < len(a) and a[p] — a[0] <= l: p += 1 if p < n: print(0) exit()

      ans = 0 u = 0 for i in range(n): ans+=a[u] if u+k<=p-1: u=u+k else: u=p-1 #print(u) print(ans)

»
5 weeks ago, # |
Rev. 4   Vote: I like it 0 Vote: I do not like it

Correct me if I am wrong.

In the first example for problem C:

4 2 1

2 2 1 2 3 2 2 3

Isn't [1, 2], [2, 2], [2, 2], [3, 3] a valid solution with answer 8?

  • »
    »
    5 weeks ago, # ^ |
    Rev. 2   Vote: I like it +3 Vote: I do not like it

    l = 1, and the volume of your first barrel is equal to 1, and the volume of your fifth is 3,so abs(3-1) is definitely bigger than 1.

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

How the solution for 985A works? what principal/logic is this? I'm struggling to understand.

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it +3 Vote: I do not like it

    You have to first sort the positions of the pieces because in one move you can either move left or right. We need to sort the positions to get optimal answer. Now the there are N positions and N/2 pieces. These N/2 pieces can be on black position or on white position since we need to place them in same color. Black positions are odd numbers till N and white pieces are even numbers till N (BWBWBW....BW) . Now the number of steps required by each piece is the absolute difference of required position where it should be and initial position of the piece given in the array. Thats what they have done for N/2 pieces. They are taking the minimum steps when placing in white position and black position.

    My Code : http://codeforces.com/contest/985/submission/38496335

»
5 weeks ago, # |
Rev. 2   Vote: I like it 0 Vote: I do not like it

For Problem C if there exists a configuration- Why only n-i barrels should be left for the barrels which appear next

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Not only n-i barrels solution says at least n-i barrels so that for remaining barrels at least one stave is available for making each barrel 'equally enough' with every other barrel.

»
5 weeks ago, # |
  Vote: I like it +8 Vote: I do not like it

In author solution of problem F, is the mt19337 random generator unpredictable enough?

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    If a seed of generator is fixed, any random generator will generate same sequence every time. So it is necessary to set a seed with diffenent values.

    On the other hand, as author, I would like to have a deterministic solution which will not depend on time it was started.

    • »
      »
      »
      5 weeks ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      I guess it is OK to use hash without unpredictable random in author solution (if the hash is strong enough, like the 8-modulo hash in author solution for F). No one beside author can see the solution and generate the anti-hash testcase until the end of the hacking phase.

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

i dont understanf why i am getting TLE in problem C my code

  • »
    »
    5 weeks ago, # ^ |
    Rev. 3   Vote: I like it +1 Vote: I do not like it

    Java's inbuilt Arrays.sort() uses a double pivot quick-sort whose worst case is O(n^2), So, it can be hacked by a testcase specially designed to break it. Some O(nlogn) sort like Merge sort or heap sort could be used or radix sort. You can also add a random shuffle before Arrays.sort or use Integer instead of int so that java's inbuilt Arrays.sort() over Objects(timsort) will be used.
    TLE with Arrays.sort()-38492535
    AC with Arrays.sort() over Integer- 38524250

    My Java solution also TLEd due to same reason :(

    • »
      »
      »
      5 weeks ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      thanxx man, got it!!! and does it only happen in educational rounds or in normal rounds we should take care of Arrays.sort()

      • »
        »
        »
        »
        5 weeks ago, # ^ |
          Vote: I like it +1 Vote: I do not like it

        Usually normal rounds are of small length, so this kind of hacks rarely happen. But creating a custom sort function which first randomly shuffles the array and then sorts or some other sort should be on the safer side.

»
5 weeks ago, # |
  Vote: I like it +4 Vote: I do not like it

Can Somebody explain DP part of Solution E

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Same question xD

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it +9 Vote: I do not like it

    Also, Necrozma

    Maybe the code of O(n2) solution will help a bit more?

    Code
    • »
      »
      »
      5 weeks ago, # ^ |
      Rev. 2   Vote: I like it 0 Vote: I do not like it

      Hey, by checking i+1-j>=k, you basically check if i and j can be taken in a segment, what if the jth is already taken in some previous segment and taking i,j together "disturbs" it?

      • »
        »
        »
        »
        5 weeks ago, # ^ |
        Rev. 2   Vote: I like it +7 Vote: I do not like it

        You update from dpj and that is the state for the first j elements being distributed. j-th element is the first one not taken.

        • »
          »
          »
          »
          »
          5 weeks ago, # ^ |
            Vote: I like it 0 Vote: I do not like it

          That was awesome !! Thanks a lot :) @PikMike

        • »
          »
          »
          »
          »
          4 weeks ago, # ^ |
            Vote: I like it 0 Vote: I do not like it

          Please can you explain the use of array f[N] . I am not able to understand.

    • »
      »
      »
      4 weeks ago, # ^ |
      Rev. 3   Vote: I like it 0 Vote: I do not like it

      Problem E. I dont understand why Time Limit is Exceeding in my code[submission:38589137]. I think time complexity of the problem is decent O(nlogn).

    • »
      »
      »
      4 weeks ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      Thank you very much. It helped me a lot.

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

can anybody help in understanding div 2 C,i havent got the idea n-i stacks to be left for each i??

  • »
    »
    5 weeks ago, # ^ |
    Rev. 2   Vote: I like it 0 Vote: I do not like it

    First, sort array in non-decreasing order.

    We must have a barrel which contains a1 stave. So each maximal total sum of volumes of barrel cannot be larger than a1 + l

    let C = (number of stave which has length  ≤ a1 + l) - N.

    If C < 0, then answer is 0.

    Otherwise, While picking staves, if C > 0, pick the smallest stave you have, and decrease C by 1. else, pick the largest stave you have.

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

How to solve F with LCP?

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

Someone please help me with this last line of code given in Ediorial for problem D.

printf("%lld\n", check(r) ? get(r) : get(l));

  • »
    »
    5 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    When check ==1 run get(r) function and that function returns a ll type , else call get (l) function and get another ll, this is basically a shorthand in c++

»
5 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

cannot wrap my head around picking staves in problem C. How do I know to pick which indices while iterating? For ex in first test case how do I know to pick 1 2 2 2?

»
4 weeks ago, # |
  Vote: I like it +3 Vote: I do not like it

PikMike, Can u give me any practice problem link similar to DP involved in Problem E?
P.S : I am new to DP.

  • »
    »
    4 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Huh, you know, this is such a basic dp usage. No particular problem comes to my head. I can recommend to search some dp trains on e-olymp, they have lots of great contests on certain topics.

»
4 weeks ago, # |
Rev. 2   Vote: I like it -29 Vote: I do not like it

I submitted using my solution but its giving TLE for test case 7. Can anyone suggest any optimization over my solution.

Soln...

include <bits/stdc++.h>

include<math.h>

define INF 2e9

define ll long long

using namespace std;

int bin[500001];

int binsearch(int* a,int low,int high,int allow){

if(low==high){
    return low;
}

int mid=(low+high+1)/2;

if(a[mid]<=allow)
    return binsearch(a,mid,high,allow);

return binsearch(a,low,mid-1,high);

}

int fn(int *a,int low,int high,int d,int k,int n){ if(low>high) return 0;

//cout<<"Start FN"<<endl;

if(high-low+1 <k){
    //cout<<"Error 1"<<endl;
    bin[low]=0;
    return 0;
}

if(bin[low]!=-1)
   return bin[low];

int index = binsearch(a,low,high,a[low]+d);
//cout<<"Index ="<<index<<endl;

if(index==n-1){
    bin[low]=1;
    return 1;
}

int elesize = index-low+1;

if(index-low+1<k){
    //cout<<"Error 2"<<endl;
    bin[low]=0;
    return 0;
}

if(fn(a,index+1,high,d,k,n)){
    bin[index+1]=1;
    return 1;
}

int l=low,r=index;
while(l<=r){
    if(l==r){

        if(fn(a,l,high,d,k,n)){
            bin[l]=1;
            return 1;
        }

    }

    int m=(l+r)/2;
    if(m-low+1>=k && fn(a,m+1,high,d,k,n)){
        bin[m+1]=1;
        return 1;
    }

    else if(m-low+1 < k){
        l=m;
        continue;
    }
    r=m-1;
}

//cout<<"Error 3"<<endl;
bin[low]=0;
return 0;

}

int main() { memset(bin,-1,sizeof(bin));

int n,k,d;cin>>n>>k>>d;

int a[n];

for(int i=0;i<n;i++)
    cin>>a[i];
sort(a,a+n);

if(fn(a,0,n-1,d,k,n))
    cout<<"YES"<<endl;

else
    cout<<"NO"<<endl;
return 0;

}

»
4 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

Well it is annoying that the third problem WA on test 13

»
4 weeks ago, # |
Rev. 2   Vote: I like it +3 Vote: I do not like it

in problem F. More over, if we sort all positions posis for all distict characters in s and sort all positions posjt for t, then posis must be equal posit for any i.

can someone expalain this?

  • »
    »
    2 weeks ago, # ^ |
      Vote: I like it +5 Vote: I do not like it

    If both strings are isomorphic, then the first occurrence of character X in S MUST be the first occurrence of some character Y in T.

    So if you sort the positions of first occurrences of distinct letters for S and T, they will be exactly same (only if they are isomorphic, also the reverse might not be true).

»
3 weeks ago, # |
  Vote: I like it 0 Vote: I do not like it

About problem 985E — Pencils and Boxes. I was thinking about proof. For "YES" condition its fairly simple. But what about "NO". Can'nt we get other segments. other than what is describe by DP solution.

  • »
    »
    2 weeks ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Usually, there is next idea behind this type of task: (solution exists)  →  (author solution finds answer) !(author solution finds answer)  →  !(solution exists).

»
10 days ago, # |
  Vote: I like it 0 Vote: I do not like it

for Problem C, Why is this going wrong ~~~~~ Arrays.sort(arr);

int limit = l+arr[0];
         int li=-1;

         for(int i=0;i<m;i++)
         {
          //pw.println(arr[i]);
          if(arr[i]>limit)
          {
              //pw.println(arr[i]);
              li=i; break;
          }
         }
       //li is the index such that all integers before are within range of l from arr[0] 
         if(li==-1)li=n;

         int noofsmallest = li;

         if(noofsmallest<n)
         {
          pw.println(0);
         }
         else
         {
          li--;
          long ans=0;
          for(int i=0;i<n-1;i++)
          {
              ans+=(long)arr[li];
              li--;
          }
          ans+=arr[0];
          pw.println(ans);

         }

~~~~~