Блог пользователя Recyclops

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

I couldn't find any discussion or editorial blog, so am creating a new one.
Was able to solve the 2nd problem with Segment Tree (standard stuff) and attempted to solve the first one with 6D DP (wasn't able to debug an RE).
It would be great if you could share your approaches for the four problems.

  • Проголосовать: нравится
  • +39
  • Проголосовать: не нравится

»
23 месяца назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Do you have questions? Can you post?

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    No, I don't have them. Thought they'd post an editorial or something.
    The IDE was kind of unintuitive as well, had to rewrite more than 60 lines of code because the text-area wouldn't accept any text that I had written in my editor.
    Plus there was a grey area around whether one could create additional functions and classes, or only the solution function could be modified.
    The testing apparatus could have also been better.
    Overall, not a great experience.

»
23 месяца назад, # |
Rev. 2   Проголосовать: нравится +8 Проголосовать: не нравится

First Problem___

You are given a range l,r. We have to count total no of beautiful numbers in that range.

Numbers are beautiful if, xor of all digits of that number is greater than the avg(max digit of that number + min digit of that number)

return result%1e9+7

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    Could someone share the intended approach / an elegant solution?
    Mine was clumsy and complicated and full of bugs; 6D DP's never a good idea, ig.

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +2 Проголосовать: не нравится

      Mine was 5D digit dp and cancerous as it involved finding the maximum and least set bit of a number. The 6d digit dp is better and easier to write.

    • »
      »
      »
      23 месяца назад, # ^ |
      Rev. 3   Проголосовать: нравится +4 Проголосовать: не нравится
      Mine is 6D Dp. Could not find the bug in contest time Later found it.
    • »
      »
      »
      23 месяца назад, # ^ |
      Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

      I dont know what is wrong with my code, but I have used 5-D dp and used digit DP . Is there any silly mistake? or some wrong approach?

      int dp[18][10][10][16][2];
      const int N = 1e9+7;
      int get(int pos , string &s, int mx, int mn, int xr, int tgt){
          if(pos == 0) return (2*xr>mx+mn);
          if(dp[pos][mx][mn][xr][tgt] != 1) return dp[pos][mx][mn][xr][tgt];
      
          int ans = 0, ub = (tgt? (s[s.size()-pos]-'0'):9);
          for(int i = 0; i <= ub; i++){
              ans += get(pos-1,s,max(mx,i),min(mn,i),(xr^i),tgt&(i==ub));
              ans %=N;
          }
          return dp[pos][mx][mn][xr][tgt] = ans;
      }
      int Solution::solve(string A, string B){
          memset(dp,-1,sizeof(dp));
          int a = get(A.size(),A,0,9,0,1);
          memset(dp,-1,sizeof(dp));
          int b = get(B.size(),B,0,9,0,1);
          int mx=0,mn=9,xr=0;
          for(auto it: A)mx=max(mx,it-'0'),mn=min(mn,it-'0'),xr^=(it-'0');
          if(2*xr>mx+mn)ans--;
          return ans;
      }
      
      • »
        »
        »
        »
        23 месяца назад, # ^ |
          Проголосовать: нравится +4 Проголосовать: не нравится

        You need one more state known to handle cases where you haven't started building numbers. For example if you start building from the second digit and put a 0 there it is not considered a part of the number as it is a leading zero.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    What was the constraint over l and r

»
23 месяца назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Btw, what is the selection criteria for the Online Round?
Is it based on the number of problems solved, or quality of submissions, or something else?
There wasn't even a rankings page, afaik.

  • »
    »
    23 месяца назад, # ^ |
    Rev. 3   Проголосовать: нравится 0 Проголосовать: не нравится

    Ig first they will go with number of problems with 100% acceptance rate, then quality of code. Then rest others.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    I think they take various factors into account like your resume quality, college, CGPA, branch etc.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    its based on points scored

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится +10 Проголосовать: не нравится

    I solved 3 and submitted 4th at the end (don't know if it was accepted or not as the test ended while compiling) and I got the clearance of Online Round mail.
    The ranking should be the only selection criteria I think which will be based on the points scored, in case of a tie, time will be considered.

    • »
      »
      »
      23 месяца назад, # ^ |
      Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

      From Interview-Bit or from Trilogy?
      I have received one for successful completion from Interview-Bit.
      Don't know if it's in any way related to the actual selection process.

»
23 месяца назад, # |
  Проголосовать: нравится +2 Проголосовать: не нравится

first was digit dp, second was usual prefix sum, third I did using Manachers algo.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    Manacher's Algo?
    Just found something new to learn.
    Thought of applying KMP or ZF in some way, but knew it wouldn't work.

  • »
    »
    23 месяца назад, # ^ |
    Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

    which one you did by prefix sum?the problem in which we need to give x1^x2 where x1 and x2 are bitwise and of range l1,r1,l2 and r2?

    If yes how?I did it using seg trees.

    upd got it from a below comment.

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +6 Проголосовать: не нравится

      Simple method: We have to find the values of x1 and x2 then xor them. So, first we create a prefix array pref[i][j] of size n*32 which represents the sum of jth bit of the indices in the range [0, i]. Now, when we want to calculate value of x1, we find the number of set bits of the jth bit and check if it is equal to r1 — l1 + 1 (size), because for bitwise and to be 1, we need all the bits to be equal to 1. Similarly, we can find x1 and x2, then xor them

  • »
    »
    23 месяца назад, # ^ |
    Rev. 3   Проголосовать: нравится +1 Проголосовать: не нравится

    Third problem can be also done by some simple string hashing and binary search.

    Make prefix and suffix hash using some mod value (I used 1e9+9). Now just binary search on the length uptil which the string [i,i+mid] from prefix hash and [i-mid,i] from suffix hash is same. This problem can be solved using binary search in log(n) time.

»
23 месяца назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

3rd Problem

Given a string s and a vector Q containing queries. For each element i in Q we have to find the length of longest odd palindrome whose middle index is i in string s. 1 <= Len(s), Len(Q) <= 1e5. We have to return vector of results of each query

»
23 месяца назад, # |
  Проголосовать: нравится +1 Проголосовать: не нравится

In second problem just store prefix of every bit for all element and check if bit[r][j] — bit[l-1][j] == r-l+1...if yes add that bit to answer. Calculate the same for both given range and store the xor of both!!

»
23 месяца назад, # |
  Проголосовать: нравится +13 Проголосовать: не нравится

Here are the problems:

problem 1
problem 2
problem 3
problem 4
»
23 месяца назад, # |
  Проголосовать: нравится +15 Проголосовать: не нравится

Here is my approaches...

Problem A
Problem B
Problem C
Problem D
  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится +7 Проголосовать: не нравится

    Just an addition, C was basically a direct implementation of Manacher's algo, which gives linear runtime.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    did u solution for the problem c got ac i got tle for a few tc.My code get 286/300.

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +6 Проголосовать: не нравится

      This was my code for 3rd problem and I got an AC for that.

      def solve(s, b):
          n = len(s)
          s = '('+s+')'
          p = [0]*(n+1)
          l = 1
          r = 1
          for i in range(1, n+1):
              p[i] = max(0, min(r - i, p[l + (r - i)]))
              while(s[i-p[i]] == s[i+p[i]]):
                  p[i] += 1
              if(i + p[i] > r):
                  l = i - p[i]
                  r = i + p[i]
          ans = []
          for i in b:
              ans.append(p[i]*2-1)
          return ans
      
      
      s = "abacaba"
      b = [2, 3, 4]
      print(solve(s, b))
      
    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +4 Проголосовать: не нравится

      Yeah, my binary search + hashing solution gave 296/300 initially, then I removed the unecessary %mod operations during addition, i.e (a%m + b%m > m) ? (a%m + b%m — m) : (a%m + b%m). that was enough for my submission to then get a full score.

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    Can You Please share Your approach for problem C of binary search with hashing

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +3 Проголосовать: не нравится

      Sure! so I just hashed the string in both the directions i.e. forward and backward/reverse. Then for a query of 'i' I searched for the maximum length that has the center 'i' and is odd. i.e. maximum x for [i — x... i + x] is palindrome. you may check palindrome nature by taking hash between [i — x .... i] for reverse hash table and [i ... i + x] for forward hash table and just compare the result. They should be equal for a substring being palindrome!

  • »
    »
    23 месяца назад, # ^ |
    Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

    In C, in test cases where the density of palindromes is high, wouldn't the effective complexity of Polynomial Hashing with Verification on Match be O(N*(LogN)*N)?
    The probability of hash collision is extremely low, I know, especially if a large prime is used, but still, without verification wouldn't it be non-deterministic?

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      Ig test cases in problem C are weak

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится +6 Проголосовать: не нравится

      Usually if we do single hash for a string with (p = 31, mod = 1e9 + 7) it don't get collision for a string of length N <= 1e5 but may fail for N <= 1e6 (would need double hash there).

      And I optimised my code to work in O(N * log(N)) with some precomputations.

»
23 месяца назад, # |
  Проголосовать: нравится +3 Проголосовать: не нравится

Was anyone able to solve the 4th problem.. The expected value one? If yes please share your approach.

»
23 месяца назад, # |
Rev. 3   Проголосовать: нравится 0 Проголосовать: не нравится

Is solving 1 problem enough to get selection mail?

  • »
    »
    23 месяца назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    I solved 2 problems(digit-dp and seg-tree) and got a mail for CCAT(Aptitude Test).

    • »
      »
      »
      23 месяца назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      what was the criteria for the selection process.I solved 2nd, 3rd and 200 pts for 1st but no mail for interviews.

»
23 месяца назад, # |
Rev. 2   Проголосовать: нравится +1 Проголосовать: не нравится

As usual I fucked up CCAT yet another time!

Anyone who successfully cleared CCAT. Any tips?