Showing posts with label key design for anagram. Show all posts
Showing posts with label key design for anagram. Show all posts

Sunday, March 27, 2016

HackerRank: Sherlock and anagrams (II)

March 27, 2016

Problem statement:

Difficulty: Moderate

Summary of practice

This problem solving gets hot. Julia found something she struggled a lot. When Julia spent more than 2 hours on a problem in the Saturday evening, she knew that she is in trouble. She needs to be trained, and she needs a mentor.

Time Spent: March 26, 2016  Saturday evening 9:30 - 11:30
                                               Sunday morning  9:00 - 12:00 

Julia's practice is here.

Code Study

Let us get ideas how other people solve the problems, study the code. Julia is training herself thinking in C# using HackerRank:

1. Hash function design 

code source is provided by a person 19 Gold, unbelievable smart and quick/ fast / great expressive code. 


Code study - code is here

The anagram string function is composed to the design of key in the Dictionary. 

Julia added some comment above the hash function 

/*
precondition:
if two string are anagram, then key of these two strings should be the same

"ab" and "ba" are the anagram, key should be the same
"ab" and "bc" are not the anagram, so keys should not be the same.

Julia's comment: 701 is confusing, why it has to be this big number?
*/

int Fun(string s, int l, int r)
    {
        var ret = new int[26];
        for (int i = l; i <= r; i++)
            ret[s[i] - 'a']++;

        int x = 0;                  // Julia's comment: should be 1  
        for (int i = 0; i < 26; i++)
            x = x * 701 + ret[i];

        return x;
    }

Julia goes over the detail to check: 

Key is designed using math formula polynomial expression:
string a -> key is integer:  0
string b -> key:   1
string ab -> key:  x = 1
                            x =  1* 701 + 1
string ba -> 11 ->  key:  x = 1 * 701 + 1
string bc -> 011-> key:  x = 0 ,  count of a is 0
                             x = 1,   count of b is 1
                             x = 1 * 701 + 1
"ba" and "bc" are not anagram, so the key should be different: both are 1 * 701 + 1

string ad -> 1001 -> key x  = 1,               count of a is 1
                                        x = 701 + 0 ,    count of b is 0
                                        x = 701 *701 + 0
                               key = 701^3 + 1

Math or computer science

Julia found out that the idea can save a lot of time, she likes to work hard. But she is also "lazy" and likes to write less code. 

Julia changed the key design, and ran the code in HackerRank, it also passed the test cases. In Julia's opinion, the code has a bug in theory but pass the HackerRank test; so, Julia fixed the code anyway. 


Just practice! It is not a science of math, it is computer science. 

C# practice code is here.

Further code review on other things

Julia is still interested in writing loops, more expressive. Let us review how the code does:

public object Solve()
{
        for (int tt = ReadInt(); tt > 0; tt--)  // Julia's comment: put ReadInt() into a loop
        {
            string s = ReadToken();
            int n     = s.Length;
            int ans = 0;
            var count = new Dictionary<int, int>();

            // Julia's comment: substring length from 1 to n-1,
            for (int i = 1; i < n; i++)  
            {
                // substring start position - j, end position: j+i-1, and check j+i <=n, easy to reason - avoid bug
                for (int j = 0; j + i <= n; j++) 
                {
                    var key = Fun(s, j, j + i - 1);
                    if ( !count.ContainsKey(key) )
                    {
                        count[key] = 0;
                    }

                    count[key]++;
                }
            }

            foreach (var p in count)
            {
                ans += p.Value * (p.Value - 1) / 2;
            }

            writer.WriteLine(ans);
        }

        return null;
    }

701 prime number vs 26

One more step, improvement:  Failed. Number from 701 to 26, it does not work. It depends on the length of string, which is <=100. Julia tried to figure out some math, algebra, but she is sure that the number should be coefficient, so, 
at least >100. 

Julia, the key design for anagram string can be modified:  
/*
precondition:
if two string are anagram, then key of these two string should be the same

"ab" and "ba" are the anagram, key should be the same

"ab" and "bc" are not the anagram, so key should not be the same.

*/
int keyForAnagramString(string s, int l, int r)
{
        var ret = new int[26];
        for (int i = l; i <= r; i++)
            ret[s[i] - 'a']++;

        int x = 1;                  // Julia's comment: should be 1  

        for (int i = 0; i < 26; i++)
        {
            x = x * 26 + ret[i];
        }

        return x;
}

Julia spent 5 hacko to buy the test case input/ output


One more try - 101 

Because the string length is <=100, so that coefficient is less than 100.

Key design can be changed to a small number 701 to 101, it passes the HackerRank test:

/*
precondition:
if two string are anagram, then key of these two string should be the same

"ab" and "ba" are the anagram, key should be the same
"ab" and "bc" are not the anagram, so key should not be the same.
*/
int keyForAnagramString(string s, int l, int r)
{
        var ret = new int[26];
        for (int i = l; i <= r; i++)
            ret[s[i] - 'a']++;

        int x = 1;                  // Julia's comment: should be 1  

        for (int i = 0; i < 26; i++)
        {
            x = x * 101 + ret[i];
        }

        return x;
}

January 8, 2017

Come back to visit the blog, and then spent 10 - 20 minutes to work on layout, fixed grammar errors. 

HackerRank: Sherlocks and Anagram (III)

March 27, 2016

Problem statement:

Difficulty: Moderate

This problem solving gets hot. Julia found something she struggled a lot. When Julia spent more than 2 hours on a problem in the Saturday evening, she knew that she is in trouble. She needs to be trained, and she needs a mentor.



Solution to study:

https://gist.github.com/jianminchen/68453786a6ea03774a16

Julia, you should warm up with C# Dictionary<string, int>, and also StringBuilder class, AppendFormat function.

Let us review how to design anagram key more efficient, more understandable way:

/*
Think about how smart and easy it is to design this key for anagram string.

Precondition:
if two strings are anagram, the key should be same.
if two strings are not anagram, the key should be different.

Test case:
ab, the key is {0}-1{1}-1
ba, the key is {0}-1{1}-1,
but
bc, the key is {0}-0{1}-1{2}-1

compare to the design of anagram key using integer, this one is much easy to understand and follow.

http://juliachencoding.blogspot.ca/2016/03/hackerrank-string-sherlock-and-anagrams.html
*/
static string GiveKey(int[] arr){
        StringBuilder sb = new StringBuilder();
        for(int i = 0 ; i < 26 ; i++){
            sb.AppendFormat("{0}-",arr[i]);
        }
        return sb.ToString();
    }




HackerRank: String - Sherlock and anagrams (I)

March 27, 2016

Problem statement:

Difficulty: Moderate

This problem solving gets hot. Julia found something she struggled a lot. When Julia spent more than 2 hours on a problem in the Saturday evening, she knew that she is in trouble. She needs to be trained, and she needs a mentor.

Time Spent: March 26, 2016 Saturday evening 9:30 - 11:30
                                               Sunday morning  9:00 - 12:00 
Several mistakes to fix:
1. Julia, improve your analysis on test cases from HackerRank
2. Julia, understand Anagram requirement.
3. loop index issues

Julia's practice:

Go over test case again:
1. abba,

Let's say S[i, j] denotes the substring(i, j-i+1)
S[0,1] = "ab",
S[0,2]="abb",
S[1,1] = "b"
S[4,4] = "b"
S[1,2] = "ab"
S[3,4] = "ba"

For S = abba, anagrammatic pairs are:

{S[1,1], S[4,4]},     //
{S[1,2], S[3,4]},
{S{2,2], S{3,3]},
{S[1,3], S[2,4]}

Notice that substring can be selected by first char of string, choice of n-m, n is string length, m is substring length.

substrings can be overlapped, but still are anagrammatic pairs.
S[1,3] and S[2,4] are overlapped, but are the anagrammatic pair.

Sample test case "abba", output should be 4, but Julia got 3. Spent time to fix index error.

2. sample case: ifailuhkqq
should be: 3
Julia got 4
Actually, Julia, you should simplify the test case first.
it is the same as ifailqq,
also it is the same as ifilqq

How many anagramammatic pairs in "ifilqq"
"i","i" - S[1, 1] and S[3,3]
"if","fi" - S[1, 2] and S[2,3] <- warm up anagram definition: same chars with same counts
"q", "q"
but 2 'i' char in the string "ifilq",
  1 'i' char in the string "filqq",

Actually, the test case can be simplified using:
"abacdd", the pairs are (a,a), (ab, ba), (d,  d). That is very simple and understandable! 
"abacd" and "bacdd" are not anagrams, because two 'a' in first string, but 1 a in second string
so two strings are not anagram.

Julia, what you spent time on:
1. List<int> constructor issue - 10 minutes, List<int>(i)
2. 10 minutes to figure out that you should work on anagrams strings 
3. 10 minutes to write, 15 minutes to debug - Wrote a wrong anagram function
"ifilq" is not an anagram of "filqq", sample test case No. 2 should be 3

Julia is not very strong on testing software, she writes the software and puts it on. The software 
she writes can be improved tremendously, but she needs to find out what to improve. 

To be continued.