Dynammic Programming - dynamic-programming

Can anyone explain to me the solution to the Longest Common Subsequence Problem? Specifically, the recurrence relation is if(xi=yj) then answer= maxL(i-1, j-1) +1
else answer=Max{MaxL(i-1, j), MaxL(i, j-1)}
xi / yi is the letter in the constructed table. MaxL corresponds to the entry in the table constructed.
My question is why is the answer maxL(i-1,j-1) + 1? Why do we have to add from the upper left diagonal only when the letters match?
Thank you

(xi=yj) in recurrence relation means both the string have same characters at their respective current position.
Lets take a simple example:
Consider the input strings AGGTAB and GXTXAYB.
Last characters 'B' match for both the strings.[That is xi==yj condition holds].
So length of LCS can be written as:
LCS("AGGTAB", "GXTXAYB") = 1 + LCS("AGGTA", "GXTXAY")
LCS("AGGTA", "GXTXAY") is stored in Table[i−1,j −1] (ie. maxL[i-1][j-1])

Related

Count number of wonderful substrings

I found below problem in one website.
A wonderful string is a string where at most one letter appears an odd number of times.
For example, "ccjjc" and "abab" are wonderful, but "ab" is not.
Given a string word that consists of the first ten lowercase English letters ('a' through 'j'), return the number of wonderful non-empty substrings in word. If the same substring appears multiple times in word, then count each occurrence separately.
A substring is a contiguous sequence of characters in a string.
Example 1 :
Input: word = "aba"
Output: 4
Explanation: The four wonderful substrings are a , b , a(last character) , aba.
I tried to solve it. I implemented a O(n^2) solution (n is input string length). But expected time complexity is O(n). I could not solve it in O(n). I found below solution but could not understood it. Can you please help me to understand below O(n) solution for this problem or come up with an O(n) solution?
My O(N^2) approach - for every substring check whether it has at most one odd count char. This check can be done in O(1) time using an 10 character array.
class Solution {
public:
long long wonderfulSubstrings(string str) {
long long ans=0;
int idx=0; long long xorsum=0;
unordered_map<long long,long long>mp;
mp[xorsum]++;
while(idx<str.length()){
xorsum=xorsum^(1<<(str[idx]-'a'));
// if xor is repeating it means it is having even ouccrences of all elements
// after the previos ouccerence of xor.
if(mp.find(xorsum)!=mp.end())
ans+=mp[xorsum];
mp[xorsum]++;
// if xor will have at most 1 odd character than check by xoring with (a to j)
// check correspondingly in the map
for(int i=0;i<10;i++){
long long temp=xorsum;
temp=temp^(1<<i);
if(mp.find(temp)!=mp.end())
ans+=mp[temp];
}
idx++;
}
return ans;
}
};
There's two main algorithmic tricks in the code, bitmasks and prefix-sums, which can be confusing if you've never seen them before. Let's look at how the problem is solved conceptually first.
For any substring of our string S, we want to count the number of appearances for each of the 10 possible letters, and ask if each number is even or odd.
For example, with a substring s = accjjc, we can summarize it as: odd# a, even# b, odd# c, even# d, even# e, even# f, even# g, even# h, even# i, even# j. This is kind of long, so we can summarize it using a bitmask: for each letter a-j, put a 1 if the count is odd, or 0 if the count is even. This gives us a 10-digit binary string, which is 1010000000 for our example.
You can treat this as a normal integer (or long long, depending on how ints are represented). When we see another character, the count flips whether it was even or odd. On bitmasks, this is the same as flipping a single bit, or an XOR operation. If we add another 'a', we can update the bitmask to start with 'even# a' by XORing it with the number 1000000000.
We want to count the number of substrings where at most one character count is odd. This is the same as counting the number of substrings whose bitmask has at most one 1. There are 11 of these bitmasks: the ten-zero string, and each string with exactly one 1 for each of the ten possible spots. If you interpret these as integers, the last ten strings are the first ten powers of 2: 1<<0, 1<<1, 1<<2, ... 1<<9.
Now, we want to count the bitmasks for all substrings in O(n) time. First, solve a simpler problem: count the bitmasks for just all prefixes, and store these counts in a hashmap. We can do this by keeping a running bitmask from the start, and performing updates by an XOR of the bit corresponding to that letter: xorsum=xorsum^(1<<(str[idx]-'a')). This can clearly be done in a single, O(n) time pass through the string.
How do we get counts of arbitrary substrings? The answer is prefix-sums: the count of letters in any substring can be expressed as a different of two prefix-counts. For example, with s = accjjc, suppose we want the bitmask corresponding to the substring 'jj'. This substring can be written as the difference of two prefixes: 'jj' = 'accjj' - 'acc'.
In the same way, we want to subtract the counts for the two prefix strings. However, we only have the bitmasks telling us whether each letter has an even or odd frequency. In the arithmetic of bitmasks, we treat each position mod 2, so coordinate-wise subtraction becomes XOR.
This means counts(jj) = counts(accjj) - counts(acc) becomes
bitmask(jj) = bitmask(accjj) ^ bitmask(acc).
There's still a problem: the algorithm I've described is still quadratic. If, at every prefix, we iterate over all previous prefix-bitmasks and check if our mask XOR the old mask is one of the 11 goal-bitmasks, we still have a quadratic runtime. Instead, you can use the fact that XOR is its own inverse: if a ^ b = c, then b = a ^ c. So, instead of doing XORs with old prefix masks, you XOR with the 11 goal masks and add the number of times we've seen that mask: ans+=mp[xorsum] counts the substrings ending at our current index whose bitmask is xorsum ^ 0000000000 = xorsum. The loop after that counts substrings whose bitmask is one of the ten goal bitmasks.
Lastly, you just have to add your current prefix-mask to update the counts: mp[xorsum]++.

Codeforces -615C, main idea

I came across the explanation of task 615c on the editorial page of codeforces #338, div.2. i can't understand the idea: "The idea is that if can make a substring t[i, j] using k coatings, then we can also make a substring t[i + 1, j] using k coatings. So we should use the longest substring each time". How can it be implied from those sentences that we should use longest substrings each time? could you explain it more clearly? Here is the task: http://codeforces.com/contest/615/problem/C
Ayrat has scissors and glue. Ayrat is going to buy some coatings(basically is a string) s, then cut out from each of them exactly one continuous piece (substring) and glue it to the end of his track coating. Moreover, he may choose to flip this block before glueing it.
substring t[i + 1, j] contains exactly 1 character less compared to substring t[i, j]. So if you can cut out substring t[i, j] from k coatings, then you can surely cut out substring t[i+1, j].
Lets say you are provided with coating abc, then you can form abc + cba = abccba using k(=2) coatings.
That also means you can form bccba (even ccba) using 2 coatings.
You just need to cut out one extra character(first char) from any one of k coatings.
Instead, reducing more character(s) from the string-formed-by-k-coatings should raise the reverse question "Can we do it in k-1 or lesser number of coatings supply?"
So we should use the longest substring each time
When you take the longest substring each time, the number of coatings decreases, which is the goal of the problem.

Minimum Character that needed to be deleted

Original Problem:
A word was K-good if for every two letters in the word, if the first appears x times and the second appears y times, then |x - y| ≤ K.
Given some word w, how many letters does he have to remove to make it K-good?
Problem Link.
I have solved the above problem and i not asking solution for the above
problem
I just misread the statement for first time and just thought how can we solve this problem in linear line time , which just give rise to a new problem
Modification Problem
A word was K-good if for every two consecutive letters in the word, if the first appears x times and the second appears y times, then |x - y| ≤ K.
Given some word w, how many letters does he have to remove to make it K-good?
Is this problem is solvable in linear time , i thought about it but could not find any valid solution.
Solution
My Approach: I could not approach my crush but her is my approach to this problem , try everything( from movie Zooptopia)
i.e.
for i range(0,1<<n): // n length of string
for j in range(0,n):
if(i&(1<<j) is not zero): delete the character
Now check if String is K good
For N in Range 10^5. Time Complexity: Time does not exist in that dimension.
Is there any linear solution to this problem , simple and sweet like people of stackoverflow.
For Ex:
String S = AABCBBCB and K=1
If we delete 'B' at index 5 so String S = AABCBCB which is good string
F[A]-F[A]=0
F[B]-F[A]=1
F[C]-F[B]=1
and so on
I guess this is a simple example there can me more complex example as deleting an I element makens (I-1) and (I+1) as consecutive
Is there any linear solution to this problem?
Consider the word DDDAAABBDC. This word is 3-good, becauseDandCare consecutive and card(D)-card(C)=3, and removing the lastDmakes it 1-good by makingDandCnon-consecutive.
Inversely if I consider DABABABBDC which is 2-good, removing the lastDmakes CandBconsecutive and increases the K-value of the word to 3.
This means that in the modified problem, the K-value of a word is determined by both the cardinals of each letter and the cardinals of each couple of consecutive letters.
By removing a letter, I reduce its cardinal of the letter as well as the cardinals of the pairs to which it belongs, but I also increase the cardinal of other pair (potentially creating new ones).
It is also important to notice that if in the original problem, all letters are equivalent (I can remove any indifferently), while it is no longer the case in the modified problem.
As a conclusion, I think we can safely assume that the "consecutive letters" constrain makes the problem not solvable in linear time for any alphabet/word.
Instead of finding the linear time solution, which i think doesn't exist (among others because there seem to be a multitude of alternative solutions to each K request), i'd like to preset the totally geeky solution.
Namely, take the parallel array processing language Dyalog APL and create these two tiny dynamic functions:
good←{1≥⍴⍵:¯1 ⋄ b←(⌈/a←(∪⍵)⍳⍵)⍴0 ⋄ b[a]+←1 ⋄ ⌈/|2-/b[a]}
make←{⍵,(good ⍵),a,⍺,(l-⍴a←⊃b),⍴b←(⍺=good¨b/¨⊂⍵)⌿(b←↓⍉~(l⍴2)⊤0,⍳2⊥(l←⍴⍵)⍴1)/¨⊂⍵}
good tells us the K-goodness of a string. A few examples below:
// fn" means the fn executes on each of the right args
good" 'AABCBBCB' 'DDDAAABBDC' 'DDDAAABBC' 'DABABABBDC' 'DABABABBC' 'STACKOVERFLOW'
2 3 1 2 3 1
make takes as arguments
[desired K] make [any string]
and returns
- original string
- K for original string
- reduced string for desired K
- how many characters were removed to achieve deired K
- how many possible solutions there are to achieve desired K
For example:
3 make 'DABABABBDC'
┌──────────┬─┬─────────┬─┬─┬──┐
│DABABABBDC│2│DABABABBC│3│1│46│
└──────────┴─┴─────────┴─┴─┴──┘
A little longer string:
1 make 'ABCACDAAFABBC'
┌─────────────┬─┬────────┬─┬─┬────┐
│ABCACDAAFABBC│4│ABCACDFB│1│5│3031│
└─────────────┴─┴────────┴─┴─┴────┘
It is possible to both increase and decrease the K-goodness.
Unfortunately, this is brute force. We generate the 2-base of all integers between 2^[lenght of string] and 1, for example:
0 1 0 1 1
Then we test the goodness of the substring, for example of:
0 1 0 1 1 / 'STACK' // Substring is now 'TCK'
We pick only those results (substrings) that match the desired K-good. Finally, out of the multitude of possible results, we pick the first one, which is the one with most characters left.
At least this was fun to code :-).

Minimum number of characters to be inserted in a string to convert it to palindrome

I need to find the minimal number of insertions needed to convert a string into a palindrome. Note: the insertions can happen at any place, at the end, or within. If it was only at the end, we have a question here.
So I found out that this can be done in O(N**2) time by this simple trick:
Let the string be s1. Reverse it. Let it be s2. Say the length is l.
Now find the longest common subsequence of s1 and s2. Let its length be x.
The answer is l-x.
For example, suppose s1 = abcda. Therefore s2 = adcba. Length is 5. Longest common subsequence is aba of length 3. So the minimal number of insertions is 5-3 = 2, which is the actual answer, with the resulting string - adcbcda.
However, I cannot understand the logic behind it. Can anyone explain it to me why it works?
And, is there any O(N) solution possible for this?
I don't know whether there is a O(N) solution but by comparing with the reverse, you find a subsequence which is a palindrome. Then you have l-x letters that are not paired. (You can consider a letter's pair as its reflection if you have a mirror right at the middle of the word. e.g. ab|ba) Later, by insertions you just complete the picture.
Now,firstly, how do we find a (maximum)subsequence that is common to two strings? There is a polynomial algorithm for finding it see it here
https://en.wikipedia.org/wiki/Longest_common_subsequence_problem
When we try to find the longest common subsequence(lcs) between s1 and s2(reverse of s1) we actually find lcs between the first half of s1 and first half s2, also second half of s1 and second half of s2.
Assume
s1 = abcddzac
so s2 = cazddcba. Here we can see it as comparison of abcd with cazd(first half) plus comparison of dzac with dcba(second half). We can see that both of comparisons are the same except they are reverse of each other so their concatenation has to be palindrome, so lcs of s1 and s2 has to be palindrome.
Once we have the lcs(ad|da) which is of length 4, we have 4 more letters that break the symmetry(b,c,z,c). Then we insert one letter for each of them to make a symmetry, i.e. a palindrome. We set our middle point as the middle point of the lcs and consider that we break s1 into two from that middle point so we have
s1 = a bc d|d z a c and we break it like a stick into two from d|d and we end up with:
dzac
dcba
now we simply fill between the letters of lcs so that they are the same. In our case steps are as follows:
dzac
dcba
dzac
dzcba
dzcac
dzcba
dzcbac
dzcba
dzcbac
dzcbac
Now we unbreak it from the same point and we have
cabczddzcbac which is a palindrome.
Note: cddc is also an ldc but that doesn't change the number of steps.

Finding length of substring

I have given n strings . I have to find a string S so that, given n strings are sub-sequence of S.
For example, I have given the following 5 strings:
AATT
CGTT
CAGT
ACGT
ATGC
Then the string is "ACAGTGCT" . . Because, ACAGTGCT contains all given strings as super-sequence.
To solve this problem I have to know the algorithm . But I have no idea how to solve this . Guys, can you help me by telling technique to solve this problem ?
This is a NP-complete problem called multiple sequence alignment.
The wiki page describes solution methods such as dynamic programming which will work for small n, but becomes prohibitively expensive for larger n.
The basic idea is to construct an array f[a,b,c,...] representing the length of the shortest string S that generates "a" characters of the first string, "b" characters of the second, and "c" characters of the third.
My Approach: using Trie
Building a Trie from the given words.
create empty string (S)
create empty string (prev)
for each layer in the trie
create empty string (curr)
for each character used in the current layer
if the character not used in the previous layer (not in prev)
add the character to S
add the character to curr
prev = curr
Hope this helps :)
1 Definitions
A sequence of length n is a concatenation of n symbols taken from an alphabet .
If S is a sequence of length n and T is a sequence of length m and n m then S is a subsequence of T if S can be obtained by deleting m-n symbols from T. The symbols need not be contiguous.
A sequence T of length m is a supersequence of S of length n if T can be obtained by inserting m-n symbols. That is, T is a supersequence of S if and only if S is a subsequence of T.
A sequence T is a common supersequence of the sequences S1 and S2 of T is a supersequence of both S1 and S2.
2 The problem
The problem is to find a shortest common supersequence (SCS), which is a common supersequence of minimal length. There could be more than one SCS for a given problem.
2.1 Example
S= {a, b, c}
S1 = bcb
S2 = baab
S3 = babc
One shortest common supersequence is babcab (babacb, baabcb, bcaabc, bacabc, baacbc).
3 Techniques
Dynamic programming Requires too much memory unless the number of input-sequences are very small.
Branch and bound Requires too much time unless the alphabet is very small.
Majority merge The best known heuristic when the number of sequences is large compared to the alphabet size. [1]
Greedy (take two sequences and replace them by their optimal shortest common supersequence until a single string is left) Worse than majority merge. [1]
Genetic algorithms Indications that it might be better than majority merge. [1]
4 Implemented heuristics
4.1 The trivial solution
The trivial solution is at most || times the optimal solution length and is obtained by concatenating the concatenation of all characters in sigma as many times as the longest sequence. That is, if = {a, b, c} and the longest input sequence is of length 4 we get abcabcabcabc.
4.2 Majority merge heuristic
The Majority merge heuristic builds up a supersequence from the empty sequence (S) in the following way:
WHILE there are non-empty input sequences
s <- The most frequent symbol at the start of non-empty input-sequences.
Add s to the end of S.
Remove s from the beginning of each input sequence that starts with s.
END WHILE
Majority merge performs very well when the number of sequences is large compared to the alphabet size.
5 My approach - Local search
My approach was to apply a local search heuristic to the SCS problem and compare it to the Majority merge heuristic to see if it might do better in the case when the alphabet size is larger than the number of sequences.
Since the length of a valid supersequence may vary and any change to the supersequence may give an invalid string a direct representation of a supersequence as a feasible solution is not an option.
I chose to view a feasible solution (S) as a sequence of mappings x1...xSl where Sl is the sum of the lengths of all sequences and xi is a mapping to a sequencenumber and an index.
That means, if L={{s1,1...s1,m1}, {s2,1...s2,m2} ...{sn,1...s3,mn}} is the set of input sequences and L(i) is the ith sequence the mappings are represented like this:
xi {k, l}, where k L and l L(k)
To be sure that any solution is valid we need to introduce the following constraints:
Every symbol in every sequence may only have one xi mapped to it.
If xi ss,k and xj ss,l and k < l then i < j.
If xi ss,k and xj ss,l and k > l then i > j.
The second constraint enforces that the order of each sequence is preserved but not its position in S. If we have two mappings xi and xj then we may only exchange mappings between them if they map to different sequences.
5.1 The initial solution
There are many ways to choose an initial solution. As long as the order of the sequences are preserved it is valid. I chose not to in some way randomize a solution but try two very different solution-types and compare them.
The first one is to create an initial solution by simply concatenating all the sequences.
The second one is to interleave the sequences one symbol at a time. That is to start with the first symbol of every sequence then, in the same order, take the second symbol of every sequence and so on.
5.2 Local change and the neighbourhood
A local change is done by exchanging two mappings in the solution.
One way of doing the iteration is to go from i to Sl and do the best exchange for each mapping.
Another way is to try to exchange the mappings in the order they are defined by the sequences. That is, first exchange s1,1, then s2,1. That is what we do.
There are two variants I have tried.
In the first one, if a single mapping exchange does not yield a better value I return otherwise I go on.
In the second one, I seperately for each sequence do as many exchanges as there are sequences so a symbol in each sequence will have a possibility of moving. The exchange that gives the best value I keep and if that value is worse than the value of the last step in the algorithm I return otherwise I go on.
A symbol may move any number of position to the left or to the right as long as the exchange does not change the order of the original sequences.
The neighbourhood in the first variant is the number of valid exchanges that can be made for the symbol. In the second variant it is the sum of valid exchanges of each symbol after the previous symbol has been exchanged.
5.3 Evaluation
Since the length of the solution is always constant it has to be compressed before the real length of the solution may be obtained.
The solution S, which consists of mappings is converted to a string by using the symbols each mapping points to. A new, initialy empty, solution T is created. Then this algorithm is performed:
T = {}
FOR i = 0 TO Sl
found = FALSE
FOR j = 0 TO |L|
IF first symbol in L(j) = the symbol xi maps to THEN
Remove first symbol from L(j)
found = TRUE
END IF
END FOR
IF found = TRUE THEN
Add the symbol xi maps to to the end of T
END IF
END FOR
Sl is as before the sum of the lengths of all sequences. L is the set of all sequences and L(j) is sequence number j.
The value of the solution S is obtained as |T|.
With Many Many Thanks to : Andreas Westling

Resources