How to add memoization in subset sum? - algorithm

I am trying to solve subset sum problem with recursive solution, but to make to make it a bit more efficient I am trying to put memoization in it. However the code without memoization gives correct solution but with memoization it doesn't work properly.
public int subsetSum(int num[], int idx, int expecedSum, int dp[]) {
if (expecedSum == 0) {
return 1;
}
else if (idx < 0 || expecedSum < 0) {
return 0;
}
else {
if (dp[expecedSum] == -1) {
int x = subsetSum(num, idx - 1, expecedSum, dp);
int y = subsetSum(num, idx - 1, expecedSum - num[idx], dp);
dp[expecedSum] = (x == 1 || y == 1) ? 1 : 0;
}
return dp[expecedSum];
}
}
public static void main(String args[]) {
Solution s = new Solution();
int num[] = new int[]{1, 2, 3, 4, 5, 6, 7};
int sum = 0;
int n = new Scanner(System.in).nextInt();
int dp[] = new int[n + 1];
for (int i = 0; i < dp.length; i++) {
dp[i] = -1;
}
dp[0] = 1;
s.subsetSum(num, num.length - 1, n, dp);
}
Can someone help me with why this is not working?
If I enter n = 14 then ideally dp[14] should contains 1 but it doesn't contain 1.

The sum is not sufficient to describe the state. The pair (sum, index) is. If you make dp an array of arrays of size (max_sum + 1) x num.length and apply memoization for a pair (idx, expectedSum) in the subsetSet method, it works.

It has been years but this may help someone today.
The idea is to take both the state - means once with including current element into the sum and once excluding current element into the sum. and then taking OR(||) for those values and store in the cache.
the code will look something like this -
#include <iostream>
#include <vector>
using namespace std;
bool sumExists(vector<int> & a, int cursum, int n, vector<vector<int>> &dp) {
if(n==0 && cursum != 0)
return 0;
if(cursum==0)
return 1;
if(dp[n][cursum] != -1) {
return dp[n][cursum];
}
int newsum = cursum - a[n-1];
bool returnval = false;
if(newsum>=0) {
dp[n][newsum] = sumExists(a, newsum, n-1, dp);
dp[n][cursum] = sumExists(a, cursum, n-1, dp);
returnval = dp[n][newsum] || dp[n][cursum];
} else {
dp[n][cursum] = sumExists(a, cursum, n-1, dp);
returnval = dp[n][cursum];
}
return returnval;
}
int main() {
// your code goes here
vector<int> a{2,0,7,8,10};
int n = a.size();
int sum = 11;
vector<vector<int>> dp(n+1, vector<int>(sum+1, -1));
for(int i=0;i<=n;++i) {
dp[i][0]=true;
}
for(int i=1;i<=sum;++i) {
dp[0][i]=false;
}
if(sumExists(a, sum, n, dp)) {
cout<<"YES"<<endl;
} else {
cout<<"NO"<<endl;
}
}

Related

Looking for a data structure to perform range element update efficiently

I currently have the following data structure:
class DataStructure {
public:
DataStructure(int n) : m_data(n, 0) {
}
void update(int i, int j, int value) {
for (int k = i; k <= j; ++k) {
m_data[k] = max(m_data[k], value);
}
}
void reset(int i) {
m_data[i] = 0;
}
int query(int i) {
return m_data[i];
}
private:
vector<int> m_data;
};
So what it does is rather simple:
Initially there is a vector of n integers initialised to zero.
update(i, j, value) updates the elements in the range [i, j] to be the max of the given value and their respective current value. The given value is in the range of [0, n].
reset(i) resets the value at index i to 0.
query(i) returns the value at index i.
I need to perform n updates, n resets and n query operations. Currently this code takes O(n*n) time, due to the update operation being O(n) in general.
I am wondering if there are some smart ways to improve this to O(n*log n) time (or better) for n updates, n resets and n query operations, while maintaining O(n) space complexity?
Thanks for #qwertman for the explanation here is an algorithm that should work
#include <iostream>
#include <cstdio>
using namespace std;
#define max(a, b) (a>b?a:b)
int tree[100005], lazy[100005];
void init(int idx, int l, int r){
if(l>r)
return ;
if(l==r){
tree[idx] = 0;
lazy[idx] = -1;
}
else {
tree[idx] = 0;
lazy[idx] = -1;
int mid = (l+r)/2;
init(2*idx, l, mid);
init(2*idx+1, mid+1, r);
}
}
// l and r is for internal use the range a-b has to be updated
void update(int idx, int l, int r, int a, int b, int val, bool isReset){
if(l>r || b<l || a>r){
return;
}
// printf("idx=%d l=%d r=%d a=%d b=%d val=%d\n",idx,l,r,a,b,val);
if(lazy[idx] != -1){
tree[idx] = max(tree[idx], lazy[idx]);
lazy[2*idx] = max(lazy[2*idx], lazy[idx]);
lazy[2*idx+1] = max(lazy[2*idx+1], lazy[idx]);
lazy[idx] = -1;
}
if(l>=a && r<=b){
// printf("updating\n");
tree[idx] = max(tree[idx], val);
if(isReset){
tree[idx] = val;
}
lazy[2*idx] = max(lazy[2*idx], val);
lazy[2*idx+1] = max(lazy[2*idx+1], val);
lazy[idx] = -1;
}
else {
int mid = (l+r)/2;
update(2*idx, l, mid, a, b, val, isReset);
update(2*idx+1, mid+1, r, a, b, val, isReset);
tree[idx] = max(tree[2*idx], tree[2*idx+1]);
}
}
int query(int idx, int l, int r, int a){
if(l>r || a<l || a>r){
return -1;
}
// printf("idx=%d l=%d r=%d a=%d\n",idx,l,r,a);
if(lazy[idx] != -1){
tree[idx] = max(tree[idx], lazy[idx]);
lazy[2*idx] = max(lazy[2*idx], lazy[idx]);
lazy[2*idx+1] = max(lazy[2*idx+1], lazy[idx]);
lazy[idx] = -1;
}
if(l==a && r==a){
// printf("----l=%d r=%d a=%d tree=%d\n",l,r,a,tree[idx]);
return tree[idx];
}
else {
int mid = (l+r)/2;
int left = query(2*idx, l, mid, a);
int right = query(2*idx+1, mid+1, r, a);
return max(left, right);
}
}
int main() {
// initializing everything to 0
init(1, 1, 10);
// updating range 1-4 with value 7
update(1, 1, 10, 1, 4, 7, false);
// query for 3 should result in 7
cout << query(1, 1, 10, 3) << endl;
// updating 3-3 with value 9
update(1, 1, 10, 3, 3, 9, false);
// should give 9
cout << query(1, 1, 10, 3) << endl;
// isReset is set to true, so the function will do a hard reset
update(1, 1, 10, 3, 3, 0, true);
// should give 0
cout << query(1, 1, 10, 3) << endl;
return 0;
}
you can run this code at http://ideone.com/Mkp4dQ
some useful links for learning segment tree with lazy propagation hackerearth
Geeksforgeeks

Find minimum cost of tickets

Find minimum cost of tickets required to buy for traveling on known days of the month (1...30). Three types of tickets are available : 1-day ticket valid for 1 days and costs 2 units, 7-days ticket valid for 7 days and costs 7 units, 30-days ticket valid for 30 days and costs 25 units.
For eg: I want to travel on [1,4,6,7,28,30] days of the month i.e. 1st, 4th, 6th ... day of the month. How to buy tickets so that the cost is minimum.
I tried to use dynamic programming to solve this but the solution is not giving me the correct answer for all cases. Here is my solution in Java :
public class TicketsCost {
public static void main(String args[]){
int[] arr = {1,5,6,9,28,30};
System.out.println(findMinCost(arr));
}
public static int findMinCost(int[] arr) {
int[][] dp = new int[arr.length][3];
int[] tDays = {1,7,30};
int[] tCost = {2,7,25};
for (int i = 0; i < arr.length; i++) {
for (int j = 0; j < 3; j++) {
if (j==0){
dp[i][j]= (i+1)*tCost[j];
}
else{
int c = arr[i]-tDays[j];
int tempCost = tCost[j];
int k;
if (c>=arr[0] && i>0){
for (k = i-1; k >= 0; k--) {
if (arr[k]<=c){
c = arr[k];
}
}
tempCost += dp[c][j];
int tempCostX = dp[i-1][j] + tCost[0];
tempCost = Math.min(tempCost,tempCostX);
}
dp[i][j] = Math.min(tempCost,dp[i][j-1]);
}
}
}
return dp[arr.length-1][2];
}
}
The solution doesn't work for {1,7,8,9,10} input, it gives 10 but the correct answer should be 9. Also, for {1,7,8,9,10,15} it give 13 but the correct is 11.
I have posted my solution not for other to debug it for me but just for reference. I was taken a bottom-up dynamic programming approach for this problem. Is this approach correct?
Let MC(d) denote the minimum cost that will pay for all trips on days 1 through d. The desired answer is then MC(30).
To calculate MC(d), observe the following:
If there's no trip on day d, then MC(d) = MC(d − 1).
As a special case, MC(d) = 0 for all d ≤ 0.
Otherwise, the minimum cost involves one of the following:
A 1-day pass on day d. In this case, MC(d) = MC(d − 1) + 2.
A 7-day pass ending on or after day d. In this case, MC(d) = min(MC(d − 7), MC(d − 6), …, MC(d − 1)) + 7.
And since MC is nondecreasing (adding a day never reduces the minimum cost), this can be simplified to MC(d) = MC(d − 7) + 7. (Hat-tip to Ravi for pointing this out.)
A 30-day pass covering the whole period. In this case, MC(d) = 25.
As you've realized, dynamic programming (bottom-up recursion) is well-suited to this.
For ease of coding, I suggest we start by converting the list of days into a lookup table for "is this a trip day?":
boolean[] isDayWithTrip = new boolean[31]; // note: initializes to false
for (final int dayWithTrip : arr) {
isDayWithTrip[dayWithTrip] = true;
}
We can then create an array to track the minimum costs, and populate it starting from index 0:
int[] minCostUpThroughDay = new int[31];
minCostUpThroughDay[0] = 0; // technically redundant
for (int d = 1; d <= 30; ++d) {
if (! isDayWithTrip[d]) {
minCostUpThroughDay[d] = minCostUpThroughDay[d-1];
continue;
}
int minCost;
// Possibility #1: one-day pass on day d:
minCost = minCostUpThroughDay[d-1] + 2;
// Possibility #2: seven-day pass ending on or after day d:
minCost =
Math.min(minCost, minCostUpThroughDay[Math.max(0, d-7)] + 7);
// Possibility #3: 30-day pass for the whole period:
minCost = Math.min(minCost, 25);
minCostUpThroughDay[d] = minCost;
}
And minCostUpThroughDay[30] is the result.
You can see the above code in action at: https://ideone.com/1Xx1fd.
One recursive solution in Python3.
from typing import List
def solution(A: List[int]) -> int:
if not any(A):
return 0
tickets = {
1: 2,
7: 7,
30: 25,
}
import sys
min_cost = sys.maxsize
size = len(A)
for length, price in tickets.items():
current_cost = price
idx = 0
last_day = A[idx] + length
while idx < size and A[idx] < last_day:
idx += 1
if current_cost > min_cost:
continue
current_cost += solution(A[idx:])
if current_cost < min_cost:
min_cost = current_cost
return min_cost
if __name__ == '__main__':
cases = {
11: [1, 4, 6, 7, 28, 30],
9: [1, 7, 8, 9, 10],
}
for expect, parameters in cases.items():
status = (expect == solution(parameters))
print("case pass status: %s, detail: %s == solution(%s)" %
(status, expect, parameters))
public class Main03v3
{
public static void main(String[] args)
{
int[] A = {1,7,8,9,10,15,16,17,18,21,25};
System.out.println("Traveling days:\r\n "+Arrays.toString(A));
int cost = solution(A);
System.out.println("\r\nMinimum cost is " + cost);
System.out.println("\r\n" + new String(new char[40]).replace("\0", "-"));
}
public static int solution(int[] A)
{
if (A == null) return -1;
int sevenDays = 7;
int dayCost = 2, weekCost = 7, monthCost = 25;
int ratio_WeekAndDays = weekCost / dayCost;
int len = A.length;
if (len == 0) return -1;
if (len <= 3) return len * dayCost;
int cost[] = new int[len];
int i = 0;
while (i < len)
{
int startIdx = i, endIdx = i + 1;
while (endIdx < len && A[endIdx]-A[startIdx] < sevenDays)
endIdx++;
if (endIdx-startIdx > ratio_WeekAndDays)
{
if (endIdx >= startIdx + sevenDays)
endIdx = startIdx + sevenDays;
int j = startIdx;
cost[j] = ((j == 0) ? 0 : cost[j-1]) + weekCost;
while (++j < endIdx) {
cost[j] = cost[j-1];
}
i = j;
}
else
{
cost[i] = ((i == 0) ? 0 : cost[i-1]) + dayCost;
i++;
}
}
int finalCost = Math.min(cost[len-1], monthCost);
return finalCost;
}
}
Find minimum cost of tickets in JavaScript
case 1 : if input is [1,7,8,9,10] then the required output is 9
case 2 : if input is [1,7,8,9,10,15] then the required output is 11
function calMinCosts(arr){
if(!arr || arr.length===0)
return 0;
var len = arr.length;
var costsOfDateArr = Array.apply(null,{length:arr[len-1]+1}).map(()=>0);
var price1=2,price2=7,price3=25;
var days=7;
var index=0,n=costsOfDateArr.length;
for(var i=1;i<n;i++){
if(i===arr[index]){
if(i>=days+1){
costsOfDateArr[i] = Math.min(costsOfDateArr[i-days-1]+price2, costsOfDateArr[i-1]+price1);
}else{
costsOfDateArr[i] = Math.min(costsOfDateArr[0]+price2, costsOfDateArr[i-1]+price1);
}
index+=1;
}else{
costsOfDateArr[i] = costsOfDateArr[i-1];
}
}
return Math.min(price3,costsOfDateArr[n-1]);
}
console.log(calMinCosts([1,7,8,9,10]))
console.log(calMinCosts([1,7,8,9,10,15]))
Here is the C++ solution including print outs
#include <vector>
#include <iostream>
#include <cmath>
#include <algorithm>
int compute(std::vector<int> &A)
{
int sum[A.size()][A.size()+1];
for (int i = 0; i < A.size(); i++)
{
for(int j =0; j < A.size(); j++)
{
sum[i][j]=2;
}
}
for (int k = 0; k < A.size();k++)
{
sum[k][A.size()]=0;
}
for (int i = 0; i < A.size(); i++)
{
for(int j = 0; j < A.size(); j++)
{
if (i!=j)
{
if (sum[i][i] != 7)
{
int temp = abs(A[j]-A[i]);
if (temp<7 && abs(j-i)>=3)
{
sum[i][i]=7;
sum[i][j]=7;
if (i>j)
{
for(int k = j;k < i;k++)
sum[i][k]=7;
}
else
{
for(int k = i;k < j;k++)
sum[i][k]=7;
}
}
}
}
}
}
for (int i = 0; i < A.size(); ++i)
{
for(int j = 0; j < A.size(); ++j)
{
if (sum[i][j]==7)
{
sum[i][A.size()]+=1;
}
}
}
for (int i = 0; i < A.size(); ++i)
{
for (int j = 0; j < A.size()+1; ++j)
std::cout<<sum[i][j]<<" ";
std::cout<<std::endl;
}
int result = 0;
int row = A.size()-1;
int column = A.size()-1;
while(1)
{
int value = sum[row][A.size()];
if (value == 0)
value=1;
int temp = sum[row][column];
result += temp;
row = row-value;
column = column-value;
while (sum[row][column+1]==7 && row>=0)
{
row-=1;
column-=1;
result+=2;
}
if (row < 0)
break;
}
return result;
}
int solution(std::vector<int> &A) {
if (A.size() > 24)
return 25;
if (A.size() <= 3)
return A.size() * 2;
return std::min(25,compute(A));
}
int main()
{
std::vector<int> AA={1,2,3,4,5,29,30};
std::vector<int> B={1,2,3,4,5};
std::vector<int> A={1,2,3,4,5,9,10,11,12,13,14,17,18,20,21};
std::vector<int> C={1,2,3,12};
std::vector<int> D={1,2,3,4,12,13,14,15,29,30};
std::vector<int> DD={1,2,3,4,5,14,17,18,19,20,23,28,29,30};
std::vector<int> CC={1,2,3,4,5,6,7,9,14,17,18,19,20,23,28,29,30};
std::cout<<solution(AA)<<std::endl;
std::cout<<solution(D)<<std::endl;
std::cout<<solution(B)<<std::endl;
std::cout<<solution(A)<<std::endl;
std::cout<<solution(C)<<std::endl;
std::cout<<solution(DD)<<std::endl;
std::cout<<solution(CC)<<std::endl;
return 0;
}
Solved using the same approach of bottom-up dynamic programming. Here is the full solution :
public class PublicTicketCost {
public static void main(String args[]){
int[] arr = {1,7,8,9,10,15,16,17,18,21,25};
int[] tDays = {1,7,30};
int[] tCost = {2,7,25};
System.out.println(minCost(arr, tDays, tCost));
}
public static int minCost(int[] arr, int[] tDays, int[] tCost) {
int[][] dp = new int[arr.length][tDays.length];
for (int i = 0; i < arr.length; i++) {
for (int j = 0; j < tDays.length; j++) {
int prevDayIndex = findPrevDayIndex(arr,i,tDays,j);
int prevCost = prevDayIndex>=0 ? dp[prevDayIndex][tDays.length-1] : 0;
int currCost = prevCost + tCost[j];
if(j-1>=0){
currCost = Math.min(currCost, dp[i][j-1]);
}
dp[i][j] = currCost;
}
}
//print(dp);
return dp[arr.length-1][tDays.length-1];
}
private static void print(int arr[][]){
for (int i = 0; i < arr.length; i++) {
for (int j = 0; j < arr[0].length; j++) {
System.out.print(arr[i][j]+" ");
}
System.out.println();
}
}
private static int findPrevDayIndex(int[] arr, int i, int[] days, int j){
int validAfterDate = arr[i] - days[j];
if (validAfterDate<1){
return -1;
}
for (int k = i-1; k >= 0; k--) {
if (arr[k]<=validAfterDate){
return k;
}
}
return -1;
}
}
http://ideone.com/sfgxGo

Algorithm for Adding/Subtracting numbers to find if number can be made?

I was wondering if there is an efficient premade algorithm for determining if the sum/difference of a group of numbers can equal a different number. Example:
5, 8, 10, 2, using + or -, to equal 9.
5 - 8 = -3 + 10 = 7 + 2 = 9
If there is a preexisting algorithm, what is it called. If not, I can figure out how to program it, though it may not be efficient.
Thank you!
Yeah, this is basically knapsack problem, but it can be computed in pseudopolynomial time using dynamic programming.
I did it few month ago, so maybe this java code can help you, if you want to implement it :
public void solve() {
while (this.isEnd() == false) {
int priceSum = this.getItemsInstance().getTotalPrice()/divide;
int numOfItems = this.getItemsInstance().itemCount();
int maxWeight = this.getItemsInstance().getMaxWeight();
int[][] array = new int[numOfItems + 1][priceSum + 1];
boolean[][] arrayCounted = new boolean[numOfItems + 1][priceSum + 1];
for (int i = 0; i < numOfItems + 1; i++) {
array[i][0] = 0;
arrayCounted[i][0] = true;
}
int max = 0;
int price = 0;
for (int j = 1; j < priceSum + 1; j++) {
for (int i = 1; i < numOfItems + 1; i++) {
int temp = W(i, j, array, arrayCounted);
if (temp <= maxWeight) {
max = temp;
price = j;
}
}
}
}
}
private int W(int i, int c, int[][] array, boolean[][] arrayCounted) {
if (c < 0) {
return MAX_PRICE / divide;
}
if (i == 0) {
if (c == 0) {
return 0;
} else {
return MAX_PRICE / divide;
}
}
if (arrayCounted[i][c]) {
return array[i][c];
}
arrayCounted[i][c] = true;
array[i][c] = Math.min(W(i - 1, c, array, arrayCounted), W(i - 1, c - this.items[i - 1].price/divide, array, arrayCounted) + this.items[i - 1].weight);
return array[i][c];
}
its not an NP problem, if the problem is to find a given number from adding or subtracting each elements of a list/array. if you think about AP. here is a sample code in C++
int Np( int mn, list<int>a, int c )
{
int size = a.size(), rst = 0, maxI = 0;
std::list<int>::iterator it;
while( size > c )
{
a.sort();
maxI += a.back();
a.pop_back();
rst = 0;
for( auto ele : a )
{
rst += ele;
cout << rst << endl;
}
if( (rst - maxI) == mn or (maxI - rst) == mn or (maxI + rst) == mn )
{
return mn;
}
size--;
}
return rst;
}
this should help. i think.
I actually wrote a simple java program, I was not actually aware of knapsack strategies. This is my own solution. Hope this helps
import java.util.ArrayList;
import java.util.List;
public class Puzzle {
public static void main(String[] args) {
int targetNumber = 0;
int min = 2147483647;
int[] numbers = {-10, -30, -20, -50};
//int[] numbers = {0,0,0,0};
//int[] numbers = {7, 2, 10};
//int[] numbers = {1, 2, 3, 4, 5};
//int[] numbers = {1000, 2, 3, 4, 100};
char set[] = {'+', '-'};
min = getNumberClosestToTarget(numbers, set, min, targetNumber);
System.out.println(String.format(" %d is closest to %d", min, targetNumber));
}
private static int getNumberClosestToTarget(int[] numbers, char[] set, int min, int targetNumber) {
List<String> operators = new ArrayList<>();
computeAllOperatorsCombination(set, "", set.length, numbers.length - 1, operators);
for (String operatorString : operators) {
String[] ops = operatorString.split("");
int sum = computeSum(numbers, ops, numbers.length - 1);
min = getClosestToTarget(min, targetNumber, sum);
}
return min;
}
static int computeSum(int[] numbers, String[] operators, int index) {
int result = numbers[index];
if (index == 0) {
return result;
} else {
switch (operators[index - 1]) {
case "+":
return computeSum(numbers, operators, index - 1) + result;
case "-":
return computeSum(numbers, operators, index - 1) - result;
}
return result;
}
}
static void computeAllOperatorsCombination(char set[], String prefix, int n, int k, List<String> result) {
if (k == 0) {
result.add(prefix);
return;
}
for (int i = 0; i < n; i++) {
String newPrefix;
newPrefix = prefix + set[i];
computeAllOperatorsCombination(set, newPrefix, n, k - 1, result);
}
}
private static int getClosestToTarget(int min, int targetNumber, int r) {
int distance = Math.abs(targetNumber - r) < Math.abs(r - targetNumber) ? Math.abs(targetNumber - r) : Math.abs(r - targetNumber);
if (distance < Math.abs(min)) {
min = distance;
if (r < 0) {
min = -distance;
}
}
return min;
}
}

How is the FlowerGarden pr0blem on TopCoder a DP-one?

I'm reading this excellent tutorial by Dumitru on DP based problems here. And I'm trying to come up with a DP based approach for the FlowerGarden problem mentioned in the list of 1D DP problems.
I can only think of a non-DP solution that would involve initially sorting the flowers in an order and then reordering them based on different condition checks mentioned in the problem. That doesn't classify as DP, does it?
The editorial also doesn't mention anything about DP.
Could anyone, by any chance, point me to a proper DP-based solution to this problem?
Thanks!
Edit:
I didn't realize the link would require registration. This is the problem:
Problem Statement
You are planting a flower garden with bulbs to give you joyous flowers throughout the year. However, you wish to plant the flowers
such that they do not block other flowers while they are visible.
You will be given a int[] height, a int[] bloom, and a int[] wilt.
Each type of flower is represented by the element at the same index of
height, bloom, and wilt. height represents how high each type of
flower grows, bloom represents the morning that each type of flower
springs from the ground, and wilt represents the evening that each
type of flower shrivels up and dies. Each element in bloom and wilt
will be a number between 1 and 365 inclusive, and wilt[i] will always
be greater than bloom[i]. You must plant all of the flowers of the
same type in a single row for appearance, and you also want to have
the tallest flowers as far forward as possible. However, if a flower
type is taller than another type, and both types can be out of the
ground at the same time, the shorter flower must be planted in front
of the taller flower to prevent blocking. A flower blooms in the
morning, and wilts in the evening, so even if one flower is blooming
on the same day another flower is wilting, one can block the other.
You should return a int[] which contains the elements of height in the
order you should plant your flowers to acheive the above goals. The
front of the garden is represented by the first element in your return
value, and is where you view the garden from. The elements of height
will all be unique, so there will always be a well-defined ordering.
Edit two:
Example 1:
height={5,4,3,2,1}
bloom={1,1,1,1,1}
wilt={365,365,365,365,365}
Returns: { 1, 2, 3, 4, 5 }
These flowers all bloom on January 1st and wilt on December 31st. Since they all may block each other, you must order them from shortest to tallest.
Example 2:
h={5,4,3,2,1}
b={1,5,10,15,20}
w={4,9,14,19,24}
Returns: { 5, 4, 3, 2, 1 }
The same set of flowers now bloom all at separate times. Since they will never block each other, you can order them from tallest to shortest to get the tallest ones as far forward as possible.
Example 3:
height={5,4,3,2,1}
bloom={1,5,10,15,20}
wilt={5,10,14,20,25}
Returns: { 3, 4, 5, 1, 2 }
The difference here is that the third type of flower wilts one day earlier than the blooming of the fourth flower. Therefore, we can put the flowers of height 3 first, then the flowers of height 4, then height 5, and finally the flowers of height 1 and 2. Note that we could have also ordered them with height 1 first, but this does not result in the maximum possible height being first in the garden.
It's not a dynamic programming problem. It's a greedy algorithm problem.
This confused me too, since topcoder's own dynamic programming tutorial links to it as a practice problem in the “Elementary” section.
Sort the flowers by height, shortest to tallest. Start with an empty list of rows. For each flower (shortest to tallest), find the forward-most place where you can insert that flower such that it blocks no flowers behind it.
In Python:
def getOrdering(height, bloom, wilt):
flowers = zip(height, bloom, wilt)
flowers.sort()
def flowersOverlap(f1, f2):
# Overlap if each blooms before the other wilts.
return f2[1] <= f1[2] and f1[1] <= f2[2]
rows = [ ]
for flower in flowers:
rowIndex = len(rows)
# Start at the back and march forward as long as
# `flower` wouldn't block any flowers behind it.
while rowIndex > 0 and not flowersOverlap(flower, rows[rowIndex - 1]):
rowIndex -= 1
rows[rowIndex:rowIndex] = [flower]
return [flower[0] for flower in rows]
public int[] getOrdering(int[] height, int[] bloom, int[] wilt) {
int[] optimal = new int[height.length];
int[] optimalBloom = new int[bloom.length];
int[] optimalWilt = new int[wilt.length];
// init state
optimal[0] = height[0];
optimalBloom[0] = bloom[0];
optimalWilt[0] = wilt[0];
// run dynamic programming
for(int i = 1; i < height.length; i ++) {
int currHeight = height[i];
int currBloom = bloom[i];
int currWilt = wilt[i];
int offset = 0; // by default, type i is to be put to 1st row
for(int j = 0; j < i; j ++) {
if(currWilt >= optimalBloom[j] && currWilt <= optimalWilt[j] ||
currBloom >= optimalBloom[j] && currBloom <= optimalWilt[j] ||
currWilt >= optimalWilt[j] && currBloom <= optimalBloom[j]) { // life period overlap
if(currHeight < optimal[j]) { // life overlap, and type i is shorter than type j
offset = j;
break;
} else {
offset = j + 1; // type i overlap with type j, and i is taller than j. Put i after j
}
} else { // not overlap with current
if(currHeight < optimal[j]) {
offset = j + 1; // type i not overlap with j, i is shorter than j, put i after j
}
// else keep offset as is considering offset is smaller than j
}
}
// shift the types after offset
for(int k = i - 1; k >= offset; k -- ) {
optimal[k+1] = optimal[k];
optimalBloom[k+1] = optimalBloom[k];
optimalWilt[k+1] = optimalWilt[k];
}
// update optimal
optimal[offset] = currHeight;
optimalBloom[offset] = currBloom;
optimalWilt[offset] = currWilt;
}
return optimal;
}
This is my tested working code.
I'have been struggling with this exact question for a whole day, and also, i couldn't find any DP solution to it.
Here is my greedy approach in java, similar to others already posted, the key point is to proceed under a height ordering. The reason is to avoid dealing with intermediate heights (referring to the already computed), given that a intermediate height can change the relative order of the previously computed ones.
int[] height = new int[]{5, 3, 4};
int[] start = new int[]{1, 3, 1};
int[] end = new int[]{2, 4, 4};
System.out.println(Arrays.toString(new FlowerGarden().getOrdering(height, start, end)));
This is the only optimal substructure I could find. But given that there is no overlapping among subproblems, this algorithm should not be considered DP but greedy.
private static boolean intersects(final int[] starts, final int[] ends, int i1, int i2) {
return !(ends[i1] < starts[i2] || ends[i2] < starts[i1]);
}
public int[] getOrdering(final int[] height, final int[] starts, final int[] ends) {
PriorityQueue<Integer> minHeap = new PriorityQueue<Integer>(new Comparator<Integer>() {
public int compare(Integer i, Integer j) {
return Integer.compare(height[i], height[j]);
}
}
);
for (int i = 0; i < height.length; i++) {
minHeap.add(i);
}
LinkedList<Integer> list = new LinkedList<Integer>();
while (minHeap.size() > 0) {
Integer index = minHeap.poll();
int p = 1;
int pos = 0;
for (Integer i : list) {
if (intersects(starts, ends, i, index)) {
pos = p;
}
p++;
}
list.add(pos, index);
}
int[] ret = new int[height.length];
int j = 0;
for (Integer i : list) {
ret[j++] = height[i];
}
return ret;
}
BTW, the DP solutions I have seen posted here fail for this example.
Cheers
I tried to solve this problem too. The main idea of my approach is to build a tree where each child is overlaped at least once by its parent.
For example, if we have three flower types of heigths 4,2 and 1 growing and dying on the same days, then, the resulting tree should be:
On the other hand, if 4 and 2 and 4 and 1 live at the same time but 2 and 1 do not coexist then, the resulting tree should be:
That will generate a tree which is consistent with the problem contraints. Nonetheless the problem statement also includes a cost function making some solutions better than others.
...you also want to have the flowers in rows which are more towards the front be as tall as possible.
The way to project this preference into our tree is to order all "brothers" (all nodes sharing the same parent) from higher to lower. So 2 comes first than 1.
I built this tree using the following code:
#define INT_MOD(a,b) ((a<0)?(b+(a%b)):(a%b))
#define DIST(a,b) ((a-b>=0)?(a-b):(b-a))
//Prev: ForAll(i), bloom[i] < wilt[i]
inline bool isOverlap(vector<int> & bloom,
vector<int> & wilt,
vector<int> & height,
unsigned int idxPrev, unsigned int idxFollowing)
{
int f1A = bloom[idxPrev];
int f1B = wilt[idxPrev];
int f2A = bloom[idxFollowing];
int f2B = wilt[idxFollowing];
bool notIntersecting =
f2A > f1B /* --[--]-(--)-- */ ||
f1A > f2B /* --(--)-[--]-- */ ;
return height[idxPrev] > height[idxFollowing] && !notIntersecting;
}
class CPreference {
public:
static vector<int> * pHeight;
static bool preference(int a, int b)
{
return (*pHeight)[a] > (*pHeight)[b];
}
};
vector<int> * CPreference::pHeight = NULL;
vector<int> getOrdering(vector<int> height,
vector<int> bloom,
vector<int> wilt)
{
int l = height.size();
vector<int> state = vector<int>(l, -1); /* Tree where each leave points to its
parent. Being that parent the first
flower type that is forced to be
after (backwards) its children */
//This loop is the dynamic programming core.
for(int i = 0; i < l; i++)
for(int j = INT_MOD((i-1),l); j != i; j = INT_MOD((j-1),l))
{
if(isOverlap(bloom, wilt, height, i, j) &&
(state[j] < 0 || DIST(height[j],height[i]) < DIST(height[j], height[state[j]])))
{
state[j] = i;
}
}
vector<vector<int> > groups; //Groups of indexes overlapped by the element at the same index
for(int i = 0; i < l+1; i++)
groups.push_back(vector<int>()); // (l+1) for no overlapped indexes group.
for(int i = 0; i < l; i++)
{
int k = state[i];
if(k < 0) k = l;
groups[k].push_back(i);
}
CPreference::pHeight = &height;
for(vector<vector<int> >::iterator it = groups.begin(); it != groups.end(); it++)
sort(it->begin(),it->end(), CPreference::preference);
At this point, Each row (i) of groups contains, ordered from higher to lower, all flower types indexes that should be placed before the flower type of index i.
One last step is needed, to flatten groups into an output vector. That is, to build a vector where each element is followed by either:
Its parent on the tree.
It next brother when sorted by height.
That can be done by a depth visit of each node of group. I think that is the weak point of my solution. I had not so much time so I just made a naive recursive implementation:
//PRE: each vector, v, in 'groups' is sorted using CPreference
void flattenTree(vector<vector<int> > & groups, vector<int> & out, int currentIdx /*parent*/, int l)
{
int pIdx = currentIdx;
if(pIdx < 0) pIdx = l;
vector<int> & elements = groups[pIdx];
vector<int> ret;
for(vector<int>::iterator it = elements.begin(); it != elements.end(); it++)
{
flattenTree(groups, out ,*it, l);
}
if(currentIdx>=0)
out.push_back(currentIdx);
}
Which is used to completed getOrdering function:
vector<int> getOrdering(vector<int> height,
vector<int> bloom,
vector<int> wilt)
{
int l = height.size();
vector<int> state = vector<int>(l, -1); /* Tree where each leave points to its
parent. Being that parent the first
flower type that is forced to be
after (backwards) its children */
for(int i = 0; i < l; i++)
for(int j = INT_MOD((i-1),l); j != i; j = INT_MOD((j-1),l))
{
if(isOverlap(bloom, wilt, height, i, j) &&
(state[j] < 0 || DIST(height[j],height[i]) < DIST(height[j], height[state[j]])))
{
state[j] = i;
}
}
vector<vector<int> > groups; //Groups of indexes overlapped by the element at the same index
for(int i = 0; i < l+1; i++)
groups.push_back(vector<int>()); // (l+1) for no overlapped indexes group.
for(int i = 0; i < l; i++)
{
int k = state[i];
if(k < 0) k = l;
groups[k].push_back(i);
}
CPreference::pHeight = &height;
for(vector<vector<int> >::iterator it = groups.begin();
it != groups.end(); it++)
sort(it->begin(),it->end(), CPreference::preference);
vector<int> ret;
flattenTree(groups, ret, -1, l);
for(unsigned int i = 0; i < ret.size(); i++)
ret[i] = height[ret[i]];
return ret;
}
Please, let my know if you found a better solution or if know any way to improve mine.
package topcoders;
import java.util.ArrayList;
import java.util.List;
public class FlowerGarden {
public int[] getOrdering(int[] height, int[] bloom, int[] wilt) {
int[] order = new int[height.length];
List<Integer> heightList = new ArrayList<Integer>();
for (int i = 0; i < height.length; i++) {
heightList.add(height[i]);
}
heightList = quickSort(heightList);
for (int i = 0; i < height.length; i++) {
height[i] = heightList.get(i);
}
order = height;
for (int i = 0; i < order.length; i++) {
int j = 0;
while (j < order.length - 1
&& isBlocking(j + 1, j, order, bloom, wilt)) {
int placeHolder = order[j];
order[j] = order[j + 1];
order[j + 1] = placeHolder;
j++;
}
}
return order;
}
public boolean isBlocking(int isBlocked, int isBlocking, int[] order,
int[] bloom, int[] wilt) {
if (order[isBlocking] > order[isBlocked]
&& bloom[isBlocked] <= wilt[isBlocking]
&& wilt[isBlocked] >= bloom[isBlocking]) {
return true;
} else {
return false;
}
}
public List<Integer> quickSort(List<Integer> array) {
if (array.size() <= 1) {
return array;
}
int pivotIndex = array.size() / 2;
int pivot = array.get(pivotIndex);
List<Integer> less = new ArrayList<Integer>();
List<Integer> greater = new ArrayList<Integer>();
int l = 0;
int g = 0;
for (int i = 0; i < array.size(); i++) {
if (i == pivotIndex) {
continue;
} else if (array.get(i) >= pivot) {
less.add(array.get(i));
} else {
greater.add(array.get(i));
}
}
List<Integer> lessResult = quickSort(less);
List<Integer> greaterResult = quickSort(greater);
List<Integer> result = new ArrayList<Integer>();
result.addAll(lessResult);
result.add(pivot);
result.addAll(greaterResult);
return result;
}
public static void main(String[] args) {
int[] height = { 5, 4, 3, 2, 1 };
int[] bloom = { 1, 5, 10, 15, 20 };
int[] wilt = { 5, 10, 14, 20, 25 };
FlowerGarden g = new FlowerGarden();
List<Integer> arrayList = new ArrayList<Integer>();
int[] array = g.getOrdering(height, bloom, wilt);
for (int i = 0; i < array.length; i++) {
System.out.println(array[i]);
}
}
}
A toplogical sort approach:
#include<stdio.h>
#include<stdlib.h>
#include <vector>
#include <queue>
using namespace std;
#define MAX_FLOWERS 50
struct flower
{
int id;
int height;
int bloom;
int wilt;
bool visited;
int ind;
};
struct flower_comp
{
bool operator()(const struct flower* lhs, const struct flower* rhs) const
{
return rhs->height > lhs->height;
}
};
inline bool overlap(const struct flower& a, const struct flower& b)
{
return !((a.bloom < b.bloom && a.wilt < b.bloom) || (a.bloom > b.bloom && a.bloom > b.wilt));
}
void getOrdering(int height[], int bloom[], int wilt[], int size)
{
struct flower flowers[MAX_FLOWERS];
for(int i = 0; i < size; i++)
{
flowers[i].id = i;
flowers[i].height = height[i];
flowers[i].bloom = bloom[i];
flowers[i].wilt = wilt[i];
flowers[i].visited = false;
flowers[i].ind = 0;
}
bool partial_order[MAX_FLOWERS][MAX_FLOWERS] = {false};
for(int i = 0; i < size; i++)
{
for(int j = i + 1; j < size; j++)
{
if(overlap(flowers[i], flowers[j]))
{
if(flowers[i].height < flowers[j].height)
{
partial_order[i][j] = true;
flowers[j].ind++;
}
else
{
partial_order[j][i] = true;
flowers[i].ind++;
}
}
}
}
priority_queue<struct flower*, vector<struct flower*>, flower_comp> pq;
for(int i = 0; i < size; i++)
{
if(flowers[i].ind == 0)
{
pq.push(&flowers[i]);
}
}
printf("{");
bool first = true;
while(!pq.empty())
{
struct flower* tmp = pq.top();
pq.pop();
tmp->visited = true;
if(!first)
{
printf(",");
}
first = false;
printf("%d", tmp->height);
for(int j = 0; j < size; j++)
{
if(!flowers[j].visited && partial_order[tmp->id][j])
{
flowers[j].ind--;
if(flowers[j].ind == 0)
{
pq.push(&flowers[j]);
}
}
}
}
printf("}\n");
}
int main(int argc, char** argv)
{
int height[] = {5,4,3,2,1};
int bloom[] = {1,1,1,1,1};
int wilt[] = {365,365,365,365,365};
getOrdering(height, bloom, wilt, sizeof(height)/sizeof(height[0]));
int height0[] = {5,4,3,2,1};
int bloom0[] = {1,5,10,15,20};
int wilt0[] = {4,9,14,19,24};
getOrdering(height0, bloom0, wilt0, sizeof(height0)/sizeof(height0[0]));
int height1[] = {5,4,3,2,1};
int bloom1[] = {1,5,10,15,20};
int wilt1[] = {5,10,15,20,25};
getOrdering(height1, bloom1, wilt1, sizeof(height1)/sizeof(height1[0]));
int height2[] = {5,4,3,2,1};
int bloom2[] = {1,5,10,15,20};
int wilt2[] = {5,10,14,20,25};
getOrdering(height2, bloom2, wilt2, sizeof(height2)/sizeof(height2[0]));
int height3[] = {1,2,3,4,5,6};
int bloom3[] = {1,3,1,3,1,3};
int wilt3[] = {2,4,2,4,2,4};
getOrdering(height3, bloom3, wilt3, sizeof(height3)/sizeof(height3[0]));
int height4[] = {3,2,5,4};
int bloom4[] = {1,2,11,10};
int wilt4[] = {4,3,12,13};
getOrdering(height4, bloom4, wilt4, sizeof(height4)/sizeof(height4[0]));
}
Same thing as Rob's but in Javascript (ES6):
function getOrdering(height, bloom, wilt) {
var n = height.length;
var idx = [];
for (var i = 0; i < n; ++i) idx[i] = i;
idx.sort( (a, b) => height[a] - height[b] );
var intersect = (a, b) => !(bloom[a] > wilt[b] || bloom[b] > wilt[a]);
for (var i = 1; i < n; ++i) {
// assume they are ordered correctly till index (i-1),
// start moving flower i to the left until it can't move because of intersection
var j = i, flw = idx[i];
while (j > 0 && !intersect(idx[j-1], flw)) {
idx[j] = idx[j-1];
idx[--j] = flw;
}
}
return idx.map( x => height[x] );
}
Similar to Rob, again in Python and slightly convoluted overlapping bloom/wilt check.
H = 0
B = 1
W = 2
def getOrdering(heights, blooms, wilts):
def _f1_after_f2(f1, f2):
fs1 = set(range(f1[B], f1[W]+1))
fs2 = set(range(f2[B], f2[W]+1))
return f1[H] > f2[H] if fs2.intersection(fs1) != set([]) else False
fs = zip(heights, blooms, wilts)
fs.sort()
ffs = []
for f1 in fs:
insert_at = len(ffs)
for f2 in reversed(ffs):
if _f1_after_f2(f1, f2): break
insert_at -= 1
ffs.insert(insert_at, f1)
return [f[H] for f in ffs]
A graph algorithm to solve the problem:
Create a directed graph(V,E):
V -> flower types
E -> relations between 2 flower types
For all pairs (v_i, v_j)
If v_i is smaller than v_j and v_j 'blocks' v_i
draw an edge starting from v_i to v_j
For all nodes v_i
Find the v_i with no incoming edges and the biggest height
-> write it at the end of the result list
-> remove v_i and all of its outgoing edges from graph
For more description checkout this forum:
Topcoder Forum - FlowerGarden
Mine is like insertion sort. For each new flower, it goes from back to front and checks to see if the one in front of it blocks it; if it does, it means it must be placed behind it. Likewise, it also searches from front to back and checks to see if the one behind it blocks it; if it does, it means it must be placed in front of it. If there are no blocks, it simply checks for the best spot height-wise.
#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include <stdbool.h>
#define uint32 uint32_t
static void
Swap(int *AIdx, int *BIdx)
{
int Tmp = *AIdx;
*AIdx = *BIdx;
*BIdx = Tmp;
}
static void
SwapTo(int Start, int End, int *Array)
{
while(Start != End)
{
Swap(&Array[Start], &Array[Start - 1]);
--Start;
}
}
static void
PrintTo(int End, int *Array)
{
for(int Idx = 0;
Idx < End;
++Idx)
{
printf("%d, ", Array[Idx]);
}
printf("\n");
}
/* Does A block B? */
static bool
Blocks(int AIdx, int BIdx, int *Heights, int *Blooms, int *Wilts)
{
bool Result = (Heights[AIdx] > Heights[BIdx] &&
Wilts[AIdx] >= Blooms[BIdx] &&
Blooms[AIdx] <= Wilts[BIdx]);
return Result;
}
static void
Order(int *Heights, int *Blooms, int *Wilts,
int FlowerCount)
{
for(int FlowerIdx = 1;
FlowerIdx < FlowerCount;
++FlowerIdx)
{
PrintTo(FlowerIdx, Heights);
/* front to back */
int MinIdx = -1;
for(int Idx = 0;
Idx < FlowerIdx;
++Idx)
{
if(Blocks(Idx, FlowerIdx, Heights, Blooms, Wilts))
{
MinIdx = Idx;
break;
}
}
/* back to front */
int MaxIdx = -1;
for(int Idx = (FlowerIdx - 1);
Idx >= 0;
--Idx)
{
if(Blocks(FlowerIdx, Idx, Heights, Blooms, Wilts))
{
MaxIdx = (Idx + 1);
break;
}
}
/* best height index */
int BestHeightIdx = -1;
if(MinIdx == -1 &&
MaxIdx == -1)
{
for(int Idx = 0;
Idx < FlowerIdx;
++Idx)
{
if(Heights[FlowerIdx] > Heights[Idx])
{
BestHeightIdx = Idx;
break;
}
}
if(BestHeightIdx == -1)
{
BestHeightIdx = FlowerIdx;
}
}
int SwapToIdx = -1;
if((MaxIdx == -1 && MinIdx != -1) ||
(MinIdx == -1 && MaxIdx != -1) ||
(MaxIdx != -1 && MinIdx != -1 && MaxIdx == MinIdx))
{
SwapToIdx = (MinIdx != -1) ? MinIdx : MaxIdx;
}
else if(BestHeightIdx != -1)
{
SwapToIdx = BestHeightIdx;
}
else
{
fprintf(stderr, "Spot-finding error:\n MinIdx: %d, MaxIdx: %d, BestHIdx: %d\n",
MinIdx, MaxIdx, BestHeightIdx);
exit(1);
}
SwapTo(FlowerIdx, SwapToIdx, Heights);
SwapTo(FlowerIdx, SwapToIdx, Blooms);
SwapTo(FlowerIdx, SwapToIdx, Wilts);
}
}
int
main(int argc, char *argv[])
{
int Heights0[] = {5,4,3,2,1};
int Blooms0[] = {1,1,1,1,1};
int Wilts0[] = {365,365,365,365,365};
int Heights1[] = {5,4,3,2,1};
int Blooms1[] = {1,5,10,15,20};
int Wilts1[] = {4,9,14,19,24};
int Heights2[] = {5,4,3,2,1};
int Blooms2[] = {1,5,10,15,20};
int Wilts2[] = {5,10,15,20,25};
int Heights3[] = {5,4,3,2,1};
int Blooms3[] = {1,5,10,15,20};
int Wilts3[] = {5,10,14,20,25};
int Heights4[] = {1,2,3,4,5,6};
int Blooms4[] = {1,3,1,3,1,3};
int Wilts4[] = {2,4,2,4,2,4};
int Heights5[] = {3,2,5,4};
int Blooms5[] = {1,2,11,10};
int Wilts5[] = {4,3,12,13};
int *AllHeights[] = {Heights0, Heights1, Heights2, Heights3, Heights4, Heights5};
int *AllBlooms[] = {Blooms0, Blooms1, Blooms2, Blooms3, Blooms4, Blooms5};
int *AllWilts[] = {Wilts0, Wilts1, Wilts2, Wilts3, Wilts4, Wilts5};
int AllFlowerCounts[] = {5, 5, 5, 5, 6, 4};
printf("\n");
for(int Idx = 0;
Idx < 6;
++Idx)
{
int *Heights = AllHeights[Idx];
int *Blooms = AllBlooms[Idx];
int *Wilts = AllWilts[Idx];
int FlowerCount = AllFlowerCounts[Idx];
printf("Test %d\n", Idx);
Order(Heights, Blooms, Wilts, FlowerCount);
printf("{ ");
for(int Idx = 0;
Idx < FlowerCount;
++Idx)
{
printf("%d", Heights[Idx]);
if(Idx != (FlowerCount - 1))
{
printf(", ");
}
}
printf(" }\n\n");
}
}
EDIT: This solution is god awful and I came up with a better one that's actually DP; it's as follows: for each flower, loop through all other flowers checking which ones it blocks; for those flowers it blocks, check for all the flowers it blocks, and so on until you get to a flower that doesn't block any other ones. Put that flower in a new array. Backtrack and put each flower before it in the next slot of that new array. If done for each flower, you will get an array full of flowers that don't block any others. You then put each flower as far forward as possible. The DP part of this solution is that sometimes you'll come across the same flower that has already been blocked by another flower previously and has already been put in the new array, so we skip that flower instead of chasing down the flowers it blocks.
I have got the implementation in c++. I have used a vector datatype to store the height, bloom and wilt respectively and then i sorted it w.r.t to height after which i took the flowers one by one and arranged them according to the values associated with them.
here is the code :-
#include<iostream>
#include<vector>
#include<utility>
#include<algorithm>
using namespace std;
bool comp(pair<int, pair<int,int> >& a,pair<int, pair<int,int> >& b ){
return (a.first > b.first);
}
bool justify(pair<int, pair<int,int> >& a,pair<int, pair<int,int> >& b, int k , int
j, vector<pair<int,pair<int,int> > >& v){
if(((b.second.first <= a.second.first) && (b.second.second>= a.second.first)) ||
((b.second.first <= a.second.second) && (b.second.second>= a.second.second)) ||
((b.second.first > a.second.first) && (b.second.second < a.second.second) )){
pair<int, pair<int,int> > temp = v[j];
int i = j-1;
while(i >= k){
v[i+1] = v[i];
i--;
}
v[k] = temp;
return true;
}
return false;
}
int main() {
vector<pair<int,pair<int,int> > > v;
int n,a,b,c;
cin>>n;
for(int i = 0;i < n;i++){
cin>>a>>b>>c;
v.push_back(make_pair(a,make_pair(b,c)));
}
sort(v.begin(), v.end(), comp);
for(int j = 1;j < n;j++){
for(int k = 0;k < j;k++){
bool res = justify(v[k],v[j], k, j, v);
if(res)
break;
}
}
cout<<"output"<<endl;
for(int i = 0;i < n;i++){
cout<<v[i].first<<" "<<v[i].second.first<<" "<<v[i].second.second<<endl;
}
return 0;
}

Find a pair of elements from an array whose sum equals a given number

Given array of n integers and given a number X, find all the unique pairs of elements (a,b), whose summation is equal to X.
The following is my solution, it is O(nLog(n)+n), but I am not sure whether or not it is optimal.
int main(void)
{
int arr [10] = {1,2,3,4,5,6,7,8,9,0};
findpair(arr, 10, 7);
}
void findpair(int arr[], int len, int sum)
{
std::sort(arr, arr+len);
int i = 0;
int j = len -1;
while( i < j){
while((arr[i] + arr[j]) <= sum && i < j)
{
if((arr[i] + arr[j]) == sum)
cout << "(" << arr[i] << "," << arr[j] << ")" << endl;
i++;
}
j--;
while((arr[i] + arr[j]) >= sum && i < j)
{
if((arr[i] + arr[j]) == sum)
cout << "(" << arr[i] << "," << arr[j] << ")" << endl;
j--;
}
}
}
There are 3 approaches to this solution:
Let the sum be T and n be the size of array
Approach 1:
The naive way to do this would be to check all combinations (n choose 2). This exhaustive search is O(n2).
Approach 2:
A better way would be to sort the array. This takes O(n log n)
Then for each x in array A,
use binary search to look for T-x. This will take O(nlogn).
So, overall search is O(n log n)
Approach 3 :
The best way
would be to insert every element into a hash table (without sorting). This takes O(n) as constant time insertion.
Then for every x,
we can just look up its complement, T-x, which is O(1).
Overall the run time of this approach is O(n).
You can refer more here.Thanks.
# Let arr be the given array.
# And K be the give sum
for i=0 to arr.length - 1 do
# key is the element and value is its index.
hash(arr[i]) = i
end-for
for i=0 to arr.length - 1 do
# if K-th element exists and it's different then we found a pair
if hash(K - arr[i]) != i
print "pair i , hash(K - arr[i]) has sum K"
end-if
end-for
Implementation in Java : Using codaddict's algorithm (Maybe slightly different)
import java.util.HashMap;
public class ArrayPairSum {
public static void main(String[] args) {
int []a = {2,45,7,3,5,1,8,9};
printSumPairs(a,10);
}
public static void printSumPairs(int []input, int k){
Map<Integer, Integer> pairs = new HashMap<Integer, Integer>();
for(int i=0;i<input.length;i++){
if(pairs.containsKey(input[i]))
System.out.println(input[i] +", "+ pairs.get(input[i]));
else
pairs.put(k-input[i], input[i]);
}
}
}
For input = {2,45,7,3,5,1,8,9} and if Sum is 10
Output pairs:
3,7
8,2
9,1
Some notes about the solution :
We iterate only once through the array --> O(n) time
Insertion and lookup time in Hash is O(1).
Overall time is O(n), although it uses extra space in terms of hash.
Solution in java. You can add all the String elements to an ArrayList of strings and return the list. Here I am just printing it out.
void numberPairsForSum(int[] array, int sum) {
HashSet<Integer> set = new HashSet<Integer>();
for (int num : array) {
if (set.contains(sum - num)) {
String s = num + ", " + (sum - num) + " add up to " + sum;
System.out.println(s);
}
set.add(num);
}
}
Python Implementation:
import itertools
list = [1, 1, 2, 3, 4, 5,]
uniquelist = set(list)
targetsum = 5
for n in itertools.combinations(uniquelist, 2):
if n[0] + n[1] == targetsum:
print str(n[0]) + " + " + str(n[1])
Output:
1 + 4
2 + 3
C++11, run time complexity O(n):
#include <vector>
#include <unordered_map>
#include <utility>
std::vector<std::pair<int, int>> FindPairsForSum(
const std::vector<int>& data, const int& sum)
{
std::unordered_map<int, size_t> umap;
std::vector<std::pair<int, int>> result;
for (size_t i = 0; i < data.size(); ++i)
{
if (0 < umap.count(sum - data[i]))
{
size_t j = umap[sum - data[i]];
result.push_back({data[i], data[j]});
}
else
{
umap[data[i]] = i;
}
}
return result;
}
Here is a solution witch takes into account duplicate entries. It is written in javascript and assumes array is sorted. The solution runs in O(n) time and does not use any extra memory aside from variable.
var count_pairs = function(_arr,x) {
if(!x) x = 0;
var pairs = 0;
var i = 0;
var k = _arr.length-1;
if((k+1)<2) return pairs;
var halfX = x/2;
while(i<k) {
var curK = _arr[k];
var curI = _arr[i];
var pairsThisLoop = 0;
if(curK+curI==x) {
// if midpoint and equal find combinations
if(curK==curI) {
var comb = 1;
while(--k>=i) pairs+=(comb++);
break;
}
// count pair and k duplicates
pairsThisLoop++;
while(_arr[--k]==curK) pairsThisLoop++;
// add k side pairs to running total for every i side pair found
pairs+=pairsThisLoop;
while(_arr[++i]==curI) pairs+=pairsThisLoop;
} else {
// if we are at a mid point
if(curK==curI) break;
var distK = Math.abs(halfX-curK);
var distI = Math.abs(halfX-curI);
if(distI > distK) while(_arr[++i]==curI);
else while(_arr[--k]==curK);
}
}
return pairs;
}
I solved this during an interview for a large corporation. They took it but not me.
So here it is for everyone.
Start at both side of the array and slowly work your way inwards making sure to count duplicates if they exist.
It only counts pairs but can be reworked to
find the pairs
find pairs < x
find pairs > x
Enjoy!
O(n)
def find_pairs(L,sum):
s = set(L)
edgeCase = sum/2
if L.count(edgeCase) ==2:
print edgeCase, edgeCase
s.remove(edgeCase)
for i in s:
diff = sum-i
if diff in s:
print i, diff
L = [2,45,7,3,5,1,8,9]
sum = 10
find_pairs(L,sum)
Methodology: a + b = c, so instead of looking for (a,b) we look for a = c -
b
Implementation in Java : Using codaddict's algorithm:
import java.util.Hashtable;
public class Range {
public static void main(String[] args) {
// TODO Auto-generated method stub
Hashtable mapping = new Hashtable();
int a[]= {80,79,82,81,84,83,85};
int k = 160;
for (int i=0; i < a.length; i++){
mapping.put(a[i], i);
}
for (int i=0; i < a.length; i++){
if (mapping.containsKey(k - a[i]) && (Integer)mapping.get(k-a[i]) != i){
System.out.println(k-a[i]+", "+ a[i]);
}
}
}
}
Output:
81, 79
79, 81
If you want duplicate pairs (eg: 80,80) also then just remove && (Integer)mapping.get(k-a[i]) != i from the if condition and you are good to go.
Just attended this question on HackerRank and here's my 'Objective C' Solution:
-(NSNumber*)sum:(NSArray*) a andK:(NSNumber*)k {
NSMutableDictionary *dict = [NSMutableDictionary dictionary];
long long count = 0;
for(long i=0;i<a.count;i++){
if(dict[a[i]]) {
count++;
NSLog(#"a[i]: %#, dict[array[i]]: %#", a[i], dict[a[i]]);
}
else{
NSNumber *calcNum = #(k.longLongValue-((NSNumber*)a[i]).longLongValue);
dict[calcNum] = a[i];
}
}
return #(count);
}
Hope it helps someone.
this is the implementation of O(n*lg n) using binary search implementation inside a loop.
#include <iostream>
using namespace std;
bool *inMemory;
int pairSum(int arr[], int n, int k)
{
int count = 0;
if(n==0)
return count;
for (int i = 0; i < n; ++i)
{
int start = 0;
int end = n-1;
while(start <= end)
{
int mid = start + (end-start)/2;
if(i == mid)
break;
else if((arr[i] + arr[mid]) == k && !inMemory[i] && !inMemory[mid])
{
count++;
inMemory[i] = true;
inMemory[mid] = true;
}
else if(arr[i] + arr[mid] >= k)
{
end = mid-1;
}
else
start = mid+1;
}
}
return count;
}
int main()
{
int arr[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
inMemory = new bool[10];
for (int i = 0; i < 10; ++i)
{
inMemory[i] = false;
}
cout << pairSum(arr, 10, 11) << endl;
return 0;
}
In python
arr = [1, 2, 4, 6, 10]
diff_hash = {}
expected_sum = 3
for i in arr:
if diff_hash.has_key(i):
print i, diff_hash[i]
key = expected_sum - i
diff_hash[key] = i
Nice solution from Codeaddict. I took the liberty of implementing a version of it in Ruby:
def find_sum(arr,sum)
result ={}
h = Hash[arr.map {|i| [i,i]}]
arr.each { |l| result[l] = sum-l if h[sum-l] && !result[sum-l] }
result
end
To allow duplicate pairs (1,5), (5,1) we just have to remove the && !result[sum-l] instruction
Here is Java code for three approaches:
1. Using Map O(n), HashSet can also be used here.
2. Sort array and then use BinarySearch to look for complement O(nLog(n))
3. Traditional BruteForce two loops O(n^2)
public class PairsEqualToSum {
public static void main(String[] args) {
int a[] = {1,10,5,8,2,12,6,4};
findPairs1(a,10);
findPairs2(a,10);
findPairs3(a,10);
}
//Method1 - O(N) use a Map to insert values as keys & check for number's complement in map
static void findPairs1(int[]a, int sum){
Map<Integer, Integer> pairs = new HashMap<Integer, Integer>();
for(int i=0; i<a.length; i++){
if(pairs.containsKey(sum-a[i]))
System.out.println("("+a[i]+","+(sum-a[i])+")");
else
pairs.put(a[i], 0);
}
}
//Method2 - O(nlog(n)) using Sort
static void findPairs2(int[]a, int sum){
Arrays.sort(a);
for(int i=0; i<a.length/2; i++){
int complement = sum - a[i];
int foundAtIndex = Arrays.binarySearch(a,complement);
if(foundAtIndex >0 && foundAtIndex != i) //to avoid situation where binarySearch would find the original and not the complement like "5"
System.out.println("("+a[i]+","+(sum-a[i])+")");
}
}
//Method 3 - Brute Force O(n^2)
static void findPairs3(int[]a, int sum){
for(int i=0; i<a.length; i++){
for(int j=i; j<a.length;j++){
if(a[i]+a[j] == sum)
System.out.println("("+a[i]+","+a[j]+")");
}
}
}
}
A Simple program in java for arrays having unique elements:
import java.util.*;
public class ArrayPairSum {
public static void main(String[] args) {
int []a = {2,4,7,3,5,1,8,9,5};
sumPairs(a,10);
}
public static void sumPairs(int []input, int k){
Set<Integer> set = new HashSet<Integer>();
for(int i=0;i<input.length;i++){
if(set.contains(input[i]))
System.out.println(input[i] +", "+(k-input[i]));
else
set.add(k-input[i]);
}
}
}
A simple Java code snippet for printing the pairs below:
public static void count_all_pairs_with_given_sum(int arr[], int S){
if(arr.length < 2){
return;
}
HashSet values = new HashSet(arr.length);
for(int value : arr)values.add(value);
for(int value : arr){
int difference = S - value;
if(values.contains(difference) && value<difference){
System.out.printf("(%d, %d) %n", value, difference);
}
}
}
Another solution in Swift: the idea is to create an hash that store values of (sum - currentValue) and compare this to the current value of the loop. The complexity is O(n).
func findPair(list: [Int], _ sum: Int) -> [(Int, Int)]? {
var hash = Set<Int>() //save list of value of sum - item.
var dictCount = [Int: Int]() //to avoid the case A*2 = sum where we have only one A in the array
var foundKeys = Set<Int>() //to avoid duplicated pair in the result.
var result = [(Int, Int)]() //this is for the result.
for item in list {
//keep track of count of each element to avoid problem: [2, 3, 5], 10 -> result = (5,5)
if (!dictCount.keys.contains(item)) {
dictCount[item] = 1
} else {
dictCount[item] = dictCount[item]! + 1
}
//if my hash does not contain the (sum - item) value -> insert to hash.
if !hash.contains(sum-item) {
hash.insert(sum-item)
}
//check if current item is the same as another hash value or not, if yes, return the tuple.
if hash.contains(item) &&
(dictCount[item] > 1 || sum != item*2) // check if we have item*2 = sum or not.
{
if !foundKeys.contains(item) && !foundKeys.contains(sum-item) {
foundKeys.insert(item) //add to found items in order to not to add duplicated pair.
result.append((item, sum-item))
}
}
}
return result
}
//test:
let a = findPair([2,3,5,4,1,7,6,8,9,5,3,3,3,3,3,3,3,3,3], 14) //will return (8,6) and (9,5)
My Solution - Java - Without duplicates
public static void printAllPairSum(int[] a, int x){
System.out.printf("printAllPairSum(%s,%d)\n", Arrays.toString(a),x);
if(a==null||a.length==0){
return;
}
int length = a.length;
Map<Integer,Integer> reverseMapOfArray = new HashMap<>(length,1.0f);
for (int i = 0; i < length; i++) {
reverseMapOfArray.put(a[i], i);
}
for (int i = 0; i < length; i++) {
Integer j = reverseMapOfArray.get(x - a[i]);
if(j!=null && i<j){
System.out.printf("a[%d] + a[%d] = %d + %d = %d\n",i,j,a[i],a[j],x);
}
}
System.out.println("------------------------------");
}
This prints the pairs and avoids duplicates using bitwise manipulation.
public static void findSumHashMap(int[] arr, int key) {
Map<Integer, Integer> valMap = new HashMap<Integer, Integer>();
for(int i=0;i<arr.length;i++)
valMap.put(arr[i], i);
int indicesVisited = 0;
for(int i=0;i<arr.length;i++) {
if(valMap.containsKey(key - arr[i]) && valMap.get(key - arr[i]) != i) {
if(!((indicesVisited & ((1<<i) | (1<<valMap.get(key - arr[i])))) > 0)) {
int diff = key-arr[i];
System.out.println(arr[i] + " " +diff);
indicesVisited = indicesVisited | (1<<i) | (1<<valMap.get(key - arr[i]));
}
}
}
}
I bypassed the bit manuplation and just compared the index values. This is less than the loop iteration value (i in this case). This will not print the duplicate pairs and duplicate array elements also.
public static void findSumHashMap(int[] arr, int key) {
Map<Integer, Integer> valMap = new HashMap<Integer, Integer>();
for (int i = 0; i < arr.length; i++) {
valMap.put(arr[i], i);
}
for (int i = 0; i < arr.length; i++) {
if (valMap.containsKey(key - arr[i])
&& valMap.get(key - arr[i]) != i) {
if (valMap.get(key - arr[i]) < i) {
int diff = key - arr[i];
System.out.println(arr[i] + " " + diff);
}
}
}
}
in C#:
int[] array = new int[] { 1, 5, 7, 2, 9, 8, 4, 3, 6 }; // given array
int sum = 10; // given sum
for (int i = 0; i <= array.Count() - 1; i++)
if (array.Contains(sum - array[i]))
Console.WriteLine("{0}, {1}", array[i], sum - array[i]);
One Solution can be this, but not optimul (The complexity of this code is O(n^2)):
public class FindPairsEqualToSum {
private static int inputSum = 0;
public static List<String> findPairsForSum(int[] inputArray, int sum) {
List<String> list = new ArrayList<String>();
List<Integer> inputList = new ArrayList<Integer>();
for (int i : inputArray) {
inputList.add(i);
}
for (int i : inputArray) {
int tempInt = sum - i;
if (inputList.contains(tempInt)) {
String pair = String.valueOf(i + ", " + tempInt);
list.add(pair);
}
}
return list;
}
}
A simple python version of the code that find a pair sum of zero and can be modify to find k:
def sumToK(lst):
k = 0 # <- define the k here
d = {} # build a dictionary
# build the hashmap key = val of lst, value = i
for index, val in enumerate(lst):
d[val] = index
# find the key; if a key is in the dict, and not the same index as the current key
for i, val in enumerate(lst):
if (k-val) in d and d[k-val] != i:
return True
return False
The run time complexity of the function is O(n) and Space: O(n) as well.
public static int[] f (final int[] nums, int target) {
int[] r = new int[2];
r[0] = -1;
r[1] = -1;
int[] vIndex = new int[0Xfff];
for (int i = 0; i < nums.length; i++) {
int delta = 0Xff;
int gapIndex = target - nums[i] + delta;
if (vIndex[gapIndex] != 0) {
r[0] = vIndex[gapIndex];
r[1] = i + 1;
return r;
} else {
vIndex[nums[i] + delta] = i + 1;
}
}
return r;
}
less than o(n) solution will be=>
function(array,k)
var map = {};
for element in array
map(element) = true;
if(map(k-element))
return {k,element}
Solution in Python using list comprehension
f= [[i,j] for i in list for j in list if j+i==X];
O(N2)
also gives two ordered pairs- (a,b) and (b,a) as well
I can do it in O(n). Let me know when you want the answer. Note it involves simply traversing the array once with no sorting, etc... I should mention too that it exploits commutativity of addition and doesn't use hashes but wastes memory.
using System;
using System.Collections.Generic;
/*
An O(n) approach exists by using a lookup table. The approach is to store the value in a "bin" that can easily be looked up(e.g., O(1)) if it is a candidate for an appropriate sum.
e.g.,
for each a[k] in the array we simply put the it in another array at the location x - a[k].
Suppose we have [0, 1, 5, 3, 6, 9, 8, 7] and x = 9
We create a new array,
indexes value
9 - 0 = 9 0
9 - 1 = 8 1
9 - 5 = 4 5
9 - 3 = 6 3
9 - 6 = 3 6
9 - 9 = 0 9
9 - 8 = 1 8
9 - 7 = 2 7
THEN the only values that matter are the ones who have an index into the new table.
So, say when we reach 9 or equal we see if our new array has the index 9 - 9 = 0. Since it does we know that all the values it contains will add to 9. (note in this cause it's obvious there is only 1 possible one but it might have multiple index values in it which we need to store).
So effectively what we end up doing is only having to move through the array once. Because addition is commutative we will end up with all the possible results.
For example, when we get to 6 we get the index into our new table as 9 - 6 = 3. Since the table contains that index value we know the values.
This is essentially trading off speed for memory.
*/
namespace sum
{
class Program
{
static void Main(string[] args)
{
int num = 25;
int X = 10;
var arr = new List<int>();
for(int i = 0; i <= num; i++) arr.Add((new Random((int)(DateTime.Now.Ticks + i*num))).Next(0, num*2));
Console.Write("["); for (int i = 0; i < num - 1; i++) Console.Write(arr[i] + ", "); Console.WriteLine(arr[arr.Count-1] + "] - " + X);
var arrbrute = new List<Tuple<int,int>>();
var arrfast = new List<Tuple<int,int>>();
for(int i = 0; i < num; i++)
for(int j = i+1; j < num; j++)
if (arr[i] + arr[j] == X)
arrbrute.Add(new Tuple<int, int>(arr[i], arr[j]));
int M = 500;
var lookup = new List<List<int>>();
for(int i = 0; i < 1000; i++) lookup.Add(new List<int>());
for(int i = 0; i < num; i++)
{
// Check and see if we have any "matches"
if (lookup[M + X - arr[i]].Count != 0)
{
foreach(var j in lookup[M + X - arr[i]])
arrfast.Add(new Tuple<int, int>(arr[i], arr[j]));
}
lookup[M + arr[i]].Add(i);
}
for(int i = 0; i < arrbrute.Count; i++)
Console.WriteLine(arrbrute[i].Item1 + " + " + arrbrute[i].Item2 + " = " + X);
Console.WriteLine("---------");
for(int i = 0; i < arrfast.Count; i++)
Console.WriteLine(arrfast[i].Item1 + " + " + arrfast[i].Item2 + " = " + X);
Console.ReadKey();
}
}
}
I implemented logic in Scala with out a Map. It gives duplicate pairs since the counter loops thru entire elements of the array. If duplicate pairs are needed, you can simply return the value pc
val arr = Array[Int](8, 7, 2, 5, 3, 1, 5)
val num = 10
var pc = 0
for(i <- arr.indices) {
if(arr.contains(Math.abs(arr(i) - num))) pc += 1
}
println(s"Pairs: ${pc/2}")
It is working with duplicates values in the array as well.
GOLANG Implementation
func findPairs(slice1 []int, sum int) [][]int {
pairMap := make(map[int]int)
var SliceOfPairs [][]int
for i, v := range slice1 {
if valuei, ok := pairMap[v]; ok {
//fmt.Println("Pair Found", i, valuei)
SliceOfPairs = append(SliceOfPairs, []int{i, valuei})
} else {
pairMap[sum-v] = i
}
}
return SliceOfPairs
}
function findPairOfNumbers(arr, targetSum) {
arr = arr.sort();
var low = 0, high = arr.length - 1, sum, result = [];
while(low < high) {
sum = arr[low] + arr[high];
if(sum < targetSum)
low++;
else if(sum > targetSum)
high--;
else if(sum === targetSum) {
result.push({val1: arr[low], val2: arr[high]});
high--;
}
}
return (result || false);
}
var pairs = findPairOfNumbers([1,2,3,4,5,6,7,8,9,0], 7);
if(pairs.length) {
console.log(pairs);
} else {
console.log("No pair of numbers found that sums to " + 7);
}

Resources