Tournament bracket placement algorithm - algorithm

Given a list of opponent seeds (for example seeds 1 to 16), I'm trying to write an algorithm that will result in the top seed playing the lowest seed in that round, the 2nd seed playing the 2nd-lowest seed, etc.
Grouping 1 and 16, 2 and 15, etc. into "matches" is fairly easy, but I also need to make sure that the higher seed will play the lower seed in subsequent rounds.
An example bracket with the correct placement:
1 vs 16
1 vs 8
8 vs 9
1 vs 4
4 vs 13
4 vs 5
5 vs 12
1 vs 2
2 vs 15
2 vs 7
7 vs 10
2 vs 3
3 vs 14
3 vs 6
6 vs 11
As you can see, seed 1 and 2 only meet up in the final.

This JavaScript returns an array where each even index plays the next odd index
function seeding(numPlayers){
var rounds = Math.log(numPlayers)/Math.log(2)-1;
var pls = [1,2];
for(var i=0;i<rounds;i++){
pls = nextLayer(pls);
}
return pls;
function nextLayer(pls){
var out=[];
var length = pls.length*2+1;
pls.forEach(function(d){
out.push(d);
out.push(length-d);
});
return out;
}
}
> seeding(2)
[1, 2]
> seeding(4)
[1, 4, 2, 3]
> seeding(8)
[1, 8, 4, 5, 2, 7, 3, 6]
> seeding(16)
[1, 16, 8, 9, 4, 13, 5, 12, 2, 15, 7, 10, 3, 14, 6, 11]

With your assumptions, players 1 and 2 will play in the final, players 1-4 in the semifinals, players 1-8 in the quarterfinals and so on, so you can build the tournament recursively backwards from the final as AakashM proposed. Think of the tournament as a tree whose root is the final.
In the root node, your players are {1, 2}.
To expand the tree recursively to the next level, take all the nodes on the bottom layer in the tree, one by one, and create two children for them each, and place one of the players of the original node to each one of the child nodes created. Then add the next layer of players and map them to the game so that the worst newly added player plays against the best pre-existing player and so on.
Here first rounds of the algorithm:
{1,2} --- create next layer
{1, _}
/ --- now fill the empty slots
{1,2}
\{2, _}
{1, 4} --- the slots filled in reverse order
/
{1,2}
\{2, 3} --- create next layer again
/{1, _}
{1, 4}
/ \{4, _}
{1,2} --- again fill
\ /{2, _}
{2, 3}
\{3, _}
/{1, 8}
{1, 4}
/ \{4, 5} --- ... and so on
{1,2}
\ /{2, 7}
{2, 3}
\{3, 6}
As you can see, it produces the same tree you posted.

I've come up with the following algorithm. It may not be super-efficient, but I don't think that it really needs to be. It's written in PHP.
<?php
$players = range(1, 32);
$count = count($players);
$numberOfRounds = log($count / 2, 2);
// Order players.
for ($i = 0; $i < $numberOfRounds; $i++) {
$out = array();
$splice = pow(2, $i);
while (count($players) > 0) {
$out = array_merge($out, array_splice($players, 0, $splice));
$out = array_merge($out, array_splice($players, -$splice));
}
$players = $out;
}
// Print match list.
for ($i = 0; $i < $count; $i++) {
printf('%s vs %s<br />%s', $players[$i], $players[++$i], PHP_EOL);
}
?>

I also wrote a solution written in PHP. I saw Patrik Bodin's answer, but thought there must be an easier way.
It does what darkangel asked for: It returns all seeds in the correct positions. The matches are the same as in his example, but in a prettier order, seed 1 and seed number 16 are on the outside of the schema (as you see in tennis tournaments).
If there are no upsets (meaning a higher seeded player always wins from a lower seeded player), you will end up with seed 1 vs seed 2 in the final.
It actually does two things more:
It shows the correct order (which is a requirement for putting byes in the correct positions)
It fills in byes in the correct positions (if required)
A perfect explanation about what a single elimination bracket should look like: http://blog.playdriven.com/2011/articles/the-not-so-simple-single-elimination-advantage-seeding/
Code example for 16 participants:
<?php
define('NUMBER_OF_PARTICIPANTS', 16);
$participants = range(1,NUMBER_OF_PARTICIPANTS);
$bracket = getBracket($participants);
var_dump($bracket);
function getBracket($participants)
{
$participantsCount = count($participants);
$rounds = ceil(log($participantsCount)/log(2));
$bracketSize = pow(2, $rounds);
$requiredByes = $bracketSize - $participantsCount;
echo sprintf('Number of participants: %d<br/>%s', $participantsCount, PHP_EOL);
echo sprintf('Number of rounds: %d<br/>%s', $rounds, PHP_EOL);
echo sprintf('Bracket size: %d<br/>%s', $bracketSize, PHP_EOL);
echo sprintf('Required number of byes: %d<br/>%s', $requiredByes, PHP_EOL);
if($participantsCount < 2)
{
return array();
}
$matches = array(array(1,2));
for($round=1; $round < $rounds; $round++)
{
$roundMatches = array();
$sum = pow(2, $round + 1) + 1;
foreach($matches as $match)
{
$home = changeIntoBye($match[0], $participantsCount);
$away = changeIntoBye($sum - $match[0], $participantsCount);
$roundMatches[] = array($home, $away);
$home = changeIntoBye($sum - $match[1], $participantsCount);
$away = changeIntoBye($match[1], $participantsCount);
$roundMatches[] = array($home, $away);
}
$matches = $roundMatches;
}
return $matches;
}
function changeIntoBye($seed, $participantsCount)
{
//return $seed <= $participantsCount ? $seed : sprintf('%d (= bye)', $seed);
return $seed <= $participantsCount ? $seed : null;
}
?>
The output:
Number of participants: 16
Number of rounds: 4
Bracket size: 16
Required number of byes: 0
C:\projects\draw\draw.php:7:
array (size=8)
0 =>
array (size=2)
0 => int 1
1 => int 16
1 =>
array (size=2)
0 => int 9
1 => int 8
2 =>
array (size=2)
0 => int 5
1 => int 12
3 =>
array (size=2)
0 => int 13
1 => int 4
4 =>
array (size=2)
0 => int 3
1 => int 14
5 =>
array (size=2)
0 => int 11
1 => int 6
6 =>
array (size=2)
0 => int 7
1 => int 10
7 =>
array (size=2)
0 => int 15
1 => int 2
If you change 16 into 6 you get:
Number of participants: 6
Number of rounds: 3
Bracket size: 8
Required number of byes: 2
C:\projects\draw\draw.php:7:
array (size=4)
0 =>
array (size=2)
0 => int 1
1 => null
1 =>
array (size=2)
0 => int 5
1 => int 4
2 =>
array (size=2)
0 => int 3
1 => int 6
3 =>
array (size=2)
0 => null
1 => int 2

# Here's one in python - it uses nested list comprehension to be succinct:
from math import log, ceil
def seed( n ):
""" returns list of n in standard tournament seed order
Note that n need not be a power of 2 - 'byes' are returned as zero
"""
ol = [1]
for i in range( ceil( log(n) / log(2) ) ):
l = 2*len(ol) + 1
ol = [e if e <= n else 0 for s in [[el, l-el] for el in ol] for e in s]
return ol

For JavaScript code, use one of the two functions below. The former embodies imperative style & is much faster. The latter is recursive & neater, but only applicable to relatively small number of teams (<16384).
// imperative style
function foo(n) {
const arr = new Array(n)
arr[0] = 0
for (let i = n >> 1, m = 1; i >= 1; i >>= 1, m = (m << 1) + 1) {
for (let j = n - i; j > 0; j -= i) {
arr[j] = m - arr[j -= i]
}
}
return arr
}
Here you fill in the spots one by one by mirroring already occupied ones. For example, the first-seeded team (that is number 0) goes to the topmost spot. The second one (1) occupies the opposite spot in the other half of the bracket. The third team (2) mirrors 1 in their half of the bracket & so on. Despite the nested loops, the algorithm has a linear time complexity depending on the number of teams.
Here is the recursive method:
// functional style
const foo = n =>
n === 1 ? [0] : foo(n >> 1).reduce((p, c) => [...p, c, n - c - 1], [])
Basically, you do the same mirroring as in the previous function, but recursively:
For n = 1 team, it's just [0].
For n = 2 teams, you apply this function to the argument n-1 (that is,
1) & get [0]. Then you double the array by inserting mirrored
elements between them at even positions. Thus, [0] becomes [0, 1].
For n = 4 teams, you do the same operation, so [0, 1] becomes [0, 3,
1, 2].
If you want to get human-readable output, increase each element of the resulting array by one:
const readableArr = arr.map(i => i + 1)

At each round sort teams by seeding criteria
(If there are n teams in a round)team at ith position plays with team n-i+1

Since this comes up when searching on the subject, and it's hopeless to find another answer that solves the problem AND puts the seeds in a "prettier" order, I will add my version of the PHP code from darkangel. I also added the possibility to give byes to the higher seed players.
This was coded in an OO environment, so the number of participants are in $this->finalists and the number of byes are in $this->byes. I have only tested the code without byes and with two byes.
public function getBracket() {
$players = range(1, $this->finalists);
for ($i = 0; $i < log($this->finalists / 2, 2); $i++) {
$out = array();
$reverse = false;
foreach ($players as $player) {
$splice = pow(2, $i);
if ($reverse) {
$out = array_merge($out, array_splice($players, -$splice));
$out = array_merge($out, array_splice($players, 0, $splice));
$reverse = false;
} else {
$out = array_merge($out, array_splice($players, 0, $splice));
$out = array_merge($out, array_splice($players, -$splice));
$reverse = true;
}
}
$players = $out;
}
if ($this->byes) {
for ($i = 0; $i < $this->byes; $i++ ) {
for ($j = (($this->finalists / pow(2, $i)) - 1); $j > 0; $j--) {
$newPlace = ($this->finalists / pow(2, $i)) - 1;
if ($players[$j] > ($this->finalists / (pow(2 ,($i + 1))))) {
$player = $players[$j];
unset($players[$j]);
array_splice($players, $newPlace, 0, $player);
}
}
}
for ($i = 0; $i < $this->finalists / (pow(2, $this->byes)); $i++ ) {
$swap[] = $players[$i];
}
for ($i = 0; $i < $this->finalists /(pow(2, $this->byes)); $i++ ) {
$players[$i] = $swap[count($swap) - 1 - $i];
}
return array_reverse($players);
}
return $players;
}

I worked on a PHP / Laravel plugin that generates brackets with / without preliminary round robin. Maybe it can be useful to you, I don't know what tech you are using. Here is the github.
https://github.com/xoco70/kendo-tournaments
Hope it helps!

A C version.
int * pctournamentSeedArray(int PlayerCnt)
{
int * Array;
int * PrevArray;
int i;
Array = meAlloc(sizeof(int) * PlayerCnt);
if (PlayerCnt == 2)
{
Array[0] = 0;
Array[1] = 1;
return Array;
}
PrevArray = pctournamentSeedArray(PlayerCnt / 2);
for (i = 0; i < PlayerCnt;i += 2)
{
Array[i] = PrevArray[i / 2];
Array[i + 1] = (PlayerCnt - 1) - Array[i] ;
}
meFree(PrevArray);
return Array;
}

Related

How to solve sequential split dynamically

There is a number of weight in array arr.
arr= [1,5,3,2,4], each of the value in arr contains weight.
n = 2, must have 2 blocks while split the weight and order cannot break for split
Combination 1:
block 0: [1] max: 1
block 1: [5,3,2,4] max: 5
----------------------------
sum of max from block 0 and 1 is 6
Combination 2:
block 0: [1,5] max: 5
block 1: [3,2,4] max: 4
----------------------------
sum of max from block 0 and 1 is 9
Combination 3:
block 0: [1,5,3] max: 5
block 1: [2,4] max: 4
----------------------------
sum of max from block 0 and 1 is 9
Combination 4:
block 0: [1,5,3, 2] max: 5
block 1: [4] max: 4
----------------------------
sum of max from block 0 and 1 is 9
So here answer is 6 from Combination 1
The hardest part of some problems is just stating them clearly. If you can do that, the code practically writes itself.
I think the problem statement is this: Find the minimum value of a function (f) applied at every index of an array (f(array, index)), where f is the sum of the max values of two subarrays formed by splitting the input array at the given index.
function f(array, index) {
let left = array.slice(0, index)
let right = array.slice(index)
return Math.max(...left) + Math.max(...right)
}
let array = [1, 5, 3, 2, 4]
let smallestMax = Infinity
for (let i=1; i<array.length; i++) {
let max = f(array, i)
smallestMax = max < smallestMax ? max : smallestMax
}
console.log(smallestMax)
#Danh has a straightforward O(N^2) solution, but linear time is also possible without too much more work, and if you have enough data, it'll make a huge difference. Dan did his in JS it looks like, so I'll try to do the same. A bit rusty on these style for loops, so they may be off by one, but quick test in the JS console gave me 6 as expected (after fixing a copy/paste error).
Idea is to pass left to right finding the max at each index (left side). Then right to left finding the max at each index (right side). Then we look at left side plus the right side to get the value. Basically dynamic programming of a sort.
let array = [1, 5, 3, 2, 4]
let maxLeft = {}
let maxRight = {}
let max,newMax;
for (let i=0; i<array.length; i++) {
if (i === 0) {
maxLeft[i] = array[i]
} else {
maxLeft[i] = array[i] < maxLeft[i-1] ? maxLeft[i-1] : array[i]
}
}
for (let i=array.length - 1; i >= 0; i--) {
if (i === array.length - 1) {
maxRight[i] = array[i]
} else {
maxRight[i] = array[i] < maxRight[i+1] ? maxRight[i+1] : array[i]
}
}
for (let i=0; i<array.length; i++) {
newMax = maxLeft[i] + maxRight[i + 1]
if (i === 0) {
max = newMax
} else {
maxLeft[i] = newMax < maxLeft[i-1] ? max : newMax
}
}
console.log(max)

Algorithm for the largest subarray of distinct values in linear time

I'm trying to come up with a fast algorithm for, given any array of length n, obtaining the largest subarray of distinct values.
For example, the largest subarray of distinct values of
[1, 4, 3, 2, 4, 2, 8, 1, 9]
would be
[4, 2, 8, 1, 9]
This is my current solution, I think it runs in O(n^2). This is because check_dups runs in linear time, and it is called every time j or i increments.
arr = [0,...,n]
i = 0
j = 1
i_best = i
j_best = j
while i < n-1 and j < n:
if check_dups(arr, i j): //determines if there's duplicates in the subarray i,j in linear time
i += 1
else:
if j - i > j_best - i_best:
i_best = i
j_best = j
j += 1
return subarray(arr, i_best, j_best)
Does anyone have a better solution, in linear time?
Please note this is pseudocode and I'm not looking for an answer that relies on specific existing functions of a defined language (such as arr.contains()).
Thanks!
Consider the problem of finding the largest distinct-valued subarray ending at a particular index j. Conceptually this is straightforward: starting at arr[j], you go backwards and include all elements until you find a duplicate.
Let's use this intuition to solve this problem for all j from 0 up to length(arr). We need to know, at any point in the iteration, how far back we can go before we find a duplicate. That is, we need to know the least i such that subarray(arr, i, j) contains distinct values. (I'm assuming subarray treats the indices as inclusive.)
If we knew i at some point in the iteration (say, when j = k), can we quickly update i when j = k+1? Indeed, if we knew when was the last occurrence of arr[k+1], then we can update i := max(i, lastOccurrence(arr[k+1]) + 1). We can compute lastOccurrence in O(1) time with a HashMap.
Pseudocode:
arr = ... (from input)
map = empty HashMap
i = 0
i_best = 0
j_best = 0
for j from 0 to length(arr) - 1 inclusive:
if map contains-key arr[j]:
i = max(i, map[arr[j]] + 1)
map[arr[j]] = j
if j - i > j_best - i_best:
i_best = i
j_best = j
return subarray(arr, i_best, j_best)
We can adapt pkpnd's algorithm to use an array rather than hash map for an O(n log n) solution or potentially O(n) if your data allows for an O(n) stable sort, but you'd need to implement a stable sorting function that also provides the original indexes of the elements.
1 4 3 2 4 2 8 1 9
0 1 2 3 4 5 6 7 8 (indexes)
Sorted:
1 1 2 2 3 4 4 8 9
0 7 3 5 2 1 4 6 8 (indexes)
--- --- ---
Now, instead of a hash map, build a new array by iterating over the sorted array and inserting the last occurrence of each element according to the duplicate index arrangements. The final array would look like:
1 4 3 2 4 2 8 1 9
-1 -1 -1 -1 1 3 -1 0 -1 (previous occurrence)
We're now ready to run pkpnd's algorithm with a slight modification:
arr = ... (from input)
map = previous occurrence array
i = 0
i_best = 0
j_best = 0
for j from 0 to length(arr) - 1 inclusive:
if map[j] >= 0:
i = max(i, map[j] + 1)
if j - i > j_best - i_best:
i_best = i
j_best = j
return subarray(arr, i_best, j_best)
JavaScript code:
function f(arr, map){
let i = 0
let i_best = 0
let j_best = 0
for (j=0; j<arr.length; j++){
if (map[j] >= 0)
i = Math.max(i, map[j] + 1)
if (j - i > j_best - i_best){
i_best = i
j_best = j
}
}
return [i_best, j_best]
}
let arr = [ 1, 4, 3, 2, 4, 2, 8, 1, 9]
let map = [-1,-1,-1,-1, 1, 3,-1, 0,-1]
console.log(f(arr, map))
arr = [ 1, 2, 2, 2, 2, 2, 1]
map = [-1,-1, 1, 2, 3, 4, 0]
console.log(f(arr, map))
We can use Hashtable(Dictionary in c#)
public int[] FindSubarrayWithDistinctEntities(int[] arr)
{
Dictionary<int, int> dic = new Dictionary<int, int>();
Result r = new Result(); //struct containing start and end index for subarray
int result = 0;
r.st = 1;
r.end = 1;
for (int i = 0; i < arr.Length; i++)
{
if (dic.ContainsKey(arr[i]))
{
int diff = i - (dic[arr[i]] + 1);
if(result<diff)
{
result = diff;
r.st = Math.Min(r.st, (dic[arr[i]] + 1));
r.end = i-1;
}
dic.Remove(arr[i]);
}
dic.Add(arr[i], i);
}
return arr.Skip(r.st).Take(r.end).ToArray();
}
Add every number to Hashset if it isn't already in it. Hashset's insert and search are both O(1). So final result will be O(n).

Median of two sorted arrays of different length

I am trying to understand the algorithm that solves this problem in O(log(n+m)) where n and m are the lengths of the arrays. I have taken the liberty to post the link to the explanation of this algorithm:
https://www.geeksforgeeks.org/median-of-two-sorted-arrays-of-different-sizes/
It's so hard for me to digest completely the idea behind this algorithm. I can see that the idea is to reduce the length of one of the arrays to either 1 or 2 and then apply the base cases. The base cases make sense, but I wonder if one can omit the base case for n = 2 and just work on n = 1. I also don't understand the remaining cases part. It looks so weird to me that we have to cut the array B[] from the start to idx. It's weird because idx can be equal to the length of B[], so we would be ignoring the whole array.
TL;DR:
The main idea is that you may delete N elements that are surely smaller than (or equal to) the median from your number set, as long as you delete the same amount that are surely greater or equal.
Let's explain it with an example:
A=[1 2 3 9 10], B=[3 4 5 6 7 8 9]
The middle elements marked:
A=[1 2 3 9 10], B=[3 4 5 6 7 8 9]
The overall median will be between 3 and 6, inclusive. So, if we delete two elements smaller than 3, and two elements greater than 6, we'll still have the same median. The smaller elements we delete from A, and the greater ones from B:
A=[3 9 10], B=[3 4 5 6 7]
Now we delete one element greater than 9 (from A) and one smaller than 5 (from B):
A=[3 9], B=[4 5 6 7]
We reached Case 4 (smaller array has 2 elements): the algorithm calls for the median of
B[M/2], B[M/2 – 1], max(A[0], B[M/2 – 2]), min(A[1], B[M/2 + 1])
being B[2], B[1], max(A[0], B[0]), min(A[1], B[3])
being 6, 5, max(3,4), min(9,7)
being [6 5 4 7]
The median of that array is 5.5. And that's the correct result.
def findmedian(A,B):
if len(A) > len(B):
return findmedian(B,A)# always ensuring that we do the binsearch on the shorter arr
x = len(A)
y = len(B)
start = 0# start and end of shorter arr
end = x
while (start <= end):
partition_x = (start + end)//2# the mid of smaller arr, partition_x is an index
partition_y = (x+y+1)//2 - partition_x# the suitable partition of larger arr to divide the arrs into equal halves
if partition_x == 0:# if there is nothing on the left
left_x = None
if partition_x == x:# if there is nothing on the right
right_x = sys.maxint# +inf
if partition_y == 0:
left_y = None# this is -inf similar to the case for smaller arr
if partition_y == y:
right_y = sys.maxint
if (left_x <= right_y) and (left_y <= right_x):# all the elems on left are smaller than all the elems on right is ensured by
#checking on the extremes only since arrs sorted. Also, the partition always makes equal halves, so found the right spot.
if (x+y) % 2 == 0:
return (max(left_x,left_y) + min(right_x,right_y))/2.0
else:
return max(left_x,left_y)# if the num of elems is odd
elif left_x > right_y:# if we have come more towards right of smaller arr, then move left on smaller arr
end = partition_x -1
else:# if we have come more to the left
start = partition_x + 1
class Solution(object):
def findMedianSortedArrays(self, nums1, nums2):
merged_array = (nums1 + nums2)
merged_array.sort()
l_m_a = len(merged_array)
count = int(l_m_a / 2)
if l_m_a % 2 == 1:
median = merged_array[count]
return median
else:
median_in_even = (merged_array[count] + merged_array[count - 1]) / 2
return median_in_even
class Solution:
def findMedianSortedArrays(self, nums1, nums2):
nums1.extend(nums2)
newArray = sorted(nums1)
if len(newArray)%2==0:
index = len(newArray)//2
median = (newArray[index] + newArray[index-1])/2
return float(median)
else:
index = len(newArray)//2
median = newArray[index]
return float(median)
if __name__ == '__main__':
obj = Solution()
print(obj.findMedianSortedArrays([1,3],[2]))
Median of two sorted Arrays | Same length | Different length
1st we need to merge both arrays in sorted order. And then we can find
the median. The Median will be the central element of the sorted array.
var findMedianSortedArrays = function(nums1, nums2) {
let array = [], leftIndex = 0, rightIndex = 0;
while (leftIndex < nums1.length && rightIndex < nums2.length) {
if (nums1[leftIndex] < nums2[rightIndex]) {
array.push(nums1[leftIndex]);
leftIndex++;
} else {
array.push(nums2[rightIndex]);
rightIndex++;
}
}
// add uninitiated remaining element from either array if any remains.
array = array.concat(nums1.slice(leftIndex)).concat(nums2.slice(rightIndex));
if (array.length % 2 == 0) {
return (array[(array.length / 2) - 1] + array[array.length / 2]) / 2;
} else {
return array[Math.floor(array.length / 2)]
}
};
findMedianSortedArrays([1 2 3 9 10], [3 4 5 6 7 8 9]);
/**
* #param {number[]} nums1
* #param {number[]} nums2
* #return {number}
*/
var findMedianSortedArrays = function (nums1, nums2) {
let newArray = [];
let median;
if (nums1.length > 0 && nums2.length > 0) {
newArray = [...nums1, ...nums2]
newArray.sort(function (a, b) {
return a - b;
})
} else if (nums1.length === 0) {
newArray = nums2
newArray.sort(function (a, b) {
return a - b;
})
} else if (nums2.length === 0) {
newArray = nums1
newArray.sort(function (a, b) {
return a - b;
})
}
if (newArray.length === 1) {
return newArray[0]
}
if (newArray.length === 3) {
return newArray[1]
}
if (newArray.length % 2 === 0) {
const findIndex = Math.floor(newArray.length / 2)
console.log("findIndexeven", findIndex)
const addValue = Math.max((newArray[findIndex - 1] +
newArray[findIndex]), 0)
median = addValue / 2
} else {
const findIndex = Math.floor(newArray.length / 2) + 1
console.log("findIndexodd", findIndex)
median = newArray[findIndex - 1]
}
console.log("medianValue",median)
return median
};
findMedianSortedArrays([1, 2], [3, 4])
For me, it's just several minutes of several lines of python codes and it passed the leetcode check with a runtime beating 62% of Python3 online submissions. My code is here:
class Solution:
def findMedianSortedArrays(self, nums1: List[int], nums2: List[int]) -> float:
n = len(nums1) + len(nums2)
nums1.extend(nums2) # merge two lists
nums1.sort() # sort it
if n % 2 == 0:
return (nums1[n//2-1] + nums1[n//2])/2 # return the median for even n
else:
return nums1[n//2] # return the median for odd n

Maximum number achievable by converting two adjacent x to one (x+1)

Given a sequence of N integers where 1 <= N <= 500 and the numbers are between 1 and 50. In a step any two adjacent equal numbers x x can be replaced with a single x + 1. What is the maximum number achievable by such steps.
For example if given 2 3 1 1 2 2 then the maximum possible is 4:
2 3 1 1 2 2 ---> 2 3 2 2 2 ---> 2 3 3 2 ---> 2 4 2.
It is evident that I should try to do better than the maximum number available in the sequence. But I can't figure out a good algorithm.
Each substring of the input can make at most one single number (invariant: the log base two of the sum of two to the power of each entry). For every x, we can find the set of substrings that can make x. For each x, this is (1) every occurrence of x (2) the union of two contiguous substrings that can make x - 1. The resulting algorithm is O(N^2)-time.
An algorithm could work like this:
Convert the input to an array where every element has a frequency attribute, collapsing repeated consecutive values in the input into one single node. For example, this input:
1 2 2 4 3 3 3 3
Would be represented like this:
{val: 1, freq: 1} {val: 2, freq: 2} {val: 4, freq: 1} {val: 3, freq: 4}
Then find local minima nodes, like the node (3 3 3 3) in 1 (2 2) 4 (3 3 3 3) 4, i.e. nodes whose neighbours both have higher values. For those local minima that have an even frequency, "lift" those by applying the step. Repeat this until no such local minima (with even frequency) exist any more.
Start of the recursive part of the algorithm:
At both ends of the array, work inwards to "lift" values as long as the more inner neighbour has a higher value. With this rule, the following:
1 2 2 3 5 4 3 3 3 1 1
will completely resolve. First from the left side inward:
1 4 5 4 3 3 3 1 1
Then from the right side:
1 4 6 3 2
Note that when there is an odd frequency (like for the 3s above), there will be a "remainder" that cannot be incremented. The remainder should in this rule always be left on the outward side, so to maximise the potential towards the inner part of the array.
At this point the remaining local minima have odd frequencies. Applying the step to such a node will always leave a "remainder" (like above) with the original value. This remaining node can appear anywhere, but it only makes sense to look at solutions where this remainder is on the left side or the right side of the lift (not in the middle). So for example:
4 1 1 1 1 1 2 3 4
Can resolve to one of these:
4 2 2 1 2 3 4
Or:
4 1 2 2 2 3 4
The 1 in either second or fourth position, is the above mentioned "remainder". Obviously, the second way of resolving is more promising in this example. In general, the choice is obvious when on one side there is a value that is too high to merge with, like the left-most 4 is too high for five 1 values to get to. The 4 is like a wall.
When the frequency of the local minimum is one, there is nothing we can do with it. It actually separates the array in a left and right side that do not influence each other. The same is true for the remainder element discussed above: it separates the array into two parts that do not influence each other.
So the next step in the algorithm is to find such minima (where the choice is obvious), apply that kind of step and separate the problem into two distinct problems which should be solved recursively (from the top). So in the last example, the following two problems would be solved separately:
4
2 2 3 4
Then the best of both solutions will count as the overall solution. In this case that is 5.
The most challenging part of the algorithm is to deal with those local minima for which the choice of where to put the remainder is not obvious. For instance;
3 3 1 1 1 1 1 2 3
This can go to either:
3 3 2 2 1 2 3
3 3 1 2 2 2 3
In this example the end result is the same for both options, but in bigger arrays it would be less and less obvious. So here both options have to be investigated. In general you can have many of them, like 2 in this example:
3 1 1 1 2 3 1 1 1 1 1 3
Each of these two minima has two options. This seems like to explode into too many possibilities for larger arrays. But it is not that bad. The algorithm can take opposite choices in neighbouring minima, and go alternating like this through the whole array. This way alternating sections are favoured, and get the most possible value drawn into them, while the other sections are deprived of value. Now the algorithm turns the tables, and toggles all choices so that the sections that were previously favoured are now deprived, and vice versa. The solution of both these alternatives is derived by resolving each section recursively, and then comparing the two "grand" solutions to pick the best one.
Snippet
Here is a live JavaScript implementation of the above algorithm.
Comments are provided which hopefully should make it readable.
"use strict";
function Node(val, freq) {
// Immutable plain object
return Object.freeze({
val: val,
freq: freq || 1, // Default frequency is 1.
// Max attainable value when merged:
reduced: val + (freq || 1).toString(2).length - 1
});
}
function compress(a) {
// Put repeated elements in a single node
var result = [], i, j;
for (i = 0; i < a.length; i = j) {
for (j = i + 1; j < a.length && a[j] == a[i]; j++);
result.push(Node(a[i], j - i));
}
return result;
}
function decompress(a) {
// Expand nodes into separate, repeated elements
var result = [], i, j;
for (i = 0; i < a.length; i++) {
for (j = 0; j < a[i].freq; j++) {
result.push(a[i].val);
}
}
return result;
}
function str(a) {
return decompress(a).join(' ');
}
function unstr(s) {
s = s.replace(/\D+/g, ' ').trim();
return s.length ? compress(s.split(/\s+/).map(Number)) : [];
}
/*
The function merge modifies an array in-place, performing a "step" on
the indicated element.
The array will get an element with an incremented value
and decreased frequency, unless a join occurs with neighboring
elements with the same value: then the frequencies are accumulated
into one element. When the original frequency was odd there will
be a "remainder" element in the modified array as well.
*/
function merge(a, i, leftWards, stats) {
var val = a[i].val+1,
odd = a[i].freq % 2,
newFreq = a[i].freq >> 1,
last = i;
// Merge with neighbouring nodes of same value:
if ((!odd || !leftWards) && a[i+1] && a[i+1].val === val) {
newFreq += a[++last].freq;
}
if ((!odd || leftWards) && i && a[i-1].val === val) {
newFreq += a[--i].freq;
}
// Replace nodes
a.splice(i, last-i+1, Node(val, newFreq));
if (odd) a.splice(i+leftWards, 0, Node(val-1));
// Update statistics and trace: this is not essential to the algorithm
if (stats) {
stats.total_applied_merges++;
if (stats.trace) stats.trace.push(str(a));
}
return i;
}
/* Function Solve
Parameters:
a: The compressed array to be reduced via merges. It is changed in-place
and should not be relied on after the call.
stats: Optional plain object that will be populated with execution statistics.
Return value:
The array after the best merges were applied to achieve the highest
value, which is stored in the maxValue custom property of the array.
*/
function solve(a, stats) {
var maxValue, i, j, traceOrig, skipLeft, skipRight, sections, goLeft,
b, choice, alternate;
if (!a.length) return a;
if (stats && stats.trace) {
traceOrig = stats.trace;
traceOrig.push(stats.trace = [str(a)]);
}
// Look for valleys of even size, and "lift" them
for (i = 1; i < a.length - 1; i++) {
if (a[i-1].val > a[i].val && a[i].val < a[i+1].val && (a[i].freq % 2) < 1) {
// Found an even valley
i = merge(a, i, false, stats);
if (i) i--;
}
}
// Check left-side elements with always increasing values
for (i = 0; i < a.length-1 && a[i].val < a[i+1].val; i++) {
if (a[i].freq > 1) i = merge(a, i, false, stats) - 1;
};
// Check right-side elements with always increasing values, right-to-left
for (j = a.length-1; j > 0 && a[j-1].val > a[j].val; j--) {
if (a[j].freq > 1) j = merge(a, j, true, stats) + 1;
};
// All resolved?
if (i == j) {
while (a[i].freq > 1) merge(a, i, true, stats);
a.maxValue = a[i].val;
} else {
skipLeft = i;
skipRight = a.length - 1 - j;
// Look for other valleys (odd sized): they will lead to a split into sections
sections = [];
for (i = a.length - 2 - skipRight; i > skipLeft; i--) {
if (a[i-1].val > a[i].val && a[i].val < a[i+1].val) {
// Odd number of elements: if more than one, there
// are two ways to merge them, but maybe
// one of both possibilities can be excluded.
goLeft = a[i+1].val > a[i].reduced;
if (a[i-1].val > a[i].reduced || goLeft) {
if (a[i].freq > 1) i = merge(a, i, goLeft, stats) + goLeft;
// i is the index of the element which has become a 1-sized valley
// Split off the right part of the array, and store the solution
sections.push(solve(a.splice(i--), stats));
}
}
}
if (sections.length) {
// Solve last remaining section
sections.push(solve(a, stats));
sections.reverse();
// Combine the solutions of all sections into one
maxValue = sections[0].maxValue;
for (i = sections.length - 1; i >= 0; i--) {
maxValue = Math.max(sections[i].maxValue, maxValue);
}
} else {
// There is no more valley that can be resolved without branching into two
// directions. Look for the remaining valleys.
sections = [];
b = a.slice(0); // take copy
for (choice = 0; choice < 2; choice++) {
if (choice) a = b; // restore from copy on second iteration
alternate = choice;
for (i = a.length - 2 - skipRight; i > skipLeft; i--) {
if (a[i-1].val > a[i].val && a[i].val < a[i+1].val) {
// Odd number of elements
alternate = !alternate
i = merge(a, i, alternate, stats) + alternate;
sections.push(solve(a.splice(i--), stats));
}
}
// Solve last remaining section
sections.push(solve(a, stats));
}
sections.reverse(); // put in logical order
// Find best section:
maxValue = sections[0].maxValue;
for (i = sections.length - 1; i >= 0; i--) {
maxValue = Math.max(sections[i].maxValue, maxValue);
}
for (i = sections.length - 1; i >= 0 && sections[i].maxValue < maxValue; i--);
// Which choice led to the highest value (choice = 0 or 1)?
choice = (i >= sections.length / 2)
// Discard the not-chosen version
sections = sections.slice(choice * sections.length/2);
}
// Reconstruct the solution from the sections.
a = [].concat.apply([], sections);
a.maxValue = maxValue;
}
if (traceOrig) stats.trace = traceOrig;
return a;
}
function randomValues(len) {
var a = [];
for (var i = 0; i < len; i++) {
// 50% chance for a 1, 25% for a 2, ... etc.
a.push(Math.min(/\.1*/.exec(Math.random().toString(2))[0].length,5));
}
return a;
}
// I/O
var inputEl = document.querySelector('#inp');
var randEl = document.querySelector('#rand');
var lenEl = document.querySelector('#len');
var goEl = document.querySelector('#go');
var outEl = document.querySelector('#out');
goEl.onclick = function() {
// Get the input and structure it
var a = unstr(inputEl.value),
stats = {
total_applied_merges: 0,
trace: a.length < 100 ? [] : undefined
};
// Apply algorithm
a = solve(a, stats);
// Output results
var output = {
value: a.maxValue,
compact: str(a),
total_applied_merges: stats.total_applied_merges,
trace: stats.trace || 'no trace produced (input too large)'
};
outEl.textContent = JSON.stringify(output, null, 4);
}
randEl.onclick = function() {
// Get input (count of numbers to generate):
len = lenEl.value;
// Generate
var a = randomValues(len);
// Output
inputEl.value = a.join(' ');
// Simulate click to find the solution immediately.
goEl.click();
}
// Tests
var tests = [
' ', '',
'1', '1',
'1 1', '2',
'2 2 1 2 2', '3 1 3',
'3 2 1 1 2 2 3', '5',
'3 2 1 1 2 2 3 1 1 1 1 3 2 2 1 1 2', '6',
'3 1 1 1 3', '3 2 1 3',
'2 1 1 1 2 1 1 1 2 1 1 1 1 1 2', '3 1 2 1 4 1 2',
'3 1 1 2 1 1 1 2 3', '4 2 1 2 3',
'1 4 2 1 1 1 1 1 1 1', '1 5 1',
];
var res;
for (var i = 0; i < tests.length; i+=2) {
var res = str(solve(unstr(tests[i])));
if (res !== tests[i+1]) throw 'Test failed: ' + tests[i] + ' returned ' + res + ' instead of ' + tests[i+1];
}
Enter series (space separated):<br>
<input id="inp" size="60" value="2 3 1 1 2 2"><button id="go">Solve</button>
<br>
<input id="len" size="4" value="30"><button id="rand">Produce random series of this size and solve</button>
<pre id="out"></pre>
As you can see the program produces a reduced array with the maximum value included. In general there can be many derived arrays that have this maximum; only one is given.
An O(n*m) time and space algorithm is possible, where, according to your stated limits, n <= 500 and m <= 58 (consider that even for a billion elements, m need only be about 60, representing the largest element ± log2(n)). m is representing the possible numbers 50 + floor(log2(500)):
Consider the condensed sequence, s = {[x, number of x's]}.
If M[i][j] = [num_j,start_idx] where num_j represents the maximum number of contiguous js ending at index i of the condensed sequence; start_idx, the index where the sequence starts or -1 if it cannot join earlier sequences; then we have the following relationship:
M[i][j] = [s[i][1] + M[i-1][j][0], M[i-1][j][1]]
when j equals s[i][0]
j's greater than s[i][0] but smaller than or equal to s[i][0] + floor(log2(s[i][1])), represent converting pairs and merging with an earlier sequence if applicable, with a special case after the new count is odd:
When M[i][j][0] is odd, we do two things: first calculate the best so far by looking back in the matrix to a sequence that could merge with M[i][j] or its paired descendants, and then set a lower bound in the next applicable cells in the row (meaning a merge with an earlier sequence cannot happen via this cell). The reason this works is that:
if s[i + 1][0] > s[i][0], then s[i + 1] could only possibly pair with the new split section of s[i]; and
if s[i + 1][0] < s[i][0], then s[i + 1] might generate a lower j that would combine with the odd j from M[i], potentially making a longer sequence.
At the end, return the largest entry in the matrix, max(j + floor(log2(num_j))), for all j.
JavaScript code (counterexamples would be welcome; the limit on the answer is set at 7 for convenient visualization of the matrix):
function f(str){
var arr = str.split(/\s+/).map(Number);
var s = [,[arr[0],0]];
for (var i=0; i<arr.length; i++){
if (s[s.length - 1][0] == arr[i]){
s[s.length - 1][1]++;
} else {
s.push([arr[i],1]);
}
}
var M = [new Array(8).fill([0,0])],
best = 0;
for (var i=1; i<s.length; i++){
M[i] = new Array(8).fill([0,i]);
var temp = s[i][1],
temp_odd,
temp_start,
odd = false;
for (var j=s[i][0]; temp>0; j++){
var start_idx = odd ? temp_start : M[i][j-1][1];
if (start_idx != -1 && M[start_idx - 1][j][0]){
temp += M[start_idx - 1][j][0];
start_idx = M[start_idx - 1][j][1];
}
if (!odd){
M[i][j] = [temp,start_idx];
temp_odd = temp;
} else {
M[i][j] = [temp_odd,-1];
temp_start = start_idx;
}
if (!odd && temp & 1 && temp > 1){
odd = true;
temp_start = start_idx;
}
best = Math.max(best,j + Math.floor(Math.log2(temp)));
temp >>= 1;
temp_odd >>= 1;
}
}
return [arr, s, best, M];
}
// I/O
var button = document.querySelector('button');
var input = document.querySelector('input');
var pre = document.querySelector('pre');
button.onclick = function() {
var val = input.value;
var result = f(val);
var text = '';
for (var i=0; i<3; i++){
text += JSON.stringify(result[i]) + '\n\n';
}
for (var i in result[3]){
text += JSON.stringify(result[3][i]) + '\n';
}
pre.textContent = text;
}
<input value ="2 2 3 3 2 2 3 3 5">
<button>Solve</button>
<pre></pre>
Here's a brute force solution:
function findMax(array A, int currentMax)
for each pair (i, i+1) of indices for which A[i]==A[i+1] do
currentMax = max(A[i]+1, currentMax)
replace A[i],A[i+1] by a single number A[i]+1
currentMax = max(currentMax, findMax(A, currentMax))
end for
return currentMax
Given the array A, let currentMax=max(A[0], ..., A[n])
print findMax(A, currentMax)
The algorithm terminates because in each recursive call the array shrinks by 1.
It's also clear that it is correct: we try out all possible replacement sequences.
The code is extremely slow when the array is large and there's lots of options regarding replacements, but actually works reasonbly fast on arrays with small number of replaceable pairs. (I'll try to quantify the running time in terms of the number of replaceable pairs.)
A naive working code in Python:
def findMax(L, currMax):
for i in range(len(L)-1):
if L[i] == L[i+1]:
L[i] += 1
del L[i+1]
currMax = max(currMax, L[i])
currMax = max(currMax, findMax(L, currMax))
L[i] -= 1
L.insert(i+1, L[i])
return currMax
# entry point
if __name__ == '__main__':
L1 = [2, 3, 1, 1, 2, 2]
L2 = [2, 3, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2]
print findMax(L1, max(L1))
print findMax(L2, max(L2))
The result of the first call is 4, as expected.
The result of the second call is 5 as expected; the sequence that gives the result: 2,3,1,1,2,2,2,2,2,2,2,2, -> 2,3,1,1,3,2,2,2,2,2,2 -> 2,3,1,1,3,3,2,2,2,2, -> 2,3,1,1,3,3,3,2,2 -> 2,3,1,1,3,3,3,3 -> 2,3,1,1,4,3, -> 2,3,1,1,4,4 -> 2,3,1,1,5

Finding all possible combinations of numbers to reach a given sum

How would you go about testing all possible combinations of additions from a given set N of numbers so they add up to a given final number?
A brief example:
Set of numbers to add: N = {1,5,22,15,0,...}
Desired result: 12345
This problem can be solved with a recursive combinations of all possible sums filtering out those that reach the target. Here is the algorithm in Python:
def subset_sum(numbers, target, partial=[]):
s = sum(partial)
# check if the partial sum is equals to target
if s == target:
print "sum(%s)=%s" % (partial, target)
if s >= target:
return # if we reach the number why bother to continue
for i in range(len(numbers)):
n = numbers[i]
remaining = numbers[i+1:]
subset_sum(remaining, target, partial + [n])
if __name__ == "__main__":
subset_sum([3,9,8,4,5,7,10],15)
#Outputs:
#sum([3, 8, 4])=15
#sum([3, 5, 7])=15
#sum([8, 7])=15
#sum([5, 10])=15
This type of algorithms are very well explained in the following Stanford's Abstract Programming lecture - this video is very recommendable to understand how recursion works to generate permutations of solutions.
Edit
The above as a generator function, making it a bit more useful. Requires Python 3.3+ because of yield from.
def subset_sum(numbers, target, partial=[], partial_sum=0):
if partial_sum == target:
yield partial
if partial_sum >= target:
return
for i, n in enumerate(numbers):
remaining = numbers[i + 1:]
yield from subset_sum(remaining, target, partial + [n], partial_sum + n)
Here is the Java version of the same algorithm:
package tmp;
import java.util.ArrayList;
import java.util.Arrays;
class SumSet {
static void sum_up_recursive(ArrayList<Integer> numbers, int target, ArrayList<Integer> partial) {
int s = 0;
for (int x: partial) s += x;
if (s == target)
System.out.println("sum("+Arrays.toString(partial.toArray())+")="+target);
if (s >= target)
return;
for(int i=0;i<numbers.size();i++) {
ArrayList<Integer> remaining = new ArrayList<Integer>();
int n = numbers.get(i);
for (int j=i+1; j<numbers.size();j++) remaining.add(numbers.get(j));
ArrayList<Integer> partial_rec = new ArrayList<Integer>(partial);
partial_rec.add(n);
sum_up_recursive(remaining,target,partial_rec);
}
}
static void sum_up(ArrayList<Integer> numbers, int target) {
sum_up_recursive(numbers,target,new ArrayList<Integer>());
}
public static void main(String args[]) {
Integer[] numbers = {3,9,8,4,5,7,10};
int target = 15;
sum_up(new ArrayList<Integer>(Arrays.asList(numbers)),target);
}
}
It is exactly the same heuristic. My Java is a bit rusty but I think is easy to understand.
C# conversion of Java solution: (by #JeremyThompson)
public static void Main(string[] args)
{
List<int> numbers = new List<int>() { 3, 9, 8, 4, 5, 7, 10 };
int target = 15;
sum_up(numbers, target);
}
private static void sum_up(List<int> numbers, int target)
{
sum_up_recursive(numbers, target, new List<int>());
}
private static void sum_up_recursive(List<int> numbers, int target, List<int> partial)
{
int s = 0;
foreach (int x in partial) s += x;
if (s == target)
Console.WriteLine("sum(" + string.Join(",", partial.ToArray()) + ")=" + target);
if (s >= target)
return;
for (int i = 0; i < numbers.Count; i++)
{
List<int> remaining = new List<int>();
int n = numbers[i];
for (int j = i + 1; j < numbers.Count; j++) remaining.Add(numbers[j]);
List<int> partial_rec = new List<int>(partial);
partial_rec.Add(n);
sum_up_recursive(remaining, target, partial_rec);
}
}
Ruby solution: (by #emaillenin)
def subset_sum(numbers, target, partial=[])
s = partial.inject 0, :+
# check if the partial sum is equals to target
puts "sum(#{partial})=#{target}" if s == target
return if s >= target # if we reach the number why bother to continue
(0..(numbers.length - 1)).each do |i|
n = numbers[i]
remaining = numbers.drop(i+1)
subset_sum(remaining, target, partial + [n])
end
end
subset_sum([3,9,8,4,5,7,10],15)
Edit: complexity discussion
As others mention this is an NP-hard problem. It can be solved in exponential time O(2^n), for instance for n=10 there will be 1024 possible solutions. If the targets you are trying to reach are in a low range then this algorithm works. So for instance:
subset_sum([1,2,3,4,5,6,7,8,9,10],100000) generates 1024 branches because the target never gets to filter out possible solutions.
On the other hand subset_sum([1,2,3,4,5,6,7,8,9,10],10) generates only 175 branches, because the target to reach 10 gets to filter out many combinations.
If N and Target are big numbers one should move into an approximate version of the solution.
The solution of this problem has been given a million times on the Internet. The problem is called The coin changing problem. One can find solutions at http://rosettacode.org/wiki/Count_the_coins and mathematical model of it at http://jaqm.ro/issues/volume-5,issue-2/pdfs/patterson_harmel.pdf (or Google coin change problem).
By the way, the Scala solution by Tsagadai, is interesting. This example produces either 1 or 0. As a side effect, it lists on the console all possible solutions. It displays the solution, but fails making it usable in any way.
To be as useful as possible, the code should return a List[List[Int]]in order to allow getting the number of solution (length of the list of lists), the "best" solution (the shortest list), or all the possible solutions.
Here is an example. It is very inefficient, but it is easy to understand.
object Sum extends App {
def sumCombinations(total: Int, numbers: List[Int]): List[List[Int]] = {
def add(x: (Int, List[List[Int]]), y: (Int, List[List[Int]])): (Int, List[List[Int]]) = {
(x._1 + y._1, x._2 ::: y._2)
}
def sumCombinations(resultAcc: List[List[Int]], sumAcc: List[Int], total: Int, numbers: List[Int]): (Int, List[List[Int]]) = {
if (numbers.isEmpty || total < 0) {
(0, resultAcc)
} else if (total == 0) {
(1, sumAcc :: resultAcc)
} else {
add(sumCombinations(resultAcc, sumAcc, total, numbers.tail), sumCombinations(resultAcc, numbers.head :: sumAcc, total - numbers.head, numbers))
}
}
sumCombinations(Nil, Nil, total, numbers.sortWith(_ > _))._2
}
println(sumCombinations(15, List(1, 2, 5, 10)) mkString "\n")
}
When run, it displays:
List(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
List(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2)
List(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2)
List(1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2)
List(1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2)
List(1, 1, 1, 1, 1, 2, 2, 2, 2, 2)
List(1, 1, 1, 2, 2, 2, 2, 2, 2)
List(1, 2, 2, 2, 2, 2, 2, 2)
List(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 5)
List(1, 1, 1, 1, 1, 1, 1, 1, 2, 5)
List(1, 1, 1, 1, 1, 1, 2, 2, 5)
List(1, 1, 1, 1, 2, 2, 2, 5)
List(1, 1, 2, 2, 2, 2, 5)
List(2, 2, 2, 2, 2, 5)
List(1, 1, 1, 1, 1, 5, 5)
List(1, 1, 1, 2, 5, 5)
List(1, 2, 2, 5, 5)
List(5, 5, 5)
List(1, 1, 1, 1, 1, 10)
List(1, 1, 1, 2, 10)
List(1, 2, 2, 10)
List(5, 10)
The sumCombinations() function may be used by itself, and the result may be further analyzed to display the "best" solution (the shortest list), or the number of solutions (the number of lists).
Note that even like this, the requirements may not be fully satisfied. It might happen that the order of each list in the solution be significant. In such a case, each list would have to be duplicated as many time as there are combination of its elements. Or we might be interested only in the combinations that are different.
For example, we might consider that List(5, 10) should give two combinations: List(5, 10) and List(10, 5). For List(5, 5, 5) it could give three combinations or one only, depending on the requirements. For integers, the three permutations are equivalent, but if we are dealing with coins, like in the "coin changing problem", they are not.
Also not stated in the requirements is the question of whether each number (or coin) may be used only once or many times. We could (and we should!) generalize the problem to a list of lists of occurrences of each number. This translates in real life into "what are the possible ways to make an certain amount of money with a set of coins (and not a set of coin values)". The original problem is just a particular case of this one, where we have as many occurrences of each coin as needed to make the total amount with each single coin value.
A Javascript version:
function subsetSum(numbers, target, partial) {
var s, n, remaining;
partial = partial || [];
// sum partial
s = partial.reduce(function (a, b) {
return a + b;
}, 0);
// check if the partial sum is equals to target
if (s === target) {
console.log("%s=%s", partial.join("+"), target)
}
if (s >= target) {
return; // if we reach the number why bother to continue
}
for (var i = 0; i < numbers.length; i++) {
n = numbers[i];
remaining = numbers.slice(i + 1);
subsetSum(remaining, target, partial.concat([n]));
}
}
subsetSum([3,9,8,4,5,7,10],15);
// output:
// 3+8+4=15
// 3+5+7=15
// 8+7=15
// 5+10=15
In Haskell:
filter ((==) 12345 . sum) $ subsequences [1,5,22,15,0,..]
And J:
(]#~12345=+/#>)(]<##~[:#:#i.2^#)1 5 22 15 0 ...
As you may notice, both take the same approach and divide the problem into two parts: generate each member of the power set, and check each member's sum to the target.
There are other solutions but this is the most straightforward.
Do you need help with either one, or finding a different approach?
There are a lot of solutions so far, but all are of the form generate then filter. Which means that they potentially spend a lot of time working on recursive paths that do not lead to a solution.
Here is a solution that is O(size_of_array * (number_of_sums + number_of_solutions)). In other words it uses dynamic programming to avoid enumerating possible solutions that will never match.
For giggles and grins I made this work with numbers that are both positive and negative, and made it an iterator. It will work for Python 2.3+.
def subset_sum_iter(array, target):
sign = 1
array = sorted(array)
if target < 0:
array = reversed(array)
sign = -1
# Checkpoint A
last_index = {0: [-1]}
for i in range(len(array)):
for s in list(last_index.keys()):
new_s = s + array[i]
if 0 < (new_s - target) * sign:
pass # Cannot lead to target
elif new_s in last_index:
last_index[new_s].append(i)
else:
last_index[new_s] = [i]
# Checkpoint B
# Now yield up the answers.
def recur(new_target, max_i):
for i in last_index[new_target]:
if i == -1:
yield [] # Empty sum.
elif max_i <= i:
break # Not our solution.
else:
for answer in recur(new_target - array[i], i):
answer.append(array[i])
yield answer
for answer in recur(target, len(array)):
yield answer
And here is an example of it being used with an array and target where the filtering approach used in other solutions would effectively never finish.
def is_prime(n):
for i in range(2, n):
if 0 == n % i:
return False
elif n < i * i:
return True
if n == 2:
return True
else:
return False
def primes(limit):
n = 2
while True:
if is_prime(n):
yield(n)
n = n + 1
if limit < n:
break
for answer in subset_sum_iter(primes(1000), 76000):
print(answer)
This prints all 522 answers in under 2 seconds. The previous approaches would be lucky to find any answers in the current lifetime of the universe. (The full space has 2^168 = 3.74144419156711e+50 possible combinations to run through. That...takes a while.)
Explanation
I was asked to explain the code, but explaining data structures is usually more revealing. So I'll explain the data structures.
Let's consider subset_sum_iter([-2, 2, -3, 3, -5, 5, -7, 7, -11, 11], 10).
At checkpoint A, we have realized that our target is positive so sign = 1. And we've sorted our input so that array = [-11, -7, -5, -3, -2, 2, 3, 5, 7, 11]. Since we wind up accessing it by index a lot, here the the map from indexes to values:
0: -11
1: -7
2: -5
3: -3
4: -2
5: 2
6: 3
7: 5
8: 7
9: 11
By checkpoint B we have used Dynamic Programming to generate our last_index data structure. What does it contain?
last_index = {
-28: [4],
-26: [3, 5],
-25: [4, 6],
-24: [5],
-23: [2, 4, 5, 6, 7],
-22: [6],
-21: [3, 4, 5, 6, 7, 8],
-20: [4, 6, 7],
-19: [3, 5, 7, 8],
-18: [1, 4, 5, 6, 7, 8],
-17: [4, 5, 6, 7, 8, 9],
-16: [2, 4, 5, 6, 7, 8],
-15: [3, 5, 6, 7, 8, 9],
-14: [3, 4, 5, 6, 7, 8, 9],
-13: [4, 5, 6, 7, 8, 9],
-12: [2, 4, 5, 6, 7, 8, 9],
-11: [0, 5, 6, 7, 8, 9],
-10: [3, 4, 5, 6, 7, 8, 9],
-9: [4, 5, 6, 7, 8, 9],
-8: [3, 5, 6, 7, 8, 9],
-7: [1, 4, 5, 6, 7, 8, 9],
-6: [5, 6, 7, 8, 9],
-5: [2, 4, 5, 6, 7, 8, 9],
-4: [6, 7, 8, 9],
-3: [3, 5, 6, 7, 8, 9],
-2: [4, 6, 7, 8, 9],
-1: [5, 7, 8, 9],
0: [-1, 5, 6, 7, 8, 9],
1: [6, 7, 8, 9],
2: [5, 6, 7, 8, 9],
3: [6, 7, 8, 9],
4: [7, 8, 9],
5: [6, 7, 8, 9],
6: [7, 8, 9],
7: [7, 8, 9],
8: [7, 8, 9],
9: [8, 9],
10: [7, 8, 9]
}
(Side note, it is not symmetric because the condition if 0 < (new_s - target) * sign stops us from recording anything past target, which in our case was 10.)
What does this mean? Well, take the entry, 10: [7, 8, 9]. It means that we can wind up at a final sum of 10 with the last number chosen being at indexes 7, 8, or 9. Namely the last number chosen could be 5, 7, or 11.
Let's take a closer look at what happens if we choose index 7. That means we end on a 5. So therefore before we came to index 7, we had to get to 10-5 = 5. And the entry for 5 reads, 5: [6, 7, 8, 9]. So we could have picked index 6, which is 3. While we get to 5 at indexes 7, 8, and 9, we didn't get there before index 7. So our second to last choice has to be the 3 at index 6.
And now we have to get to 5-3 = 2 before index 6. The entry 2 reads: 2: [5, 6, 7, 8, 9]. Again, we only care about the answer at index 5 because the others happened too late. So the third to last choice is has to be the 2 at index 5.
And finally we have to get to 2-2 = 0 before index 5. The entry 0 reads: 0: [-1, 5, 6, 7, 8, 9]. Again we only care about the -1. But -1 isn't an index - in fact I'm using it to signal we're done choosing.
So we just found the solution 2+3+5 = 10. Which is the very first solution we print out.
And now we get to the recur subfunction. Because it is defined inside of our main function, it can see last_index.
The first thing to note is that it calls yield, not return. This makes it into a generator. When you call it you return a special kind of iterator. When you loop over that iterator, you'll get a list of all of the things it can yield. But you get them as it generates them. If it is a long list, you don't put it in memory. (Kind of important because we could get a long list.)
What recur(new_target, max_i) will yield are all of the ways that you could have summed up to new_target using only elements of array with maximum index max_i. That is it answers: "We have to get to new_target before index max_i+1." It is, of course, recursive.
Therefore recur(target, len(array)) is all solutions that reach target using any index at all. Which is what we want.
C++ version of the same algorithm
#include <iostream>
#include <list>
void subset_sum_recursive(std::list<int> numbers, int target, std::list<int> partial)
{
int s = 0;
for (std::list<int>::const_iterator cit = partial.begin(); cit != partial.end(); cit++)
{
s += *cit;
}
if(s == target)
{
std::cout << "sum([";
for (std::list<int>::const_iterator cit = partial.begin(); cit != partial.end(); cit++)
{
std::cout << *cit << ",";
}
std::cout << "])=" << target << std::endl;
}
if(s >= target)
return;
int n;
for (std::list<int>::const_iterator ai = numbers.begin(); ai != numbers.end(); ai++)
{
n = *ai;
std::list<int> remaining;
for(std::list<int>::const_iterator aj = ai; aj != numbers.end(); aj++)
{
if(aj == ai)continue;
remaining.push_back(*aj);
}
std::list<int> partial_rec=partial;
partial_rec.push_back(n);
subset_sum_recursive(remaining,target,partial_rec);
}
}
void subset_sum(std::list<int> numbers,int target)
{
subset_sum_recursive(numbers,target,std::list<int>());
}
int main()
{
std::list<int> a;
a.push_back (3); a.push_back (9); a.push_back (8);
a.push_back (4);
a.push_back (5);
a.push_back (7);
a.push_back (10);
int n = 15;
//std::cin >> n;
subset_sum(a, n);
return 0;
}
C# version of #msalvadores code answer
void Main()
{
int[] numbers = {3,9,8,4,5,7,10};
int target = 15;
sum_up(new List<int>(numbers.ToList()),target);
}
static void sum_up_recursive(List<int> numbers, int target, List<int> part)
{
int s = 0;
foreach (int x in part)
{
s += x;
}
if (s == target)
{
Console.WriteLine("sum(" + string.Join(",", part.Select(n => n.ToString()).ToArray()) + ")=" + target);
}
if (s >= target)
{
return;
}
for (int i = 0;i < numbers.Count;i++)
{
var remaining = new List<int>();
int n = numbers[i];
for (int j = i + 1; j < numbers.Count;j++)
{
remaining.Add(numbers[j]);
}
var part_rec = new List<int>(part);
part_rec.Add(n);
sum_up_recursive(remaining,target,part_rec);
}
}
static void sum_up(List<int> numbers, int target)
{
sum_up_recursive(numbers,target,new List<int>());
}
Java non-recursive version that simply keeps adding elements and redistributing them amongst possible values. 0's are ignored and works for fixed lists (what you're given is what you can play with) or a list of repeatable numbers.
import java.util.*;
public class TestCombinations {
public static void main(String[] args) {
ArrayList<Integer> numbers = new ArrayList<>(Arrays.asList(0, 1, 2, 2, 5, 10, 20));
LinkedHashSet<Integer> targets = new LinkedHashSet<Integer>() {{
add(4);
add(10);
add(25);
}};
System.out.println("## each element can appear as many times as needed");
for (Integer target: targets) {
Combinations combinations = new Combinations(numbers, target, true);
combinations.calculateCombinations();
for (String solution: combinations.getCombinations()) {
System.out.println(solution);
}
}
System.out.println("## each element can appear only once");
for (Integer target: targets) {
Combinations combinations = new Combinations(numbers, target, false);
combinations.calculateCombinations();
for (String solution: combinations.getCombinations()) {
System.out.println(solution);
}
}
}
public static class Combinations {
private boolean allowRepetitions;
private int[] repetitions;
private ArrayList<Integer> numbers;
private Integer target;
private Integer sum;
private boolean hasNext;
private Set<String> combinations;
/**
* Constructor.
*
* #param numbers Numbers that can be used to calculate the sum.
* #param target Target value for sum.
*/
public Combinations(ArrayList<Integer> numbers, Integer target) {
this(numbers, target, true);
}
/**
* Constructor.
*
* #param numbers Numbers that can be used to calculate the sum.
* #param target Target value for sum.
*/
public Combinations(ArrayList<Integer> numbers, Integer target, boolean allowRepetitions) {
this.allowRepetitions = allowRepetitions;
if (this.allowRepetitions) {
Set<Integer> numbersSet = new HashSet<>(numbers);
this.numbers = new ArrayList<>(numbersSet);
} else {
this.numbers = numbers;
}
this.numbers.removeAll(Arrays.asList(0));
Collections.sort(this.numbers);
this.target = target;
this.repetitions = new int[this.numbers.size()];
this.combinations = new LinkedHashSet<>();
this.sum = 0;
if (this.repetitions.length > 0)
this.hasNext = true;
else
this.hasNext = false;
}
/**
* Calculate and return the sum of the current combination.
*
* #return The sum.
*/
private Integer calculateSum() {
this.sum = 0;
for (int i = 0; i < repetitions.length; ++i) {
this.sum += repetitions[i] * numbers.get(i);
}
return this.sum;
}
/**
* Redistribute picks when only one of each number is allowed in the sum.
*/
private void redistribute() {
for (int i = 1; i < this.repetitions.length; ++i) {
if (this.repetitions[i - 1] > 1) {
this.repetitions[i - 1] = 0;
this.repetitions[i] += 1;
}
}
if (this.repetitions[this.repetitions.length - 1] > 1)
this.repetitions[this.repetitions.length - 1] = 0;
}
/**
* Get the sum of the next combination. When 0 is returned, there's no other combinations to check.
*
* #return The sum.
*/
private Integer next() {
if (this.hasNext && this.repetitions.length > 0) {
this.repetitions[0] += 1;
if (!this.allowRepetitions)
this.redistribute();
this.calculateSum();
for (int i = 0; i < this.repetitions.length && this.sum != 0; ++i) {
if (this.sum > this.target) {
this.repetitions[i] = 0;
if (i + 1 < this.repetitions.length) {
this.repetitions[i + 1] += 1;
if (!this.allowRepetitions)
this.redistribute();
}
this.calculateSum();
}
}
if (this.sum.compareTo(0) == 0)
this.hasNext = false;
}
return this.sum;
}
/**
* Calculate all combinations whose sum equals target.
*/
public void calculateCombinations() {
while (this.hasNext) {
if (this.next().compareTo(target) == 0)
this.combinations.add(this.toString());
}
}
/**
* Return all combinations whose sum equals target.
*
* #return Combinations as a set of strings.
*/
public Set<String> getCombinations() {
return this.combinations;
}
#Override
public String toString() {
StringBuilder stringBuilder = new StringBuilder("" + sum + ": ");
for (int i = 0; i < repetitions.length; ++i) {
for (int j = 0; j < repetitions[i]; ++j) {
stringBuilder.append(numbers.get(i) + " ");
}
}
return stringBuilder.toString();
}
}
}
Sample input:
numbers: 0, 1, 2, 2, 5, 10, 20
targets: 4, 10, 25
Sample output:
## each element can appear as many times as needed
4: 1 1 1 1
4: 1 1 2
4: 2 2
10: 1 1 1 1 1 1 1 1 1 1
10: 1 1 1 1 1 1 1 1 2
10: 1 1 1 1 1 1 2 2
10: 1 1 1 1 2 2 2
10: 1 1 2 2 2 2
10: 2 2 2 2 2
10: 1 1 1 1 1 5
10: 1 1 1 2 5
10: 1 2 2 5
10: 5 5
10: 10
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2
25: 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2
25: 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2
25: 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2
25: 1 1 1 2 2 2 2 2 2 2 2 2 2 2
25: 1 2 2 2 2 2 2 2 2 2 2 2 2
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 5
25: 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 5
25: 1 1 1 1 1 1 1 1 2 2 2 2 2 2 5
25: 1 1 1 1 1 1 2 2 2 2 2 2 2 5
25: 1 1 1 1 2 2 2 2 2 2 2 2 5
25: 1 1 2 2 2 2 2 2 2 2 2 5
25: 2 2 2 2 2 2 2 2 2 2 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 2 5 5
25: 1 1 1 1 1 1 1 1 1 1 1 2 2 5 5
25: 1 1 1 1 1 1 1 1 1 2 2 2 5 5
25: 1 1 1 1 1 1 1 2 2 2 2 5 5
25: 1 1 1 1 1 2 2 2 2 2 5 5
25: 1 1 1 2 2 2 2 2 2 5 5
25: 1 2 2 2 2 2 2 2 5 5
25: 1 1 1 1 1 1 1 1 1 1 5 5 5
25: 1 1 1 1 1 1 1 1 2 5 5 5
25: 1 1 1 1 1 1 2 2 5 5 5
25: 1 1 1 1 2 2 2 5 5 5
25: 1 1 2 2 2 2 5 5 5
25: 2 2 2 2 2 5 5 5
25: 1 1 1 1 1 5 5 5 5
25: 1 1 1 2 5 5 5 5
25: 1 2 2 5 5 5 5
25: 5 5 5 5 5
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 10
25: 1 1 1 1 1 1 1 1 1 1 1 1 1 2 10
25: 1 1 1 1 1 1 1 1 1 1 1 2 2 10
25: 1 1 1 1 1 1 1 1 1 2 2 2 10
25: 1 1 1 1 1 1 1 2 2 2 2 10
25: 1 1 1 1 1 2 2 2 2 2 10
25: 1 1 1 2 2 2 2 2 2 10
25: 1 2 2 2 2 2 2 2 10
25: 1 1 1 1 1 1 1 1 1 1 5 10
25: 1 1 1 1 1 1 1 1 2 5 10
25: 1 1 1 1 1 1 2 2 5 10
25: 1 1 1 1 2 2 2 5 10
25: 1 1 2 2 2 2 5 10
25: 2 2 2 2 2 5 10
25: 1 1 1 1 1 5 5 10
25: 1 1 1 2 5 5 10
25: 1 2 2 5 5 10
25: 5 5 5 10
25: 1 1 1 1 1 10 10
25: 1 1 1 2 10 10
25: 1 2 2 10 10
25: 5 10 10
25: 1 1 1 1 1 20
25: 1 1 1 2 20
25: 1 2 2 20
25: 5 20
## each element can appear only once
4: 2 2
10: 1 2 2 5
10: 10
25: 1 2 2 20
25: 5 20
Thank you.. ephemient
i have converted above logic from python to php..
<?php
$data = array(array(2,3,5,10,15),array(4,6,23,15,12),array(23,34,12,1,5));
$maxsum = 25;
print_r(bestsum($data,$maxsum)); //function call
function bestsum($data,$maxsum)
{
$res = array_fill(0, $maxsum + 1, '0');
$res[0] = array(); //base case
foreach($data as $group)
{
$new_res = $res; //copy res
foreach($group as $ele)
{
for($i=0;$i<($maxsum-$ele+1);$i++)
{
if($res[$i] != 0)
{
$ele_index = $i+$ele;
$new_res[$ele_index] = $res[$i];
$new_res[$ele_index][] = $ele;
}
}
}
$res = $new_res;
}
for($i=$maxsum;$i>0;$i--)
{
if($res[$i]!=0)
{
return $res[$i];
break;
}
}
return array();
}
?>
Another python solution would be to use the itertools.combinations module as follows:
#!/usr/local/bin/python
from itertools import combinations
def find_sum_in_list(numbers, target):
results = []
for x in range(len(numbers)):
results.extend(
[
combo for combo in combinations(numbers ,x)
if sum(combo) == target
]
)
print results
if __name__ == "__main__":
find_sum_in_list([3,9,8,4,5,7,10], 15)
Output: [(8, 7), (5, 10), (3, 8, 4), (3, 5, 7)]
I thought I'd use an answer from this question but I couldn't, so here is my answer. It is using a modified version of an answer in Structure and Interpretation of Computer Programs. I think this is a better recursive solution and should please the purists more.
My answer is in Scala (and apologies if my Scala sucks, I've just started learning it). The findSumCombinations craziness is to sort and unique the original list for the recursion to prevent dupes.
def findSumCombinations(target: Int, numbers: List[Int]): Int = {
cc(target, numbers.distinct.sortWith(_ < _), List())
}
def cc(target: Int, numbers: List[Int], solution: List[Int]): Int = {
if (target == 0) {println(solution); 1 }
else if (target < 0 || numbers.length == 0) 0
else
cc(target, numbers.tail, solution)
+ cc(target - numbers.head, numbers, numbers.head :: solution)
}
To use it:
> findSumCombinations(12345, List(1,5,22,15,0,..))
* Prints a whole heap of lists that will sum to the target *
Excel VBA version below. I needed to implement this in VBA (not my preference, don't judge me!), and used the answers on this page for the approach. I'm uploading in case others also need a VBA version.
Option Explicit
Public Sub SumTarget()
Dim numbers(0 To 6) As Long
Dim target As Long
target = 15
numbers(0) = 3: numbers(1) = 9: numbers(2) = 8: numbers(3) = 4: numbers(4) = 5
numbers(5) = 7: numbers(6) = 10
Call SumUpTarget(numbers, target)
End Sub
Public Sub SumUpTarget(numbers() As Long, target As Long)
Dim part() As Long
Call SumUpRecursive(numbers, target, part)
End Sub
Private Sub SumUpRecursive(numbers() As Long, target As Long, part() As Long)
Dim s As Long, i As Long, j As Long, num As Long
Dim remaining() As Long, partRec() As Long
s = SumArray(part)
If s = target Then Debug.Print "SUM ( " & ArrayToString(part) & " ) = " & target
If s >= target Then Exit Sub
If (Not Not numbers) <> 0 Then
For i = 0 To UBound(numbers)
Erase remaining()
num = numbers(i)
For j = i + 1 To UBound(numbers)
AddToArray remaining, numbers(j)
Next j
Erase partRec()
CopyArray partRec, part
AddToArray partRec, num
SumUpRecursive remaining, target, partRec
Next i
End If
End Sub
Private Function ArrayToString(x() As Long) As String
Dim n As Long, result As String
result = "{" & x(n)
For n = LBound(x) + 1 To UBound(x)
result = result & "," & x(n)
Next n
result = result & "}"
ArrayToString = result
End Function
Private Function SumArray(x() As Long) As Long
Dim n As Long
SumArray = 0
If (Not Not x) <> 0 Then
For n = LBound(x) To UBound(x)
SumArray = SumArray + x(n)
Next n
End If
End Function
Private Sub AddToArray(arr() As Long, x As Long)
If (Not Not arr) <> 0 Then
ReDim Preserve arr(0 To UBound(arr) + 1)
Else
ReDim Preserve arr(0 To 0)
End If
arr(UBound(arr)) = x
End Sub
Private Sub CopyArray(destination() As Long, source() As Long)
Dim n As Long
If (Not Not source) <> 0 Then
For n = 0 To UBound(source)
AddToArray destination, source(n)
Next n
End If
End Sub
Output (written to the Immediate window) should be:
SUM ( {3,8,4} ) = 15
SUM ( {3,5,7} ) = 15
SUM ( {8,7} ) = 15
SUM ( {5,10} ) = 15
Here's a solution in R
subset_sum = function(numbers,target,partial=0){
if(any(is.na(partial))) return()
s = sum(partial)
if(s == target) print(sprintf("sum(%s)=%s",paste(partial[-1],collapse="+"),target))
if(s > target) return()
for( i in seq_along(numbers)){
n = numbers[i]
remaining = numbers[(i+1):length(numbers)]
subset_sum(remaining,target,c(partial,n))
}
}
Perl version (of the leading answer):
use strict;
sub subset_sum {
my ($numbers, $target, $result, $sum) = #_;
print 'sum('.join(',', #$result).") = $target\n" if $sum == $target;
return if $sum >= $target;
subset_sum([#$numbers[$_ + 1 .. $#$numbers]], $target,
[#{$result||[]}, $numbers->[$_]], $sum + $numbers->[$_])
for (0 .. $#$numbers);
}
subset_sum([3,9,8,4,5,7,10,6], 15);
Result:
sum(3,8,4) = 15
sum(3,5,7) = 15
sum(9,6) = 15
sum(8,7) = 15
sum(4,5,6) = 15
sum(5,10) = 15
Javascript version:
const subsetSum = (numbers, target, partial = [], sum = 0) => {
if (sum < target)
numbers.forEach((num, i) =>
subsetSum(numbers.slice(i + 1), target, partial.concat([num]), sum + num));
else if (sum == target)
console.log('sum(%s) = %s', partial.join(), target);
}
subsetSum([3,9,8,4,5,7,10,6], 15);
Javascript one-liner that actually returns results (instead of printing it):
const subsetSum=(n,t,p=[],s=0,r=[])=>(s<t?n.forEach((l,i)=>subsetSum(n.slice(i+1),t,[...p,l],s+l,r)):s==t?r.push(p):0,r);
console.log(subsetSum([3,9,8,4,5,7,10,6], 15));
And my favorite, one-liner with callback:
const subsetSum=(n,t,cb,p=[],s=0)=>s<t?n.forEach((l,i)=>subsetSum(n.slice(i+1),t,cb,[...p,l],s+l)):s==t?cb(p):0;
subsetSum([3,9,8,4,5,7,10,6], 15, console.log);
Here is a Java version which is well suited for small N and very large target sum, when complexity O(t*N) (the dynamic solution) is greater than the exponential algorithm. My version uses a meet in the middle attack, along with a little bit shifting in order to reduce the complexity from the classic naive O(n*2^n) to O(2^(n/2)).
If you want to use this for sets with between 32 and 64 elements, you should change the int which represents the current subset in the step function to a long although performance will obviously drastically decrease as the set size increases. If you want to use this for a set with odd number of elements, you should add a 0 to the set to make it even numbered.
import java.util.ArrayList;
import java.util.List;
public class SubsetSumMiddleAttack {
static final int target = 100000000;
static final int[] set = new int[]{ ... };
static List<Subset> evens = new ArrayList<>();
static List<Subset> odds = new ArrayList<>();
static int[][] split(int[] superSet) {
int[][] ret = new int[2][superSet.length / 2];
for (int i = 0; i < superSet.length; i++) ret[i % 2][i / 2] = superSet[i];
return ret;
}
static void step(int[] superSet, List<Subset> accumulator, int subset, int sum, int counter) {
accumulator.add(new Subset(subset, sum));
if (counter != superSet.length) {
step(superSet, accumulator, subset + (1 << counter), sum + superSet[counter], counter + 1);
step(superSet, accumulator, subset, sum, counter + 1);
}
}
static void printSubset(Subset e, Subset o) {
String ret = "";
for (int i = 0; i < 32; i++) {
if (i % 2 == 0) {
if ((1 & (e.subset >> (i / 2))) == 1) ret += " + " + set[i];
}
else {
if ((1 & (o.subset >> (i / 2))) == 1) ret += " + " + set[i];
}
}
if (ret.startsWith(" ")) ret = ret.substring(3) + " = " + (e.sum + o.sum);
System.out.println(ret);
}
public static void main(String[] args) {
int[][] superSets = split(set);
step(superSets[0], evens, 0,0,0);
step(superSets[1], odds, 0,0,0);
for (Subset e : evens) {
for (Subset o : odds) {
if (e.sum + o.sum == target) printSubset(e, o);
}
}
}
}
class Subset {
int subset;
int sum;
Subset(int subset, int sum) {
this.subset = subset;
this.sum = sum;
}
}
Very efficient algorithm using tables i wrote in c++ couple a years ago.
If you set PRINT 1 it will print all combinations(but it wont be use the efficient method).
Its so efficient that it calculate more than 10^14 combinations in less than 10ms.
#include <stdio.h>
#include <stdlib.h>
//#include "CTime.h"
#define SUM 300
#define MAXNUMsSIZE 30
#define PRINT 0
long long CountAddToSum(int,int[],int,const int[],int);
void printr(const int[], int);
long long table1[SUM][MAXNUMsSIZE];
int main()
{
int Nums[]={3,4,5,6,7,9,13,11,12,13,22,35,17,14,18,23,33,54};
int sum=SUM;
int size=sizeof(Nums)/sizeof(int);
int i,j,a[]={0};
long long N=0;
//CTime timer1;
for(i=0;i<SUM;++i)
for(j=0;j<MAXNUMsSIZE;++j)
table1[i][j]=-1;
N = CountAddToSum(sum,Nums,size,a,0); //algorithm
//timer1.Get_Passd();
//printf("\nN=%lld time=%.1f ms\n", N,timer1.Get_Passd());
printf("\nN=%lld \n", N);
getchar();
return 1;
}
long long CountAddToSum(int s, int arr[],int arrsize, const int r[],int rsize)
{
static int totalmem=0, maxmem=0;
int i,*rnew;
long long result1=0,result2=0;
if(s<0) return 0;
if (table1[s][arrsize]>0 && PRINT==0) return table1[s][arrsize];
if(s==0)
{
if(PRINT) printr(r, rsize);
return 1;
}
if(arrsize==0) return 0;
//else
rnew=(int*)malloc((rsize+1)*sizeof(int));
for(i=0;i<rsize;++i) rnew[i]=r[i];
rnew[rsize]=arr[arrsize-1];
result1 = CountAddToSum(s,arr,arrsize-1,rnew,rsize);
result2 = CountAddToSum(s-arr[arrsize-1],arr,arrsize,rnew,rsize+1);
table1[s][arrsize]=result1+result2;
free(rnew);
return result1+result2;
}
void printr(const int r[], int rsize)
{
int lastr=r[0],count=0,i;
for(i=0; i<rsize;++i)
{
if(r[i]==lastr)
count++;
else
{
printf(" %d*%d ",count,lastr);
lastr=r[i];
count=1;
}
}
if(r[i-1]==lastr) printf(" %d*%d ",count,lastr);
printf("\n");
}
This is similar to a coin change problem
public class CoinCount
{
public static void main(String[] args)
{
int[] coins={1,4,6,2,3,5};
int count=0;
for (int i=0;i<coins.length;i++)
{
count=count+Count(9,coins,i,0);
}
System.out.println(count);
}
public static int Count(int Sum,int[] coins,int index,int curSum)
{
int count=0;
if (index>=coins.length)
return 0;
int sumNow=curSum+coins[index];
if (sumNow>Sum)
return 0;
if (sumNow==Sum)
return 1;
for (int i= index+1;i<coins.length;i++)
count+=Count(Sum,coins,i,sumNow);
return count;
}
}
I ported the C# sample to Objective-c and didn't see it in the responses:
//Usage
NSMutableArray* numberList = [[NSMutableArray alloc] init];
NSMutableArray* partial = [[NSMutableArray alloc] init];
int target = 16;
for( int i = 1; i<target; i++ )
{ [numberList addObject:#(i)]; }
[self findSums:numberList target:target part:partial];
//*******************************************************************
// Finds combinations of numbers that add up to target recursively
//*******************************************************************
-(void)findSums:(NSMutableArray*)numbers target:(int)target part:(NSMutableArray*)partial
{
int s = 0;
for (NSNumber* x in partial)
{ s += [x intValue]; }
if (s == target)
{ NSLog(#"Sum[%#]", partial); }
if (s >= target)
{ return; }
for (int i = 0;i < [numbers count];i++ )
{
int n = [numbers[i] intValue];
NSMutableArray* remaining = [[NSMutableArray alloc] init];
for (int j = i + 1; j < [numbers count];j++)
{ [remaining addObject:#([numbers[j] intValue])]; }
NSMutableArray* partRec = [[NSMutableArray alloc] initWithArray:partial];
[partRec addObject:#(n)];
[self findSums:remaining target:target part:partRec];
}
}
Here is a better version with better output formatting and C++ 11 features:
void subset_sum_rec(std::vector<int> & nums, const int & target, std::vector<int> & partialNums)
{
int currentSum = std::accumulate(partialNums.begin(), partialNums.end(), 0);
if (currentSum > target)
return;
if (currentSum == target)
{
std::cout << "sum([";
for (auto it = partialNums.begin(); it != std::prev(partialNums.end()); ++it)
cout << *it << ",";
cout << *std::prev(partialNums.end());
std::cout << "])=" << target << std::endl;
}
for (auto it = nums.begin(); it != nums.end(); ++it)
{
std::vector<int> remaining;
for (auto it2 = std::next(it); it2 != nums.end(); ++it2)
remaining.push_back(*it2);
std::vector<int> partial = partialNums;
partial.push_back(*it);
subset_sum_rec(remaining, target, partial);
}
}
Deduce 0 in the first place. Zero is an identiy for addition so it is useless by the monoid laws in this particular case. Also deduce negative numbers as well if you want to climb up to a positive number. Otherwise you would also need subtraction operation.
So... the fastest algorithm you can get on this particular job is as follows given in JS.
function items2T([n,...ns],t){
var c = ~~(t/n);
return ns.length ? Array(c+1).fill()
.reduce((r,_,i) => r.concat(items2T(ns, t-n*i).map(s => Array(i).fill(n).concat(s))),[])
: t % n ? []
: [Array(c).fill(n)];
};
var data = [3, 9, 8, 4, 5, 7, 10],
result;
console.time("combos");
result = items2T(data, 15);
console.timeEnd("combos");
console.log(JSON.stringify(result));
This is a very fast algorithm but if you sort the data array descending it will be even faster. Using .sort() is insignificant since the algorithm will end up with much less recursive invocations.
PHP Version, as inspired by Keith Beller's C# version.
bala's PHP version did not work for me, because I did not need to group numbers. I wanted a simpler implementation with one target value, and a pool of numbers. This function will also prune any duplicate entries.
Edit 25/10/2021: Added the precision argument to support floating point numbers (now requires the bcmath extension).
/**
* Calculates a subset sum: finds out which combinations of numbers
* from the numbers array can be added together to come to the target
* number.
*
* Returns an indexed array with arrays of number combinations.
*
* Example:
*
* <pre>
* $matches = subset_sum(array(5,10,7,3,20), 25);
* </pre>
*
* Returns:
*
* <pre>
* Array
* (
* [0] => Array
* (
* [0] => 3
* [1] => 5
* [2] => 7
* [3] => 10
* )
* [1] => Array
* (
* [0] => 5
* [1] => 20
* )
* )
* </pre>
*
* #param number[] $numbers
* #param number $target
* #param array $part
* #param int $precision
* #return array[number[]]
*/
function subset_sum($numbers, $target, $precision=0, $part=null)
{
// we assume that an empty $part variable means this
// is the top level call.
$toplevel = false;
if($part === null) {
$toplevel = true;
$part = array();
}
$s = 0;
foreach($part as $x)
{
$s = $s + $x;
}
// we have found a match!
if(bccomp((string) $s, (string) $target, $precision) === 0)
{
sort($part); // ensure the numbers are always sorted
return array(implode('|', $part));
}
// gone too far, break off
if($s >= $target)
{
return null;
}
$matches = array();
$totalNumbers = count($numbers);
for($i=0; $i < $totalNumbers; $i++)
{
$remaining = array();
$n = $numbers[$i];
for($j = $i+1; $j < $totalNumbers; $j++)
{
$remaining[] = $numbers[$j];
}
$part_rec = $part;
$part_rec[] = $n;
$result = subset_sum($remaining, $target, $precision, $part_rec);
if($result)
{
$matches = array_merge($matches, $result);
}
}
if(!$toplevel)
{
return $matches;
}
// this is the top level function call: we have to
// prepare the final result value by stripping any
// duplicate results.
$matches = array_unique($matches);
$result = array();
foreach($matches as $entry)
{
$result[] = explode('|', $entry);
}
return $result;
}
Example:
$result = subset_sum(array(5, 10, 7, 3, 20), 25);
This will return an indexed array with two number combination arrays:
3, 5, 7, 10
5, 20
Example with floating point numbers:
// Specify the precision in the third argument
$result = subset_sum(array(0.40, 0.03, 0.05), 0.45, 2);
This will return a single match:
0.40, 0.05
To find the combinations using excel - (its fairly easy).
(You computer must not be too slow)
Go to this site
Go to the "Sum to Target" page
Download the "Sum to Target" excel file.
Follow the directions on the website page.
hope this helps.
Swift 3 conversion of Java solution: (by #JeremyThompson)
protocol _IntType { }
extension Int: _IntType {}
extension Array where Element: _IntType {
func subsets(to: Int) -> [[Element]]? {
func sum_up_recursive(_ numbers: [Element], _ target: Int, _ partial: [Element], _ solution: inout [[Element]]) {
var sum: Int = 0
for x in partial {
sum += x as! Int
}
if sum == target {
solution.append(partial)
}
guard sum < target else {
return
}
for i in stride(from: 0, to: numbers.count, by: 1) {
var remaining = [Element]()
for j in stride(from: i + 1, to: numbers.count, by: 1) {
remaining.append(numbers[j])
}
var partial_rec = [Element](partial)
partial_rec.append(numbers[i])
sum_up_recursive(remaining, target, partial_rec, &solution)
}
}
var solutions = [[Element]]()
sum_up_recursive(self, to, [Element](), &solutions)
return solutions.count > 0 ? solutions : nil
}
}
usage:
let numbers = [3, 9, 8, 4, 5, 7, 10]
if let solution = numbers.subsets(to: 15) {
print(solution) // output: [[3, 8, 4], [3, 5, 7], [8, 7], [5, 10]]
} else {
print("not possible")
}
This can be used to print all the answers as well
public void recur(int[] a, int n, int sum, int[] ans, int ind) {
if (n < 0 && sum != 0)
return;
if (n < 0 && sum == 0) {
print(ans, ind);
return;
}
if (sum >= a[n]) {
ans[ind] = a[n];
recur(a, n - 1, sum - a[n], ans, ind + 1);
}
recur(a, n - 1, sum, ans, ind);
}
public void print(int[] a, int n) {
for (int i = 0; i < n; i++)
System.out.print(a[i] + " ");
System.out.println();
}
Time Complexity is exponential. Order of 2^n
I was doing something similar for a scala assignment. Thought of posting my solution here:
def countChange(money: Int, coins: List[Int]): Int = {
def getCount(money: Int, remainingCoins: List[Int]): Int = {
if(money == 0 ) 1
else if(money < 0 || remainingCoins.isEmpty) 0
else
getCount(money, remainingCoins.tail) +
getCount(money - remainingCoins.head, remainingCoins)
}
if(money == 0 || coins.isEmpty) 0
else getCount(money, coins)
}
#KeithBeller's answer with slightly changed variable names and some comments.
public static void Main(string[] args)
{
List<int> input = new List<int>() { 3, 9, 8, 4, 5, 7, 10 };
int targetSum = 15;
SumUp(input, targetSum);
}
public static void SumUp(List<int> input, int targetSum)
{
SumUpRecursive(input, targetSum, new List<int>());
}
private static void SumUpRecursive(List<int> remaining, int targetSum, List<int> listToSum)
{
// Sum up partial
int sum = 0;
foreach (int x in listToSum)
sum += x;
//Check sum matched
if (sum == targetSum)
Console.WriteLine("sum(" + string.Join(",", listToSum.ToArray()) + ")=" + targetSum);
//Check sum passed
if (sum >= targetSum)
return;
//Iterate each input character
for (int i = 0; i < remaining.Count; i++)
{
//Build list of remaining items to iterate
List<int> newRemaining = new List<int>();
for (int j = i + 1; j < remaining.Count; j++)
newRemaining.Add(remaining[j]);
//Update partial list
List<int> newListToSum = new List<int>(listToSum);
int currentItem = remaining[i];
newListToSum.Add(currentItem);
SumUpRecursive(newRemaining, targetSum, newListToSum);
}
}'
Recommended as an answer:
Here's a solution using es2015 generators:
function* subsetSum(numbers, target, partial = [], partialSum = 0) {
if(partialSum === target) yield partial
if(partialSum >= target) return
for(let i = 0; i < numbers.length; i++){
const remaining = numbers.slice(i + 1)
, n = numbers[i]
yield* subsetSum(remaining, target, [...partial, n], partialSum + n)
}
}
Using generators can actually be very useful because it allows you to pause script execution immediately upon finding a valid subset. This is in contrast to solutions without generators (ie lacking state) which have to iterate through every single subset of numbers
I did not like the Javascript Solution I saw above. Here is the one I build using partial applying, closures and recursion:
Ok, I was mainly concern about, if the combinations array could satisfy the target requirement, hopefully this approached you will start to find the rest of combinations
Here just set the target and pass the combinations array.
function main() {
const target = 10
const getPermutationThatSumT = setTarget(target)
const permutation = getPermutationThatSumT([1, 4, 2, 5, 6, 7])
console.log( permutation );
}
the currently implementation I came up with
function setTarget(target) {
let partial = [];
return function permute(input) {
let i, removed;
for (i = 0; i < input.length; i++) {
removed = input.splice(i, 1)[0];
partial.push(removed);
const sum = partial.reduce((a, b) => a + b)
if (sum === target) return partial.slice()
if (sum < target) permute(input)
input.splice(i, 0, removed);
partial.pop();
}
return null
};
}
An iterative C++ stack solution for a flavor of this problem. Unlike some other iterative solutions, it doesn't make unnecessary copies of intermediate sequences.
#include <vector>
#include <iostream>
// Given a positive integer, return all possible combinations of
// positive integers that sum up to it.
std::vector<std::vector<int>> print_all_sum(int target){
std::vector<std::vector<int>> output;
std::vector<int> stack;
int curr_min = 1;
int sum = 0;
while (curr_min < target) {
sum += curr_min;
if (sum >= target) {
if (sum == target) {
output.push_back(stack); // make a copy
output.back().push_back(curr_min);
}
sum -= curr_min + stack.back();
curr_min = stack.back() + 1;
stack.pop_back();
} else {
stack.push_back(curr_min);
}
}
return output;
}
int main()
{
auto vvi = print_all_sum(6);
for (auto const& v: vvi) {
for(auto const& i: v) {
std::cout << i;
}
std::cout << "\n";
}
return 0;
}
Output print_all_sum(6):
111111
11112
1113
1122
114
123
15
222
24
33
function solve(n){
let DP = [];
DP[0] = DP[1] = DP[2] = 1;
DP[3] = 2;
for (let i = 4; i <= n; i++) {
DP[i] = DP[i-1] + DP[i-3] + DP[i-4];
}
return DP[n]
}
console.log(solve(5))
This is a Dynamic Solution for JS to tell how many ways anyone can get the certain sum. This can be the right solution if you think about time and space complexity.

Resources