How to build an algorithm to find a combination, which summation is nearest to a number and its difference is within a range in c#

Question

I have a list of random values as below

319, 4, 90, 50, 20, 99, 500, 95, 900

and i have to find a values which sum is within selected range say 5% to 10%.

for example if the number is 300 and the range was 5% to 10%

then the difference should be in the range 15 to 30

then the list which satisfy this condition is

319 => 319-300=-19 which nearest to 300 and difference within range 5% to 10% 319,4 => 319+4=323 => 323-300=-23 which nearest to 300 and difference within range 5% to 10% 90,99,97 => 90+99+95=284 => 284-300=16 which nearest to 300 and difference within range 5% to 10%

the result will be
319,
319,4
90,99,95

i have tried with modifying the recursive algorithm ( Efficient algorithm to find a combination, which summation is equal to a known number, in a set of number ) but it is able to return only few matched sequence and not all.

Code:

   public static IEnumerable<string> GetSequence(decimal[] set, decimal? sum, decimal? startPercent, decimal? endPercent, string values = "")
    {            
        for (int i = 0; i < set.Length; i++)
        {
            decimal? left = sum - set[i];
            string vals = set[i] + "," + values;
            if (Math.Abs(decimal.Parse(left.ToString())) >= startPercent && Math.Abs(decimal.Parse(left.ToString())) <= endPercent)
            {
                yield return vals;
            }
            else
            {
                decimal[] possible = set.Take(i).Where(n => n <= sum).ToArray();
                if (possible.Length > 0)
                {
                    foreach (string s in GetSequence(possible, left, startPercent, endPercent, vals))
                    {
                        yield return s;
                    }
                }
            }
        }
    }

Could anybody help me with this.

Answer 1

Possibly a better approach is to generate all possible combinations using code like so:

public static IEnumerable<IEnumerable<T>> Combinations<T>(IList<T> items)
{
    return Combinations(items.Count).Select(comb => comb.Select(index => items[index]));
}

public static IEnumerable<IEnumerable<int>> Combinations(int n)
{
    long m = 1 << n;

    for (long i = 1; i < m; ++i)
        yield return bitIndices((uint)i);
}

static IEnumerable<int> bitIndices(uint n)
{
    uint mask = 1;

    for (int bit = 0; bit < 32; ++bit, mask <<= 1)
        if ((n & mask) != 0)
            yield return bit;
}

Then you can write a method to sum each possible combination:

static IEnumerable<(int Sum, List<int> Values)> SummedCombinations(IList<int> values)
{
    return 
        Combinations(values)
        .Select(comb => comb.ToList())
        .Select(comb => (comb.Sum(), comb));
}

Then you can write a method to find all the combinations where the sum matches the range you're looking for:

static IEnumerable<List<int>> FindMatches(IList<int> values, int target, int toleranceLow, int toleranceHigh)
{
    int minDiff = (target * toleranceLow)  / 100;
    int maxDiff = (target * toleranceHigh) / 100;

    foreach (var sum in SummedCombinations(values))
    {
        int diff = Math.Abs(sum.Sum - target);

        if (minDiff <= diff && diff <= maxDiff)
            yield return sum.Values;
    }
}

Putting this all together into a compilable console app:

using System;
using System.Collections.Generic;
using System.Linq;

namespace ConsoleApp1
{
    class Program
    {
        static void Main()
        {
            int[] values = {319, 4, 90, 50, 20, 99, 500, 95, 900};

            foreach (var combination in FindMatches(values, 300, 5, 10))
            {
                Console.WriteLine(string.Join(", ", combination));
            }
        }

        static IEnumerable<List<int>> FindMatches(IList<int> values, int target, int toleranceLow, int toleranceHigh)
        {
            int minDiff = (target * toleranceLow)  / 100;
            int maxDiff = (target * toleranceHigh) / 100;

            foreach (var sum in SummedCombinations(values))
            {
                int diff = Math.Abs(sum.Sum - target);

                if (minDiff <= diff && diff <= maxDiff)
                    yield return sum.Values;
            }
        }

        static IEnumerable<(int Sum, List<int> Values)> SummedCombinations(IList<int> values)
        {
            return 
                Combinations(values)
                .Select(comb => comb.ToList())
                .Select(comb => (comb.Sum(), comb));
        }

        public static IEnumerable<IEnumerable<T>> Combinations<T>(IList<T> items)
        {
            return Combinations(items.Count).Select(comb => comb.Select(index => items[index]));
        }

        public static IEnumerable<IEnumerable<int>> Combinations(int n)
        {
            long m = 1 << n;

            for (long i = 1; i < m; ++i)
                yield return bitIndices((uint)i);
        }

        static IEnumerable<int> bitIndices(uint n)
        {
            uint mask = 1;

            for (int bit = 0; bit < 32; ++bit, mask <<= 1)
                if ((n & mask) != 0)
                    yield return bit;
        }
    }
}

This outputs:

319
319, 4
90, 99, 95

Which is your expected output.

---

Note: The code above is using C# 7 tuples - if you are using an earlier version you'll have to change FindMatches() and SummedCombinations() to:

static IEnumerable<List<int>> FindMatches(IList<int> values, int target, int toleranceLow, int toleranceHigh)
{
    int minDiff = (target * toleranceLow)  / 100;
    int maxDiff = (target * toleranceHigh) / 100;

    foreach (var sum in SummedCombinations(values))
    {
        int diff = Math.Abs(sum.Item1 - target);

        if (minDiff <= diff && diff <= maxDiff)
            yield return sum.Item2;
    }
}

static IEnumerable<Tuple<int, List<int>>> SummedCombinations(IList<int> values)
{
    return 
        Combinations(values)
        .Select(comb => comb.ToList())
        .Select(comb => Tuple.Create(comb.Sum(), comb));
}

---

Explanation of the Combinations part

The combinations work as follows:

• Select i from 1 to 2^N-1 where N is the number of items to combine.
• For each bit set in i , return the item from the corresponding location in the input values.

So for example if you have 3 values; A, B and C:

i will go from 1 to (2^3-1) = 7.

Look at the binary values we will get for 1..7, and look at the corresponding elements of the A, B, C input:

C B A (Input)
2 1 0 (Bit number, i.e. power of two)
---------------------------------------
0 0 1 [1] = A
0 1 0 [2] = B
0 1 1 [3] = A B
1 0 0 [4] = C
1 0 1 [5] = A C
1 1 0 [5] = B C
1 1 1 [6] = A B C