Given an array: [dog, cat, mouse]
what is the most elegant way to create:
[,,]
[,,mouse]
[,cat,]
[,cat,mouse]
[dog,,]
[dog,,mouse]
[dog,
Easy to understand version (with descriptions)
I assumed that source = {1,2,3,4}
public static IEnumerable<IEnumerable<T>> GetSubSets<T>(IEnumerable<T> source)
{
var result = new List<IEnumerable<T>>() { new List<T>() }; // empty cluster added
for (int i = 0; i < source.Count(); i++)
{
var elem = source.Skip(i).Take(1);
// for elem = 2
// and currently result = [ [],[1] ]
var matchUps = result.Select(x => x.Concat(elem));
//then matchUps => [ [2],[1,2] ]
result = result.Concat(matchUps).ToList();
// matchUps and result concat operation
// finally result = [ [],[1],[2],[1,2] ]
}
return result;
}
string[] source = new string[] { "dog", "cat", "mouse" };
for (int i = 0; i < Math.Pow(2, source.Length); i++)
{
string[] combination = new string[source.Length];
for (int j = 0; j < source.Length; j++)
{
if ((i & (1 << (source.Length - j - 1))) != 0)
{
combination[j] = source[j];
}
}
Console.WriteLine("[{0}, {1}, {2}]", combination[0], combination[1], combination[2]);
}
Elegant? Why not Linq it.
public static IEnumerable<IEnumerable<T>> SubSetsOf<T>(IEnumerable<T> source)
{
if (!source.Any())
return Enumerable.Repeat(Enumerable.Empty<T>(), 1);
var element = source.Take(1);
var haveNots = SubSetsOf(source.Skip(1));
var haves = haveNots.Select(set => element.Concat(set));
return haves.Concat(haveNots);
}
Here is a variant of mqp's answer, that uses as state a BigInteger instead of an int
, to avoid overflow for collections containing more than 30 elements:
using System.Numerics;
public static IEnumerable<IEnumerable<T>> GetSubsets<T>(IList<T> source)
{
BigInteger combinations = BigInteger.One << source.Count;
for (BigInteger i = 0; i < combinations; i++)
{
yield return Enumerable.Range(0, source.Count)
.Select(j => (i & (BigInteger.One << j)) != 0 ? source[j] : default);
}
}
The way this is written, it is more of a Product (Cartesian product) rather than a list of all subsets.
You have three sets: (Empty,"dog"), (Empty,"cat"),(Empty,"mouse")
.
There are several posts on general solutions for products. As noted though, since you really just have 2 choices for each axis a single bit can represent the presence or not of the item.
So the total set of sets is all numbers from 0
to 2^N-1
. If N < 31
an int will work.
This is a small change to Mehrdad's solution above:
static IEnumerable<T[]> GetSubsets<T>(T[] set) {
bool[] state = new bool[set.Length+1];
for (int x; !state[set.Length]; state[x] = true ) {
yield return Enumerable.Range(0, state.Length)
.Where(i => state[i])
.Select(i => set[i])
.ToArray();
for (x = 0; state[x]; state[x++] = false);
}
}
or with pointers
static IEnumerable<T[]> GetSubsets<T>(T[] set) {
bool[] state = new bool[set.Length+1];
for (bool *x; !state[set.Length]; *x = true ) {
yield return Enumerable.Range(0, state.Length)
.Where(i => state[i])
.Select(i => set[i])
.ToArray();
for (x = state; *x; *x++ = false);
}
}