Question

static IEnumerable<T> FindUniqueNumbersInCollection<T>(ICollection<T> value)
{
    Dictionary<T, byte> hash = new Dictionary<T, byte>();

    foreach (T val in value)
    {
        if (hash.ContainsKey(val)) { hash[val] = 1; continue; }
        hash.Add(val, 0);
    }

    List<T> unique = new List<T>();

    foreach (KeyValuePair<T, byte> kvp in hash)
    {
        if (kvp.Value == 0) unique.Add(kvp.Key);

    }

    return unique;
}

Answer 1

Edit:

I think this is finally correct. 🙂

var dict = new Dictionary<T, bool>();
foreach (var v in value)
    dict[v] = dict.ContainsKey(v);
foreach (var pair in dict)
    if (!pair.Value)
        yield return pair.Key;  //Algorithmically, as fast as possible

Or if you’d like some LINQ:

var dict = new Dictionary<T, bool>();
foreach (var v in value)
    dict[v] = dict.ContainsKey(v);
return dict.Keys.Where(k => !dict[k]);  //*COULD* be slower on lots of collisions

Or even

var dict = new Dictionary<T, bool>();
foreach (var v in value)
    dict[v] = dict.ContainsKey(v);
return dict.Where(p => !p.Value).Select(p => p.Key); //"Fastest".. but not really

I wouldn’t say it’s "more efficient" than yours (it’s really not — they’re pretty much the same), but it’s definitely shorter.

---

If you want overkill efficiency (at the cost of accuracy), you can always use a Bloom Filter!