/// <summary>
/// Adds items to the Markov chain, including termini.
/// </summary>
/// <param name="items">The items to add.</param>
public void AddItems(IEnumerable <T> items)
{
Queue <T> previous = new Queue <T>();
// Add each item to the chain.
foreach (T item in items)
{
AddItem(previous, item);
previous.Enqueue(item);
if (previous.Count > order)
{
previous.Dequeue();
}
}
// Add to the terminal weights
MarkovNode <T> key = new MarkovNode <T>(previous);
if (!terminalWeights.ContainsKey(key))
{
terminalWeights.Add(key, 0);
}
terminalWeights[key]++;
}
/// <summary>
/// Gets an series of items, generated from the Markov chain.
/// </summary>
/// <returns>An IEnumerable of items.</returns>
public IEnumerable <T> Generate()
{
Random random = new Random();
Queue <T> previous = new Queue <T>();
while (true)
{
while (previous.Count > order)
{
previous.Dequeue();
}
MarkovNode <T> key = new MarkovNode <T>(previous);
Dictionary <T, int> weights;
// If terminus is reached.
if (!chain.TryGetValue(key, out weights))
{
yield break;
}
int terminalWeight = 0;
terminalWeights.TryGetValue(key, out terminalWeight);
int sumWeights = weights.Sum(w => w.Value);
int randomValue = random.Next(sumWeights + terminalWeight) + 1;
// If terminus is chosen.
if (randomValue > sumWeights)
{
yield break;
}
// Loop through the chain, adding the weights of each step.
// When 'randomValue' is less than or equal to the cumulative
// weights so far, the current item is chosen.
int cumulativeWeight = 0;
foreach (var pair in weights)
{
cumulativeWeight += pair.Value;
if (randomValue <= cumulativeWeight)
{
yield return(pair.Key);
previous.Enqueue(pair.Key);
break;
}
}
}
}
/// <summary>
/// Adds an item to the Markov chain.
/// </summary>
/// <param name="previous">The previous items in the chain.</param>
/// <param name="item">The item to add.</param>
private void AddItem(Queue <T> previous, T item)
{
if (previous.Count > order)
{
throw new ArgumentException("The queue is longer than the Markov chain order", "previous");
}
MarkovNode <T> key = new MarkovNode <T>(previous);
if (!chain.ContainsKey(key))
{
chain.Add(key, new Dictionary <T, int>());
}
if (!chain[key].ContainsKey(item))
{
chain[key].Add(item, 0);
}
chain[key][item]++;
}
public bool Equals(MarkovNode <T> other)
{
if (other == null)
{
return(false);
}
if (items.Length != other.items.Length)
{
return(false);
}
for (int i = 0; i < items.Length; i++)
{
if (!items[i].Equals(other.items[i]))
{
return(false);
}
}
return(true);
}
