public int CompareTo(object obj) // implement IComparable to enable sort by ranking
{
int result = 1;
if (obj != null && obj is PoissonSample)
{
PoissonSample p = obj as PoissonSample;
result = this.ranking.CompareTo(p.ranking);
}
return(result);
}
/// <summary>
/// Toroidal Dart Throwing is when we apply the minimum-distance criterion also
/// wrapping around on the edges of the plane, so that we can put the resulting
/// distributions next to each other without clustering on the seams.
/// </summary>
private void ToroidalDartThrowing(int desiredSamples,
float initialRadius, // e.g. 0.15f
int attemptsPerRadius, // e.g. 1000
float radiusDecreaseFactor // e.g. 0.99f
)
{
Random random = new Random();
samples = new List <PoissonSample>();
float radius = initialRadius;
while (samples.Count < desiredSamples && radius > 0.0f)
{
int attemtps = 0;
for (; attemtps < attemptsPerRadius; attemtps++)
{
PoissonSample candidate = new PoissonSample((float)random.NextDouble(),
(float)random.NextDouble(),
radius,
(float)samples.Count / desiredSamples);
// As opposed to regular dart throwing, we plan to compose
// tiles made out of these Poisson distributions, meaning
// that a given distribution may fit with itself on the
// opposite edges. The 'toroidal' bit is in charge of fixing
// this, by extending the distance checking wrapping around edges
if (IsCandidateValid(candidate, radius))
{
samples.Add(candidate);
if (UpdateFrequency > 0 && samples.Count % UpdateFrequency == 0 && sampleAdded != null)
{
sampleAdded(this, EventArgs.Empty);
}
break;
}
}
if (attemtps == attemptsPerRadius)
{
radius *= radiusDecreaseFactor;
}
}
}
private bool IsCandidateValid(PoissonSample c, float minDist)
{
float minDist2 = minDist * minDist;
float cx = c.x;
float cy = c.y;
bool res = true;
for (int i = 0; res && i < samples.Count; i++)
{
// regular dart throwing condition
if (samples[i].DistSquared(c) < minDist2)
{
res = false;
}
// wrapped around edges
c.x = cx + 1.0f;
if (res && samples[i].DistSquared(c) < minDist2)
{
res = false;
}
c.x = cx - 1.0f;
if (res && samples[i].DistSquared(c) < minDist2)
{
res = false;
}
c.x = cx;
c.y = cy + 1.0f;
if (res && samples[i].DistSquared(c) < minDist2)
{
res = false;
}
c.y = cy - 1.0f;
if (res && samples[i].DistSquared(c) < minDist2)
{
res = false;
}
c.y = cy;
}
return(res);
}
public float DistSquared(PoissonSample s)
{
return((x - s.x) * (x - s.x) + (y - s.y) * (y - s.y));
}
