public bool UniformTwoDimensionalDistributionTest(IGenerator random, int sampleSize, int t)
{
List <double> sample = SampleGenerator.GenerateSample(random, sampleSize);
double
deltaR = 0.5 / (t - 1),
hi = 0;
int[] frequency = new int[t];
int m = sampleSize / 2;
for (int i = 0; i < sampleSize; i += 2)
{
double
x = sample[i],
y = sample[i + 1];
bool isInnerPoint = false;
for (int j = 0; j < t - 1; j++)
{
if (Math.Pow(x - 0.5, 2) + Math.Pow(y - 0.5, 2)
< Math.Pow(deltaR * (j + 1), 2))
{
frequency[j]++;
isInnerPoint = true;
break;
}
}
if (!isInnerPoint)
{
frequency[t - 1]++;
}
}
double p;
for (int i = 0; i < t - 1; i++)
{
p = Math.PI *
(Math.Pow(deltaR * (i + 1), 2) - Math.Pow(deltaR * i, 2));
hi += Math.Pow((frequency[i] - m * p), 2) / (m * p);
}
p = 1 - Math.PI * 0.25;
hi += Math.Pow((frequency[t - 1] - m * p), 2) / (m * p);
double P = 1 - MathNet.Numerics.Distributions.ChiSquared.CDF(t - 1, hi);
return(epsilon < P);
}
public bool CovarianceTest(IGenerator random, int sampleSize, int t)
{
List <double> sample = SampleGenerator.GenerateSample(random, sampleSize);
List <double>
R = new List <double>(), //^cov
P = new List <double>();
double E = sample.Average();
for (int i = 0; i < sampleSize; i++)
{
int to = sampleSize - i;
double sum = 0;
for (int j = 0; j < to; j++)
{
sum += sample[j] * sample[j + sampleSize - to];
}
R.Add(1.0 / (sampleSize - i - 1) * sum
- sampleSize / (sampleSize - 1) * E * E);
}
double
r = 1.0 / 12, //cov
c = Math.Sqrt(2);
P.Add(2 * (1.0 - NormalDistribution(12 * Math.Sqrt(sampleSize - 1)
* Math.Abs(R[0] - r) / c)));
r = 0;
c = 1;
for (int i = 1; i <= t; i++)
{
P.Add(2 * (1.0 - NormalDistribution(12 * Math.Sqrt(sampleSize - 1)
* Math.Abs(R[i] - r) / c)));
}
for (int i = 0; i <= t; i++)
{
if (epsilon >= P[i])
{
return(false);
}
}
return(true);
}
private void runButton_Click(object sender, EventArgs e)
{
methodLabel.Text = "";
element100Label.Text = "";
element900Label.Text = "";
element1000Label.Text = "";
gmmTestLabel.Text = "";
covariationLabel.Text = "";
uniOfTwoTestLabel.Text = "";
dispersionPatternBox.Visible = false;
if (nextDistribution == 0)
{
methodLabel.Text = "Мультипликативный конгруэнтный метод";
rand = new RandomBSV(24149775, 19581355);
}
else
{
methodLabel.Text = "Метод Макларена-Марсальи";
rand = new MacLarenRandom(
new RandomBSV(24149775, 19581355),
new RandomBSV(179029053, 457816087));
}
sample = SampleGenerator.GenerateSample(rand, sampleSize);
element100Label.Text = sample[99].ToString();
element900Label.Text = sample[899].ToString();
element1000Label.Text = sample[999].ToString();
PaintChart();
dispersionPatternBox.Visible = true;
TestAccuracy();
nextDistribution++;
nextDistribution %= 2;
}
public bool TestGMM(IGenerator random, int sampleSize)
{
List <double> sample = SampleGenerator.GenerateSample(random, sampleSize);
double
E = sample.Average(),
D = 0;
for (int i = 0; i < sampleSize; i++)
{
D += Math.Pow(E - sample[i], 2);
}
D /= sampleSize - 1;
double[] x = new double[2];
x[0] = E - 0.5;
x[1] = D - 1 / 12;
double[] c = new double[2];
c[0] = Math.Sqrt(12 * sampleSize);
c[1] = (sampleSize - 1) / sampleSize / Math.Sqrt((0.0056 / sampleSize
+ 0.0028 / Math.Pow(sampleSize, 2) - 0.0083 / Math.Pow(sampleSize, 3)));
double[] P = new double[2];
bool[] res = new bool[2];
for (int i = 0; i < 2; i++)
{
P[i] = 2 * (1 - NormalDistribution(c[i] * Math.Abs(x[i])));
res[i] = epsilon < P[i];
}
return(res[0] && res[1]);
}