public static BaseDistribution GetSum(BivariateContinuousDistribution distribution)
{
return(Bivariate(distribution, DistributionsOperation.Add));
}
private static DiscreteDistribution Bivariate(BivariateContinuousDistribution distribution, DistributionsOperation operation)
{
if (operation == DistributionsOperation.PowerInv)
{
distribution = distribution.Rotate();
}
int samples = distribution.Samples;
double[] range = CommonRandomMath.GetRange(distribution.SupportMinLeft, distribution.SupportMaxLeft, distribution.SupportMinRight, distribution.SupportMaxRight, operation);
double[] rightAxis = CommonRandomMath.GenerateXAxis(distribution.SupportMinRight, distribution.SupportMaxRight, samples, out double rightStep);
double[] xAxis = CommonRandomMath.GenerateXAxis(range[0], range[1], samples, out double step);
double[] result = new double[samples];
switch (operation)
{
case DistributionsOperation.Add:
{
Parallel.For(0, xAxis.Length, i =>
{
double x = xAxis[i];
double sum = 0;
// Trap rule is useless because both normal and t-distributions are smoooth.
for (int j = 1; j < samples; j++)
{
double m = rightAxis[j];
sum += distribution.ProbabilityDensityFunction(x - m, m);
}
result[i] = sum * rightStep;
});
break;
}
case DistributionsOperation.Sub:
{
Parallel.For(0, xAxis.Length, i =>
{
double x = xAxis[i];
double sum = 0;
for (int j = 1; j < samples; j++)
{
double m = rightAxis[j];
sum += distribution.ProbabilityDensityFunction(x + m, m);
}
result[i] = sum * rightStep;
});
break;
}
case DistributionsOperation.Muliply:
{
Parallel.For(0, xAxis.Length, i =>
{
double x = xAxis[i];
double sum = 0;
for (int j = 1; j < samples; j++)
{
double m = rightAxis[j];
if (m != 0)
{
sum += distribution.ProbabilityDensityFunction(x / m, m) / Math.Abs(m);
}
}
result[i] = sum * rightStep;
});
break;
}
case DistributionsOperation.Divide:
{
Parallel.For(0, xAxis.Length, i =>
{
double x = xAxis[i];
double sum = 0;
for (int j = 1; j < samples; j++)
{
double m = rightAxis[j];
if (m != 0)
{
sum += distribution.ProbabilityDensityFunction(x * m, m) * Math.Abs(m);
}
}
result[i] = sum * rightStep;
});
break;
}
case DistributionsOperation.PowerInv:
{
Parallel.For(0, xAxis.Length, i =>
{
double m = 0;
double sum = 0;
double x = xAxis[i];
double d = 0;
double k = 0;
for (int j = 1; j < samples; j++)
{
m = rightAxis[j];
d = Math.Log(x, m);
k = Math.Abs(Math.Log(m) * x);
sum += distribution.ProbabilityDensityFunction(d, m) / k;
}
result[i] = sum * rightStep;
});
break;
}
case DistributionsOperation.Log:
{
Parallel.For(0, xAxis.Length, i =>
{
double m = 0;
double sum = 0;
double x = xAxis[i];
double d = 0;
double k = 0;
for (int j = 1; j < samples; j++)
{
m = rightAxis[j];
d = Math.Pow(m, x);
k = Math.Abs(Math.Log(m) * d);
sum += distribution.ProbabilityDensityFunction(d, m) * k;
}
result[i] = sum * rightStep;
});
break;
}
default:
{
throw new DistributionsInvalidOperationException();
}
}
return(new DiscreteDistribution(xAxis, result));
}
public static BaseDistribution GetPower(BivariateContinuousDistribution distribution)
{
return(Bivariate(distribution, DistributionsOperation.PowerInv));
}
public static BaseDistribution GetRatio(BivariateContinuousDistribution distribution)
{
return(Bivariate(distribution, DistributionsOperation.Divide));
}
public static BaseDistribution GetProduct(BivariateContinuousDistribution distribution)
{
return(Bivariate(distribution, DistributionsOperation.Muliply));
}
public static BaseDistribution GetDifference(BivariateContinuousDistribution distribution)
{
return(Bivariate(distribution, DistributionsOperation.Sub));
}