| ... | ... |
@@ -769,74 +769,70 @@ |
| 769 | 769 |
double V1=1.0-real<double>(); |
| 770 | 770 |
double V2=1.0-real<double>(); |
| 771 | 771 |
if(V2<=v0) |
| 772 | 772 |
{
|
| 773 | 773 |
xi=std::pow(V1,1.0/delta); |
| 774 | 774 |
nu=V0*std::pow(xi,delta-1.0); |
| 775 | 775 |
} |
| 776 | 776 |
else |
| 777 | 777 |
{
|
| 778 | 778 |
xi=1.0-std::log(V1); |
| 779 | 779 |
nu=V0*std::exp(-xi); |
| 780 | 780 |
} |
| 781 | 781 |
} while(nu>std::pow(xi,delta-1.0)*std::exp(-xi)); |
| 782 | 782 |
return theta*(xi-gamma(int(std::floor(k)))); |
| 783 | 783 |
} |
| 784 | 784 |
|
| 785 | 785 |
/// Weibull distribution |
| 786 | 786 |
|
| 787 | 787 |
/// This function generates a Weibull distribution random number. |
| 788 | 788 |
/// |
| 789 | 789 |
///\param k shape parameter (<tt>k>0</tt>) |
| 790 | 790 |
///\param lambda scale parameter (<tt>lambda>0</tt>) |
| 791 | 791 |
/// |
| 792 | 792 |
double weibull(double k,double lambda) |
| 793 | 793 |
{
|
| 794 | 794 |
return lambda*pow(-std::log(1.0-real<double>()),1.0/k); |
| 795 | 795 |
} |
| 796 | 796 |
|
| 797 | 797 |
/// Pareto distribution |
| 798 | 798 |
|
| 799 | 799 |
/// This function generates a Pareto distribution random number. |
| 800 | 800 |
/// |
| 801 |
///\param k shape parameter (<tt>k>0</tt>) |
|
| 801 | 802 |
///\param x_min location parameter (<tt>x_min>0</tt>) |
| 802 |
///\param k shape parameter (<tt>k>0</tt>) |
|
| 803 | 803 |
/// |
| 804 |
///\warning This function used inverse transform sampling, therefore may |
|
| 805 |
///suffer from numerical unstability. |
|
| 806 |
/// |
|
| 807 |
///\todo Implement a numerically stable method |
|
| 808 |
double pareto(double |
|
| 804 |
double pareto(double k,double x_min) |
|
| 809 | 805 |
{
|
| 810 |
return |
|
| 806 |
return exponential(gamma(k,1.0/x_min)); |
|
| 811 | 807 |
} |
| 812 | 808 |
|
| 813 | 809 |
///@} |
| 814 | 810 |
|
| 815 | 811 |
///\name Two dimensional distributions |
| 816 | 812 |
/// |
| 817 | 813 |
|
| 818 | 814 |
///@{
|
| 819 | 815 |
|
| 820 | 816 |
/// Uniform distribution on the full unit circle. |
| 821 | 817 |
dim2::Point<double> disc() |
| 822 | 818 |
{
|
| 823 | 819 |
double V1,V2; |
| 824 | 820 |
do {
|
| 825 | 821 |
V1=2*real<double>()-1; |
| 826 | 822 |
V2=2*real<double>()-1; |
| 827 | 823 |
|
| 828 | 824 |
} while(V1*V1+V2*V2>=1); |
| 829 | 825 |
return dim2::Point<double>(V1,V2); |
| 830 | 826 |
} |
| 831 | 827 |
/// A kind of two dimensional Gauss distribution |
| 832 | 828 |
|
| 833 | 829 |
/// This function provides a turning symmetric two-dimensional distribution. |
| 834 | 830 |
/// Both coordinates are of standard normal distribution, but they are not |
| 835 | 831 |
/// independent. |
| 836 | 832 |
/// |
| 837 | 833 |
/// \note The coordinates are the two random variables provided by |
| 838 | 834 |
/// the Box-Muller method. |
| 839 | 835 |
dim2::Point<double> gauss2() |
| 840 | 836 |
{
|
| 841 | 837 |
double V1,V2,S; |
| 842 | 838 |
do {
|
0 comments (0 inline)