public static void chain(ref typeMethods.r8vecNormalData data, ref int seed, int n, ref double[] xem, ref double[] vem, ref double diff)
//****************************************************************************80
//
// Purpose:
//
// CHAIN tests the stochastic Chain Rule.
//
// Discussion:
//
// This function solves a stochastic differential equation for
//
// V = sqrt(X)
//
// where X satisfies the stochastic differential equation:
//
// dX = ( alpha - X ) * dt + beta * sqrt(X) dW,
// X(0) = Xzero,
//
// with
//
// alpha = 2,
// beta = 1,
// Xzero = 1.
//
// From the stochastic Chain Rule, the SDE for V is therefore:
//
// dV = ( ( 4 * alpha - beta^2 ) / ( 8 * V ) - 1/2 V ) dt + 1/2 beta dW
// V(0) = sqrt ( Xzero ).
//
// Xem is the Euler-Maruyama solution for X.
//
// Vem is the Euler-Maruyama solution of the SDE for V from
// the stochastic Chain Rule.
//
// Hence, we compare sqrt(Xem) and Vem.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input, int &SEED, a seed for the random number generator.
//
// Input, int N, the number of time steps.
//
// Output, double XEM[N+1], the computed value of X.
//
// Output, double VEM[N+1], the computed value of V.
//
// Output, double &DIFF, the maximum value of |sqrt(XEM)-V|.
//
{
int i;
int j;
//
// Set problem parameters.
//
const double alpha = 2.0;
const double beta = 1.0;
//
// Stepping parameters.
// dt2 is the size of the Euler-Maruyama steps.
//
const double tmax = 1.0;
double dt = tmax / n;
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
for (i = 0; i < n; i++)
{
dw[i] = Math.Sqrt(dt) * dw[i];
}
//
// Solve for X(t).
//
xem[0] = 1.0;
for (j = 1; j <= n; j++)
{
xem[j] = xem[j - 1] + (alpha - xem[j - 1]) * dt
+ beta * Math.Sqrt(xem[j - 1]) * dw[j - 1];
}
//
// Solve for V(t).
//
vem[0] = Math.Sqrt(xem[0]);
for (j = 1; j <= n; j++)
{
vem[j] = vem[j - 1]
+ ((4.0 * alpha - beta * beta) / (8.0 * vem[j - 1])
- 0.5 * vem[j - 1]) * dt
+ 0.5 * beta * dw[j - 1];
}
//
// Compare sqrt(X) and V.
//
diff = 0.0;
for (i = 0; i <= n; i++)
{
diff = Math.Max(diff, Math.Abs(Math.Sqrt(xem[i]) - vem[i]));
}
}
public static double[] hypersphere01_sample(int m, int n, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// HYPERSPHERE01_SAMPLE uniformly samples the surface of the unit hypersphere.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 04 January 2014
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Russell Cheng,
// Random Variate Generation,
// in Handbook of Simulation,
// edited by Jerry Banks,
// Wiley, 1998, pages 168.
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int M, the spatial dimension.
//
// Input, int N, the number of points.
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Output, double X[M*N], the points.
//
{
int j;
double[] x = typeMethods.r8mat_normal_01_new(m, n, ref data, ref seed);
for (j = 0; j < n; j++)
{
//
// Compute the length of the vector.
//
double norm = 0.0;
int i;
for (i = 0; i < m; i++)
{
norm += Math.Pow(x[i + j * m], 2);
}
norm = Math.Sqrt(norm);
//
// Normalize the vector.
//
for (i = 0; i < m; i++)
{
x[i + j * m] /= norm;
}
}
return(x);
}
public static void stab_meansquare(ref typeMethods.r8vecNormalData vdata, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// STAB_MEANSQUARE examines mean-square stability.
//
// Discussion:
//
// The function tests the mean-square stability
// of the Euler-Maruyama method applied to a stochastic differential
// equation (SDE).
//
// The SDE is
//
// dX = lambda*X dt + mu*X dW,
// X(0) = Xzero,
//
// where
//
// lambda is a constant,
// mu is a constant,
// Xzero = 1.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input, int &SEED, a seed for the random number generator.
// In the reference, this value is set to 100.
//
{
const string command_filename = "stab_meansquare_commands.txt";
List <string> command = new();
const string data_filename0 = "stab_meansquare0_data.txt";
List <string> data = new();
int k;
Console.WriteLine("");
Console.WriteLine("STAB_MEANSQUARE:");
Console.WriteLine(" Investigate mean square stability of Euler-Maruyama");
Console.WriteLine(" solution with stepsize DT and MU.");
Console.WriteLine("");
Console.WriteLine(" SDE is mean square stable if");
Console.WriteLine(" Real ( lambda + 1/2 |mu|^2 ) < 0.");
Console.WriteLine("");
Console.WriteLine(" EM with DT is mean square stable if");
Console.WriteLine(" |1+dt^2| + dt * |mu|^2 - 1.0 < 0.");
//
// Set problem parameters.
//
const double tmax = 20.0;
const int m = 50000;
const double xzero = 1.0;
//
// Problem parameters.
//
const double lambda = -3.0;
double mu = Math.Sqrt(3.0);
//
// Test the SDE.
//
Console.WriteLine("");
Console.WriteLine(" Lambda = " + lambda + "");
Console.WriteLine(" Mu = " + mu + "");
double test = lambda + 0.5 * mu * mu;
Console.WriteLine(" SDE mean square stability test = " + test + "");
//
// XMS is the mean square estimate of M paths.
//
string data_filename = data_filename0;
for (k = 0; k < 3; k++)
{
double dt = Math.Pow(2.0, -k);
int n = 20 * (int)Math.Pow(2, k);
//
// Test the EM for this DT.
//
Console.WriteLine("");
Console.WriteLine(" dt = " + dt + "");
test = Math.Pow(1.0 + dt * lambda, 2) + dt * mu * mu - 1.0;
Console.WriteLine(" EM mean square stability test = " + test + "");
double[] xms = new double[n + 1];
double[] xtemp = new double[m];
int i;
for (i = 0; i < m; i++)
{
xtemp[i] = xzero;
}
xms[0] = xzero;
int j;
for (j = 0; j <= n; j++)
{
double[] winc = typeMethods.r8vec_normal_01_new(m, ref vdata, ref seed);
for (i = 0; i < m; i++)
{
winc[i] = Math.Sqrt(dt) * winc[i];
}
for (i = 0; i < m; i++)
{
xtemp[i] = xtemp[i]
+ dt * lambda * xtemp[i]
+ mu * xtemp[i] * winc[i];
}
xms[j] = 0.0;
for (i = 0; i < m; i++)
{
xms[j] += xtemp[i] * xtemp[i];
}
xms[j] /= m;
}
//
// Write this data to a file.
//
Files.filename_inc(ref data_filename);
for (j = 0; j <= n; j++)
{
double t = tmax * j / n;
data.Add(" " + t
+ " " + xms[j] + "");
}
File.WriteAllLines(data_filename, data);
Console.WriteLine("");
Console.WriteLine(" Data for DT = " + dt + " stored in \"" + data_filename + "\".");
command.Add("# stab_meansquare_commands.txt");
command.Add("# created by sde::stab_meansquare.");
command.Add("#");
command.Add("# Usage:");
command.Add("# gnuplot < stab_meansquare_commands.txt");
command.Add("#");
command.Add("set term png");
command.Add("set output 'stab_meansquare.png'");
command.Add("set xlabel 't'");
command.Add("set ylabel 'E|X^2(t)|'");
command.Add("set title 'Mean Square of EM Solution'");
command.Add("set grid");
command.Add("set logscale y 10");
command.Add("set style data lines");
data_filename = data_filename0;
Files.filename_inc(ref data_filename);
command.Add("plot '" + data_filename + "' using 1:2, \\");
for (k = 1; k <= 1; k++)
{
Files.filename_inc(ref data_filename);
command.Add(" '" + data_filename + "' using 1:2, \\");
}
Files.filename_inc(ref data_filename);
command.Add(" '" + data_filename + "' using 1:2");
command.Add("quit");
File.WriteAllLines(command_filename, command);
Console.WriteLine(" STAB_MEANSQUARE plot commands stored in \"" + command_filename + "\".");
}
}
public static double[] disk_sample(double[] center, double r, int n, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// DISK_SAMPLE uniformly samples the general disk in 2D.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 05 July 2018
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, double CENTER[2], coordinates of the center.
// This information is not actually needed to compute the area.
//
// Input, double R, the radius of the disk.
//
// Input, int N, the number of points.
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Output, double X[2*N], the points.
//
{
int j;
double[] x = new double[2 * n];
for (j = 0; j < n; j++)
{
double[] v = typeMethods.r8vec_normal_01_new(2, ref data, ref seed);
//
// Compute the length of the vector.
//
double norm = Math.Sqrt(Math.Pow(v[0], 2) + Math.Pow(v[1], 2));
//
// Normalize the vector.
//
int i;
for (i = 0; i < 2; i++)
{
v[i] /= norm;
}
//
// Now compute a value to map the point ON the circle INTO the circle.
//
double r2 = UniformRNG.r8_uniform_01(ref seed);
for (i = 0; i < 2; i++)
{
x[i + j * 2] = center[i] + r * Math.Sqrt(r2) * v[i];
}
}
return(x);
}
public static double[] bpath(ref typeMethods.r8vecNormalData data, ref int seed, int n)
//****************************************************************************80
//
// Purpose:
//
// BPATH performs a Brownian path simulation.
//
// Discussion:
//
// This routine computes one simulation of discretized Brownian
// motion over the time interval [0,1] using N time steps.
// The user specifies a random number seed. Different values of
// the seed will result in different realizations of the path.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original MATLAB version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random number
// generator.
//
// Input, int N, the number of steps.
//
// Output, double BPATH[N+1], the Brownian path.
//
{
int j;
const double tmax = 1.0;
double dt = tmax / n;
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
for (j = 0; j < n; j++)
{
dw[j] = Math.Sqrt(dt) * dw[j];
}
//
// W is the sum of the previous increments.
//
double[] w = new double[n + 1];
w[0] = 0.0;
for (j = 1; j < n; j++)
{
w[j] = w[j - 1] + dw[j];
}
return(w);
}
public static double[] mc(double s0, double e, double r, double sigma, double t1, int m,
ref typeMethods.r8vecNormalData data,
ref int seed)
//****************************************************************************80
//
// Purpose:
//
// MC uses Monte Carlo valuation on a European call.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 08 June 2016
//
// Author:
//
// Original MATLAB version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// Black-Scholes for Scientific Computing Students,
// Computing in Science and Engineering,
// November/December 2004, Volume 6, Number 6, pages 72-79.
//
// Parameters:
//
// Input, double S0, the asset price at time 0.
//
// Input, double E, the exercise price.
//
// Input, double R, the interest rate.
//
// Input, double SIGMA, the volatility of the asset.
//
// Input, double T1, the expiry date.
//
// Input, int M, the number of simulations.
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Output, double MC[2], the estimated range of the valuation.
//
{
int i;
double[] u = typeMethods.r8vec_normal_01_new(m, ref data, ref seed);
double[] svals = new double[m];
for (i = 0; i < m; i++)
{
svals[i] = s0 * Math.Exp((r - 0.5 * sigma * sigma) * t1
+ sigma * Math.Sqrt(t1) * u[i]);
}
double[] pvals = new double[m];
for (i = 0; i < m; i++)
{
pvals[i] = Math.Exp(-r * t1) * Math.Max(svals[i] - e, 0.0);
}
double pmean = 0.0;
for (i = 0; i < m; i++)
{
pmean += pvals[i];
}
pmean /= m;
double std = 0.0;
for (i = 0; i < m; i++)
{
std += Math.Pow(pvals[i] - pmean, 2);
}
std = Math.Sqrt(std / (m - 1));
double width = 1.96 * std / Math.Sqrt(m);
double[] conf = new double[2];
conf[0] = pmean - width;
conf[1] = pmean + width;
return(conf);
}
public static double[] uniform_on_sphere01_map(int dim_num, int n, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// UNIFORM_ON_SPHERE01_MAP maps uniform points onto the unit sphere.
//
// Discussion:
//
// The sphere has center 0 and radius 1.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 12 November 2010
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Russell Cheng,
// Random Variate Generation,
// in Handbook of Simulation,
// edited by Jerry Banks,
// Wiley, 1998, pages 168.
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int DIM_NUM, the dimension of the space.
//
// Input, int N, the number of points.
//
// Input/output, int &SEED, a seed for the random number generator.
//
// Output, double UNIFORM_ON_SPHERE01_MAP[DIM_NUM*N], the points.
//
{
int j;
double[] x = typeMethods.r8mat_normal_01_new(dim_num, n, ref data, ref seed);
for (j = 0; j < n; j++)
{
//
// Compute the length of the vector.
//
double norm = 0.0;
int i;
for (i = 0; i < dim_num; i++)
{
norm += x[i + j * dim_num] * x[i + j * dim_num];
}
norm = Math.Sqrt(norm);
//
// Normalize the vector.
//
for (i = 0; i < dim_num; i++)
{
x[i + j * dim_num] /= norm;
}
}
return(x);
}
public static void emweak(ref typeMethods.r8vecNormalData data, ref int seed, int method, int m, int p_max, ref double[] dtvals,
ref double[] xerr)
//****************************************************************************80
//
// Purpose:
//
// EMWEAK tests the weak convergence of the Euler-Maruyama method.
//
// Discussion:
//
// The SDE is
//
// dX = lambda * X dt + mu * X dW,
// X(0) = Xzero,
//
// where
//
// lambda = 2,
// mu = 1,
// Xzero = 1.
//
// The discretized Brownian path over [0,1] has dt = 2^(-9).
//
// The Euler-Maruyama method will use 5 different timesteps:
//
// 2^(p-10), p = 1,2,3,4,5.
//
// We examine weak convergence at T=1:
//
// | E (X_L) - E (X(T)) |.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 25 September 2012
//
// Author:
//
// Original MATLAB version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546
//
// Parameters:
//
// Input, int &SEED, a seed for the random number generator.
//
// Input, int METHOD.
// 0, use the standard Euler-Maruyama method;
// 1, use the weak Euler-Maruyama method.
//
// Input, int M, the number of simulations to perform.
// A typical value is M = 1000.
//
// Input, int P_MAX, the number of time step sizes to use.
// A typical value is 5.
//
// Output, double DTVALS[P_MAX], the time steps used.
//
// Output, double XERR[P_MAX], the averaged absolute error in the
// solution estimate at the final time.
//
{
int i;
int p;
//
// Problem parameters;
//
const double lambda = 2.0;
const double mu = 0.1;
const double xzero = 1.0;
//
// Stepping parameters.
//
for (p = 0; p < p_max; p++)
{
dtvals[p] = Math.Pow(2.0, p - 9);
}
//
// Take various Euler timesteps.
// For stepsize dt, we will need to take L Euler steps to reach time TMAX.
//
double[] xtemp = new double[m];
double[] xem = new double[p_max];
for (p = 0; p < p_max; p++)
{
int l = (int)Math.Pow(2, 9 - p);
double dt = dtvals[p];
for (i = 0; i < m; i++)
{
xtemp[i] = xzero;
}
int j;
for (j = 0; j < l; j++)
{
double[] winc = typeMethods.r8vec_normal_01_new(m, ref data, ref seed);
switch (method)
{
case 0:
{
for (i = 0; i < m; i++)
{
winc[i] = Math.Sqrt(dt) * winc[i];
}
break;
}
default:
{
for (i = 0; i < m; i++)
{
winc[i] = Math.Sqrt(dt) * typeMethods.r8_sign(winc[i]);
}
break;
}
}
for (i = 0; i < m; i++)
{
xtemp[i] = xtemp[i] + dt * lambda * xtemp[i]
+ mu * xtemp[i] * winc[i];
}
}
//
// Average the M results for this stepsize.
//
xem[p] = typeMethods.r8vec_mean(m, xtemp);
}
//
// Compute the error in the estimates for each stepsize.
//
for (p = 0; p < p_max; p++)
{
xerr[p] = Math.Abs(xem[p] - Math.Exp(lambda));
}
//
// Least squares fit of error = c * dt^q.
//
double[] a = new double[p_max * 2];
double[] rhs = new double[p_max];
for (i = 0; i < p_max; i++)
{
a[i + 0 * p_max] = 1.0;
a[i + 1 * p_max] = Math.Log(dtvals[i]);
rhs[i] = Math.Log(xerr[i]);
}
double[] sol = QRSolve.qr_solve(p_max, 2, a, rhs);
Console.WriteLine("");
Console.WriteLine("EMWEAK:");
switch (method)
{
case 0:
Console.WriteLine(" Using standard Euler-Maruyama method.");
break;
default:
Console.WriteLine(" Using weak Euler-Maruyama method.");
break;
}
Console.WriteLine(" Least squares solution to Error = c * dt ^ q");
Console.WriteLine(" (Expecting Q to be about 1.)");
Console.WriteLine(" Computed Q = " + sol[1] + "");
double resid = 0.0;
for (i = 0; i < p_max; i++)
{
double e = a[i + 0 * p_max] * sol[0] + a[i + 1 * p_max] * sol[1] - rhs[i];
resid += e * e;
}
resid = Math.Sqrt(resid);
Console.WriteLine(" Residual is " + resid + "");
}
public static double[] sample_paths2_cholesky(int n, int n2, double rhomin, double rhomax,
double rho0, Func <int, int, double[], double[], double, double[]> correlation2, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// SAMPLE_PATHS2_CHOLESKY: sample paths for stationary correlation functions.
//
// Discussion:
//
// This method uses the Cholesky factorization of the correlation matrix.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 12 November 2012
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int N, the number of points on each path.
//
// Input, int N2, the number of paths.
//
// Input, double RHOMIN, RHOMAX, the range of RHO.
//
// Input, double RHO0, the correlation length.
//
// Input, double *CORRELATION2 ( int m, int n, double s[], double t[],
// double rho0 ), the name of the function which evaluates the correlation.
//
// Input/output, int &SEED, a seed for the random number
// generator.
//
// Output, double X[N*N2], the sample paths.
//
{
int flag = 0;
//
// Choose N equally spaced sample points from RHOMIN to RHOMAX.
//
double[] s = typeMethods.r8vec_linspace_new(n, rhomin, rhomax);
//
// Evaluate the correlation function.
//
double[] cor = correlation2(n, n, s, s, rho0);
//
// Get the Cholesky factorization of COR:
//
// COR = L * L'.
//
double[] l = typeMethods.r8mat_cholesky_factor(n, cor, ref flag);
switch (flag)
{
//
// The matrix might not be nonnegative definite.
//
case 2:
Console.WriteLine("");
Console.WriteLine("SAMPLE_PATHS2_CHOLESKY - Fatal error!");
Console.WriteLine(" The correlation matrix is not");
Console.WriteLine(" symmetric nonnegative definite.");
return(null);
}
//
// Compute a matrix of N by N2 normally distributed values.
//
double[] r = typeMethods.r8mat_normal_01_new(n, n2, ref data, ref seed);
//
// Compute the sample path.
//
double[] x = typeMethods.r8mat_mm_new(n, n, n2, l, r);
return(x);
}
public static Correlation.CorrelationResult sample_paths_cholesky(FullertonLib.BesselData globaldata, FullertonLib.r8BESKData kdata, int n, int n2, double rhomax, double rho0,
Func <FullertonLib.BesselData, FullertonLib.r8BESKData, int, double[], double, Correlation.CorrelationResult> correlation, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// SAMPLE_PATHS_CHOLESKY: sample paths for stationary correlation functions.
//
// Discussion:
//
// This method uses the Cholesky factorization of the correlation matrix.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 11 November 2012
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int N, the number of points on each path.
//
// Input, int N2, the number of paths.
//
// Input, double RHOMAX, the maximum value of RHO.
//
// Input, double RHO0, the correlation length.
//
// Input, double *CORRELATION ( int n, double rho_vec[], double rho0),
// the name of the function which evaluates the correlation.
//
// Input/output, int &SEED, a seed for the random number
// generator.
//
// Output, double X[N*N2], the sample paths.
//
{
int flag = 0;
int j;
//
// Choose N equally spaced sample points from 0 to RHOMAX.
//
const double rhomin = 0.0;
double[] rho_vec = typeMethods.r8vec_linspace_new(n, rhomin, rhomax);
//
// Evaluate the correlation function.
//
Correlation.CorrelationResult tr = correlation(globaldata, kdata, n, rho_vec, rho0);
double[] cor_vec = tr.result;
globaldata = tr.data;
//
// Construct the correlation matrix;
//
// From the vector
// [ C(0), C(1), C(2), ... C(N-1) ]
// construct the vector
// [ C(N-1), ..., C(2), C(1), C(0), C(1), C(2), ... C(N-1) ]
// Every row of the correlation matrix can be constructed by a subvector
// of this vector.
//
double[] cor = new double[n * n];
for (j = 0; j < n; j++)
{
int i;
for (i = 0; i < n; i++)
{
int k = typeMethods.i4_wrap(j - i, 0, n - 1);
cor[i + j * n] = cor_vec[k];
}
}
//
// Get the Cholesky factorization of COR:
//
// COR = L * L'.
//
double[] l = typeMethods.r8mat_cholesky_factor(n, cor, ref flag);
switch (flag)
{
//
// The matrix might not be nonnegative definite.
//
case 2:
Console.WriteLine("");
Console.WriteLine("SAMPLE_PATHS_CHOLESKY - Fatal error!");
Console.WriteLine(" The correlation matrix is not");
Console.WriteLine(" symmetric nonnegative definite.");
return(null);
}
//
// Compute a matrix of N by N2 normally distributed values.
//
double[] r = typeMethods.r8mat_normal_01_new(n, n2, ref data, ref seed);
//
// Compute the sample path.
//
double[] x = typeMethods.r8mat_mm_new(n, n, n2, l, r);
Correlation.CorrelationResult res = new() { result = x, data = globaldata };
return(res);
}
public static Correlation.CorrelationResult sample_paths_eigen(FullertonLib.BesselData globaldata, FullertonLib.r8BESJ0Data jdata, int n, int n2, double rhomax, double rho0,
Func <FullertonLib.BesselData, FullertonLib.r8BESJ0Data, int, double[], double, Correlation.CorrelationResult> correlation, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// SAMPLE_PATHS_EIGEN: sample paths for stationary correlation functions.
//
// Discussion:
//
// This method uses the eigen-decomposition of the correlation matrix.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 12 November 2012
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int N, the number of points on each path.
//
// Input, int N2, the number of paths.
//
// Input, double RHOMAX, the maximum value of RHO.
//
// Input, double RHO0, the correlation length.
//
// Input, double *CORRELATION ( int n, double rho_vec[], double rho0),
// the name of the function which evaluates the correlation.
//
// Input/output, int &SEED, a seed for the random number
// generator.
//
// Output, double X[N*N2], the sample paths.
//
{
int i;
int j;
int k;
//
// Choose N equally spaced sample points from 0 to RHOMAX.
//
const double rhomin = 0.0;
double[] rho_vec = typeMethods.r8vec_linspace_new(n, rhomin, rhomax);
//
// Evaluate the correlation function.
//
Correlation.CorrelationResult tr = correlation(globaldata, jdata, n, rho_vec, rho0);
double[] cor_vec = tr.result;
globaldata = tr.data;
jdata = tr.j0data;
//
// Construct the correlation matrix;
//
// From the vector
// [ C(0), C(1), C(2), ... C(N-1) ]
// construct the vector
// [ C(N-1), ..., C(2), C(1), C(0), C(1), C(2), ... C(N-1) ]
// Every row of the correlation matrix can be constructed by a subvector
// of this vector.
//
double[] cor = new double[n * n];
for (j = 0; j < n; j++)
{
for (i = 0; i < n; i++)
{
k = typeMethods.i4_wrap(Math.Abs(i - j), 0, n - 1);
cor[i + j * n] = cor_vec[k];
}
}
//
// Get the eigendecomposition of COR:
//
// COR = V * D * V'.
//
// Because COR is symmetric, V is orthogonal.
//
double[] d = new double[n];
double[] w = new double[n];
double[] v = new double[n * n];
TRED2.tred2(n, cor, ref d, ref w, ref v);
TQL2.tql2(n, ref d, ref w, v);
//
// We assume COR is non-negative definite, and hence that there
// are no negative eigenvalues.
//
double dmin = typeMethods.r8vec_min(n, d);
if (dmin < -Math.Sqrt(typeMethods.r8_epsilon()))
{
Console.WriteLine("");
Console.WriteLine("SAMPLE_PATHS_EIGEN - Warning!");
Console.WriteLine(" Negative eigenvalues observed as low as " + dmin + "");
}
for (i = 0; i < n; i++)
{
d[i] = Math.Max(d[i], 0.0);
}
//
// Compute the eigenvalues of the factor C.
//
for (i = 0; i < n; i++)
{
d[i] = Math.Sqrt(d[i]);
}
//
// Compute C, such that C' * C = COR.
//
double[] c = new double[n * n];
for (j = 0; j < n; j++)
{
for (i = 0; i < n; i++)
{
c[i + j * n] = 0.0;
for (k = 0; k < n; k++)
{
c[i + j * n] += d[k] * v[i + k * n] * v[j + k * n];
}
}
}
//
// Compute N by N2 independent random normal values.
//
double[] r = typeMethods.r8mat_normal_01_new(n, n2, ref data, ref seed);
//
// Multiply to get the variables X which have correlation COR.
//
double[] x = typeMethods.r8mat_mm_new(n, n, n2, c, r);
Correlation.CorrelationResult result = new() { result = x, data = globaldata };
return(result);
}
public static double[] sample_paths2_eigen(int n, int n2, double rhomin, double rhomax,
double rho0, Func <int, int, double[], double[], double, double[]> correlation2, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// SAMPLE_PATHS2_EIGEN: sample paths for stationary correlation functions.
//
// Discussion:
//
// This method uses the eigen-decomposition of the correlation matrix.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 12 November 2012
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int N, the number of points on each path.
//
// Input, int N2, the number of paths.
//
// Input, double RHOMIN, RHOMAX, the range of RHO.
//
// Input, double RHO0, the correlation length.
//
// Input, double *CORRELATION2 ( int m, int n, double s[], double t[],
// double rho0 ), the name of the function which evaluates the correlation.
//
// Input/output, int &SEED, a seed for the random number
// generator.
//
// Output, double X[N*N2], the sample paths.
//
{
int i;
int j;
//
// Choose N equally spaced sample points from RHOMIN to RHOMAX.
//
double[] s = typeMethods.r8vec_linspace_new(n, rhomin, rhomax);
//
// Evaluate the correlation function.
//
double[] cor = correlation2(n, n, s, s, rho0);
//
// Get the eigendecomposition of COR:
//
// COR = V * D * V'.
//
// Because COR is symmetric, V is orthogonal.
//
double[] d = new double[n];
double[] w = new double[n];
double[] v = new double[n * n];
TRED2.tred2(n, cor, ref d, ref w, ref v);
TQL2.tql2(n, ref d, ref w, v);
//
// We assume COR is non-negative definite, and hence that there
// are no negative eigenvalues.
//
double dmin = typeMethods.r8vec_min(n, d);
if (dmin < -Math.Sqrt(typeMethods.r8_epsilon()))
{
Console.WriteLine("");
Console.WriteLine("SAMPLE_PATHS2_EIGEN - Warning!");
Console.WriteLine(" Negative eigenvalues observed as low as " + dmin + "");
}
for (i = 0; i < n; i++)
{
d[i] = Math.Max(d[i], 0.0);
}
//
// Compute the eigenvalues of the factor C.
//
for (i = 0; i < n; i++)
{
d[i] = Math.Sqrt(d[i]);
}
//
// Compute C, such that C' * C = COR.
//
double[] c = new double[n * n];
for (j = 0; j < n; j++)
{
for (i = 0; i < n; i++)
{
c[i + j * n] = 0.0;
int k;
for (k = 0; k < n; k++)
{
c[i + j * n] += d[k] * v[i + k * n] * v[j + k * n];
}
}
}
//
// Compute N by N2 independent random normal values.
//
double[] r = typeMethods.r8mat_normal_01_new(n, n2, ref data, ref seed);
//
// Multiply to get the variables X which have correlation COR.
//
double[] x = typeMethods.r8mat_mm_new(n, n, n2, c, r);
return(x);
}
public static void em(ref typeMethods.r8vecNormalData data, ref int seed, int n, ref double[] t, ref double[] xtrue, ref double[] t2,
ref double[] xem, ref double error)
//****************************************************************************80
//
// Purpose:
//
// EM applies the Euler-Maruyama method to a linear SDE.
//
// Discussion:
//
// The SDE is
//
// dX = lambda * X dt + mu * X dW,
// X(0) = Xzero,
//
// where
//
// lambda = 2,
// mu = 1,
// Xzero = 1.
//
// The discretized Brownian path over [0,1] uses
// a stepsize dt = 2^(-8).
//
// The Euler-Maruyama method uses a larger timestep Dt = R*dt,
// where R is an integer. For an SDE of the form
//
// dX = f(X(t)) dt + g(X(t)) dW(t)
//
// it has the form
//
// X(j) = X(j-1) + f(X(j-1)) * Dt + g(X(j-1)) * ( W(j*Dt) - W((j-1)*Dt) )
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Input, int N, the number of time steps. A typical
// value is 2^8. N should be a multiple of 4.
//
// Output, double T[N+1], the time values for the exact solution.
//
// Output, double XTRUE[N+1], the exact solution.
//
// Output, double T2[N/4+1], the time values for the
// Euler-Maruyama solution.
//
// Output, double XEM[N/4+1], the Euler-Maruyama solution.
//
// Output, double &ERROR, the value of | XEM(T) - XTRUE(T) |.
//
{
int j;
//
// Set problem parameters.
//
const double lambda = 2.0;
const double mu = 1.0;
const double xzero = 1.0;
//
// Set stepping parameters.
//
const double tmax = 1.0;
double dt = tmax / n;
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
for (j = 0; j < n; j++)
{
dw[j] = Math.Sqrt(dt) * dw[j];
}
//
// Sum the Brownian increments.
//
double[] w = new double[n + 1];
w[0] = 0.0;
for (j = 1; j <= n; j++)
{
w[j] = w[j - 1] + dw[j - 1];
}
for (j = 0; j <= n; j++)
{
t[j] = j * tmax / n;
}
//
// Compute the discretized Brownian path.
//
for (j = 0; j <= n; j++)
{
xtrue[j] = xzero * Math.Exp((lambda - 0.5 * mu * mu) * (t[j] + mu * w[j]));
}
//
// Set:
// R, the multiplier for the EM step,
// Dt, the EM stepsize,
// L, the number of EM steps (we need N to be a multiple of R!)
//
const int r = 4;
double dt2 = r * dt;
int l = n / r;
for (j = 0; j <= l; j++)
{
t2[j] = j * tmax / l;
}
//
// Compute XEM.
//
xem[0] = xzero;
for (j = 1; j <= l; j++)
{
double dw2 = 0.0;
int i;
for (i = r * (j - 1); i < r * j; i++)
{
dw2 += dw[i];
}
xem[j] = xem[j - 1] + dt2 * lambda * xem[j - 1] + mu * xem[j - 1] * dw2;
}
error = Math.Abs(xem[l] - xtrue[n]);
}
public static double[] ellipsoid_sample(int m, int n, double[] a, double[] v, double r,
ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// ELLIPSOID_SAMPLE samples points uniformly from an ellipsoid.
//
// Discussion:
//
// The points X in the ellipsoid are described by a M by M
// positive definite symmetric matrix A, a "center" V, and
// a "radius" R, such that
//
// (X-V)' * A * (X-V) <= R * R
//
// The algorithm computes the Cholesky factorization of A:
//
// A = U' * U.
//
// A set of uniformly random points Y is generated, satisfying:
//
// Y' * Y <= R * R.
//
// The appropriate points in the ellipsoid are found by solving
//
// U * X = Y
// X = X + V
//
// Thanks to Dr Karl-Heinz Keil for pointing out that the original
// coding was actually correct only if A was replaced by its inverse.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 14 August 2014
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int M, the dimension of the space.
//
// Input, int N, the number of points.
//
// Input, double A[M*M], the matrix that describes
// the ellipsoid.
//
// Input, double V[M], the "center" of the ellipsoid.
//
// Input, double R, the "radius" of the ellipsoid.
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Output, double ELLIPSE_SAMPLE[M*N], the points.
//
{
int i;
int j;
double[] u;
//
// Get the Cholesky factor U.
//
try
{
u = typeMethods.r8po_fa(m, a);
}
catch (Exception)
{
Console.WriteLine("");
Console.WriteLine("ELLIPSOID_SAMPLE - Fatal error!");
Console.WriteLine(" R8PO_FA reports that the matrix A");
Console.WriteLine(" is not positive definite symmetric.");
return(null);
}
//
// Get the points Y that satisfy Y' * Y <= 1.
//
double[] x = Uniform.Sphere.uniform_in_sphere01_map(m, n, ref data, ref seed);
//
// Get the points Y that satisfy Y' * Y <= R * R.
//
for (j = 0; j < n; j++)
{
for (i = 0; i < m; i++)
{
x[i + j * m] = r * x[i + j * m];
}
}
//
// Solve U * X = Y.
//
for (j = 0; j < n; j++)
{
double[] t = typeMethods.r8po_sl(m, u, x, bIndex: +j * m);
for (i = 0; i < m; i++)
{
x[i + j * m] = t[i];
}
}
//
// X = X + V.
//
for (j = 0; j < n; j++)
{
for (i = 0; i < m; i++)
{
x[i + j * m] += v[i];
}
}
return(x);
}
public static double[] uniform_in_sphere01_map(int m, int n, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// UNIFORM_IN_SPHERE01_MAP maps uniform points into the unit sphere.
//
// Discussion:
//
// The sphere has center 0 and radius 1.
//
// We first generate a point ON the sphere, and then distribute it
// IN the sphere.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 14 August 2014
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Russell Cheng,
// Random Variate Generation,
// in Handbook of Simulation,
// edited by Jerry Banks,
// Wiley, 1998, pages 168.
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int M, the dimension of the space.
//
// Input, int N, the number of points.
//
// Input/output, int &SEED, a seed for the random number generator.
//
// Output, double X[M*N], the points.
//
{
int j;
double exponent = 1.0 / m;
double[] x = new double[m * n];
for (j = 0; j < n; j++)
{
//
// Fill a vector with normally distributed values.
//
double[] v = typeMethods.r8vec_normal_01_new(m, ref data, ref seed);
//
// Compute the length of the vector.
//
double norm = typeMethods.r8vec_norm(m, v);
//
// Normalize the vector.
//
int i;
for (i = 0; i < m; i++)
{
v[i] /= norm;
}
//
// Now compute a value to map the point ON the sphere INTO the sphere.
//
double r = UniformRNG.r8_uniform_01(ref seed);
r = Math.Pow(r, exponent);
for (i = 0; i < m; i++)
{
x[i + j * m] = r * v[i];
}
}
return(x);
}
public static double[] uniform_in_ellipsoid_map(int dim_num, int n, double[] a, double r,
ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// UNIFORM_IN_ELLIPSOID_MAP maps uniform points into an ellipsoid.
//
// Discussion:
//
// The points X in the ellipsoid are described by a DIM_NUM by DIM_NUM positive
// definite symmetric matrix A, and a "radius" R, such that
//
// X' * A * X <= R * R
//
// The algorithm computes the Cholesky factorization of A:
//
// A = U' * U.
//
// A set of uniformly random points Y is generated, satisfying:
//
// Y' * Y <= R * R.
//
// The appropriate points in the ellipsoid are found by solving
//
// U * X = Y
//
// Thanks to Dr Karl-Heinz Keil for pointing out that the original
// coding was actually correct only if A was replaced by its inverse.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 23 May 2005
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int DIM_NUM, the dimension of the space.
//
// Input, int N, the number of points.
//
// Input, double A[DIM_NUM*DIM_NUM], the matrix that describes the ellipsoid.
//
// Input, double R, the right hand side of the ellipsoid equation.
//
// Input/output, int &SEED, a seed for the random number generator.
//
// Output, double UNIFORM_IN_ELLIPSOID_MAP[DIM_NUM*N], the points.
//
{
int i;
int j;
//
// Get the upper triangular Cholesky factor U of A.
//
double[] u = new double[dim_num * dim_num];
for (j = 0; j < dim_num; j++)
{
for (i = 0; i < dim_num; i++)
{
u[i + j * dim_num] = a[i + j * dim_num];
}
}
int info = typeMethods.r8po_fa(ref u, dim_num, dim_num);
if (info != 0)
{
Console.WriteLine("");
Console.WriteLine("UNIFORM_IN_ELLIPSOID_MAP - Fatal error!");
Console.WriteLine(" R8PO_FA reports that the matrix A");
Console.WriteLine(" is not positive definite symmetric.");
return(null);
}
//
// Get the points Y that satisfy Y' * Y <= R * R.
//
double[] x = Sphere.uniform_in_sphere01_map(dim_num, n, ref data, ref seed);
for (j = 0; j < n; j++)
{
for (i = 0; i < dim_num; i++)
{
x[i + j * dim_num] = r * x[i + j * dim_num];
}
}
//
// Solve U * X = Y.
//
for (j = 0; j < n; j++)
{
typeMethods.r8po_sl(u, dim_num, dim_num, ref x, bIndex: +j * dim_num);
}
return(x);
}
public static void stochastic_integral_strat(int n, ref typeMethods.r8vecNormalData data, ref int seed, ref double estimate,
ref double exact, ref double error)
//****************************************************************************80
//
// Purpose:
//
// STOCHASTIC_INTEGRAL_STRAT approximates the Stratonovich integral of W(t) dW.
//
// Discussion:
//
// This function estimates the Stratonovich integral of W(t) dW over
// the interval [0,1].
//
// The estimates is made by taking N steps.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input, int N, the number of steps to take.
//
// Input, int &SEED, a seed for the random number generator.
//
// Output, double &ESTIMATE, the estimate of the integral.
//
// Output, double &EXACT, the exact value of the integral.
//
// Output, double &ERROR, the error in the integral estimate.
//
{
int j;
//
// Set step parameters.
//
const double tmax = 1.0;
double dt = tmax / n;
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
for (j = 0; j < n; j++)
{
dw[j] = Math.Sqrt(dt) * dw[j];
}
//
// Sum the increments to get the Brownian path.
//
double[] w = new double[n + 1];
w[0] = 0.0;
for (j = 1; j <= n; j++)
{
w[j] = w[j - 1] + dw[j - 1];
}
//
// Approximate the Stratonovich integral.
//
double[] u = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
double[] v = new double[n];
for (j = 0; j < n; j++)
{
v[j] = 0.5 * (w[j] + w[j + 1]) + 0.5 * Math.Sqrt(dt) * u[j];
}
estimate = typeMethods.r8vec_dot_product(n, v, dw);
//
// Compare with the exact solution.
//
exact = 0.5 * w[n - 1] * w[n - 1];
error = Math.Abs(estimate - exact);
}
public static void bpath_average(ref typeMethods.r8vecNormalData data, ref int seed, int m, int n, ref double[] u, ref double[] umean,
ref double error)
//****************************************************************************80
//
// Purpose:
//
// BPATH_AVERAGE: displays the average of 1000 Brownian paths.
//
// Discussion:
//
// This routine computes M simulations of discretized Brownian
// motion W(t) over the time interval [0,1] using N time steps.
// The user specifies a random number seed. Different values of
// the seed will result in a different set of realizations of the path.
//
// Actually, we are interested in a function u(W(t)):
//
// u(W(t)) = exp ( t + W(t)/2 )
//
// The routine plots 5 of the simulations, as well as the average
// of all the simulations.
//
// The plot of the average should be quite smooth. Its expected
// value is exp ( 9 * t / 8 ), and we compute the 'error', that is,
// the difference between the averaged value and this expected
// value. This 'error' should decrease as the number of simulation
// is increased.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Input, int M, the number of simulations to compute
// and average. A typical value is 1000.
//
// Input, int N, the number of steps. A typical value
// is 500.
//
// Output, double U[M*(N+1)], the M paths.
//
// Output, double UMEAN[N+1], the averaged path.
//
// Output, double &ERROR, the maximum difference between the
// averaged path and the exact expected value.
//
{
int i;
int j;
const double tmax = 1.0;
double dt = tmax / n;
double[] t = new double[n + 1];
for (j = 0; j <= n; j++)
{
t[j] = j * tmax / n;
}
double[] w = new double[n + 1];
for (i = 0; i < m; i++)
{
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
for (j = 0; j < n; j++)
{
dw[j] = Math.Sqrt(dt) * dw[j];
}
//
// W is the sum of the previous increments.
//
w[0] = 0.0;
for (j = 1; j <= n; j++)
{
w[j] = w[j - 1] + dw[j - 1];
}
for (j = 0; j <= n; j++)
{
u[i + j * m] = Math.Exp(t[j] + 0.5 * w[j]);
}
}
//
// Average the M estimates of the path.
//
for (j = 0; j <= n; j++)
{
umean[j] = 0.0;
for (i = 0; i < m; i++)
{
umean[j] += u[i + j * m];
}
umean[j] /= m;
}
error = 0.0;
for (j = 0; j <= n; j++)
{
error = Math.Max(error, Math.Abs(umean[j] - Math.Exp(9.0 * t[j] / 8.0)));
}
}
public static void milstrong(ref typeMethods.r8vecNormalData data, ref int seed, int p_max, ref double[] dtvals, ref double[] xerr)
//****************************************************************************80
//
// Purpose:
//
// MILSTRONG tests the strong convergence of the Milstein method.
//
// Discussion:
//
// This function solves the stochastic differential equation
//
// dX = sigma * X * ( k - X ) dt + beta * X dW,
// X(0) = Xzero,
//
// where
//
// sigma = 2,
// k = 1,
// beta = 1,
// Xzero = 0.5.
//
// The discretized Brownian path over [0,1] has dt = 2^(-11).
//
// The Milstein method uses timesteps 128*dt, 64*dt, 32*dt, 16*dt
// (also dt for reference).
//
// We examine strong convergence at T=1:
//
// E | X_L - X(T) |.
//
// The code is vectorized: all paths computed simultaneously.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original MATLAB version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Input, int P_MAX, the number of time step sizes to use.
// A typical value is 4.
//
// Output, double DTVALS[P_MAX], the time steps used.
//
// Output, double XERR[P_MAX], the averaged absolute error in the
// solution estimate at the final time.
//
{
int i;
int j;
int p;
//
// Set problem parameters.
//
const double sigma = 2.0;
const double k = 1.0;
const double beta = 0.25;
const double xzero = 0.5;
//
// Set stepping parameters.
//
const double tmax = 1.0;
int n = (int)Math.Pow(2, 11);
double dt = tmax / n;
//
// Number of paths sampled.
//
const int m = 500;
//
// Define the increments dW.
//
double[] dw = typeMethods.r8mat_normal_01_new(m, n, ref data, ref seed);
for (j = 0; j < n; j++)
{
for (i = 0; i < m; i++)
{
dw[i + j * m] = Math.Sqrt(dt) * dw[i + j * m];
}
}
//
// Estimate the reference solution at time T M times.
//
double[] xref = new double[m];
for (i = 0; i < m; i++)
{
xref[i] = xzero;
}
for (j = 0; j < n; j++)
{
for (i = 0; i < m; i++)
{
xref[i] = xref[i]
+ dt * sigma * xref[i] * (k - xref[i])
+ beta * xref[i] * dw[i + j * m]
+ 0.5 * beta * beta * xref[i] * (dw[i + j * m] * dw[i + j * m] - dt);
}
}
//
// Now compute M Milstein approximations at each of 4 timesteps,
// and record the average errors.
//
for (p = 0; p < p_max; p++)
{
dtvals[p] = dt * 8.0 * Math.Pow(2.0, p + 1);
}
for (p = 0; p < p_max; p++)
{
xerr[p] = 0.0;
}
double[] xtemp = new double[m];
for (p = 0; p < p_max; p++)
{
int r = 8 * (int)Math.Pow(2, p + 1);
double dtp = dtvals[p];
int l = n / r;
for (i = 0; i < m; i++)
{
xtemp[i] = xzero;
}
for (j = 0; j < l; j++)
{
for (i = 0; i < m; i++)
{
double winc = 0.0;
int i2;
for (i2 = r * j; i2 < r * (j + 1); i2++)
{
winc += dw[i + i2 * m];
}
xtemp[i] = xtemp[i]
+ dtp * sigma * xtemp[i] * (k - xtemp[i])
+ beta * xtemp[i] * winc
+ 0.5 * beta * beta * xtemp[i] * (winc * winc - dtp);
}
}
xerr[p] = 0.0;
for (i = 0; i < m; i++)
{
xerr[p] += Math.Abs(xtemp[i] - xref[i]);
}
xerr[p] /= m;
}
//
// Least squares fit of error = C * dt^q
//
double[] a = new double[p_max * 2];
double[] rhs = new double[p_max];
for (p = 0; p < p_max; p++)
{
a[p + 0 * p_max] = 1.0;
a[p + 1 * p_max] = Math.Log(dtvals[p]);
rhs[p] = Math.Log(xerr[p]);
}
double[] sol = QRSolve.qr_solve(p_max, 2, a, rhs);
Console.WriteLine("");
Console.WriteLine("MILSTEIN:");
Console.WriteLine(" Least squares solution to Error = c * dt ^ q");
Console.WriteLine(" Expecting Q to be about 1.");
Console.WriteLine(" Computed Q = " + sol[1] + "");
double resid = 0.0;
for (i = 0; i < p_max; i++)
{
double e = a[i + 0 * p_max] * sol[0] + a[i + 1 * p_max] * sol[1] - rhs[i];
resid += e * e;
}
resid = Math.Sqrt(resid);
Console.WriteLine(" Residual is " + resid + "");
}
public static void emstrong(ref typeMethods.r8vecNormalData data, ref int seed, int m, int n, int p_max, ref double[] dtvals,
ref double[] xerr)
//****************************************************************************80
//
// Purpose:
//
// EMSTRONG tests the strong convergence of the EM method.
//
// Discussion:
//
// The SDE is
//
// dX = lambda * X dt + mu * X dW,
// X(0) = Xzero,
//
// where
//
// lambda = 2,
// mu = 1,
// Xzero = 1.
//
// The discretized Brownian path over [0,1] has dt = 2^(-9).
//
// The Euler-Maruyama method uses 5 different timesteps:
// 16*dt, 8*dt, 4*dt, 2*dt, dt.
//
// We are interested in examining strong convergence at T=1,
// that is
//
// E | X_L - X(T) |.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Input, int M, the number of simulations to perform.
// A typical value is M = 1000.
//
// Input, int N, the number of time steps to take.
// A typical value is N = 512.
//
// Input, int P_MAX, the number of time step sizes to use.
// A typical value is 5.
//
// Output, double DTVALS[P_MAX], the time steps used.
//
// Output, double XERR[P_MAX], the averaged absolute error in the
// solution estimate at the final time.
//
{
int i;
int p;
int s;
//
// Set problem parameters.
//
const double lambda = 2.0;
const double mu = 1.0;
const double xzero = 1.0;
//
// Set stepping parameters.
//
const double tmax = 1.0;
double dt = tmax / n;
for (p = 0; p < p_max; p++)
{
dtvals[p] = dt * Math.Pow(2.0, p);
}
//
// Sample over discrete Brownian paths.
//
for (p = 0; p < p_max; p++)
{
xerr[p] = 0.0;
}
for (s = 0; s < m; s++)
{
//
// Define the increments dW.
//
double[] dw = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
int j;
for (j = 0; j < n; j++)
{
dw[j] = Math.Sqrt(dt) * dw[j];
}
//
// Sum the increments to get the Brownian path.
//
double[] w = new double[n + 1];
w[0] = 0.0;
for (j = 1; j <= n; j++)
{
w[j] = w[j - 1] + dw[j - 1];
}
//
// Determine the true solution.
//
double xtrue = xzero * Math.Exp(lambda - 0.5 * mu * mu + mu * w[n]);
//
// Use the Euler-Maruyama method with 5 different time steps dt2 = r * dt
// to estimate the solution value at time TMAX.
//
for (p = 0; p < p_max; p++)
{
double dt2 = dtvals[p];
int r = (int)Math.Pow(2, p);
int l = n / r;
double xtemp = xzero;
for (j = 0; j < l; j++)
{
double winc = 0.0;
int k;
for (k = r * j; k < r * (j + 1); k++)
{
winc += dw[k];
}
xtemp = xtemp + dt2 * lambda * xtemp + mu * xtemp * winc;
}
xerr[p] += Math.Abs(xtemp - xtrue);
}
}
for (p = 0; p < p_max; p++)
{
xerr[p] /= m;
}
//
// Least squares fit of error = c * dt^q.
//
double[] a = new double[p_max * 2];
double[] rhs = new double[p_max];
for (i = 0; i < p_max; i++)
{
a[i + 0 * p_max] = 1.0;
a[i + 1 * p_max] = Math.Log(dtvals[i]);
rhs[i] = Math.Log(xerr[i]);
}
double[] sol = QRSolve.qr_solve(p_max, 2, a, rhs);
Console.WriteLine("");
Console.WriteLine("EMSTRONG:");
Console.WriteLine(" Least squares solution to Error = c * dt ^ q");
Console.WriteLine(" (Expecting Q to be about 1/2.)");
Console.WriteLine(" Computed Q = " + sol[1] + "");
double resid = 0.0;
for (i = 0; i < p_max; i++)
{
double e = a[i + 0 * p_max] * sol[0] + a[i + 1 * p_max] * sol[1] - rhs[i];
resid += e * e;
}
resid = Math.Sqrt(resid);
Console.WriteLine(" Residual is " + resid + "");
}
public static double[] ellipse_sample(int n, double[] a, double r, ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// ELLIPSE_SAMPLE samples points in an ellipse.
//
// Discussion:
//
// The points X in the ellipsoid are described by a 2 by 2 positive
// definite symmetric matrix A, and a "radius" R, such that
// X' * A * X <= R * R
// The algorithm computes the Cholesky factorization of A:
// A = U' * U.
// A set of uniformly random points Y is generated, satisfying:
// Y' * Y <= R * R.
// The appropriate points in the ellipsoid are found by solving
// U * X = Y
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 23 May 2005
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Reuven Rubinstein,
// Monte Carlo Optimization, Simulation, and Sensitivity
// of Queueing Networks,
// Krieger, 1992,
// ISBN: 0894647644,
// LC: QA298.R79.
//
// Parameters:
//
// Input, int N, the number of points.
//
// Input, double A[2*2], the matrix that describes the ellipse.
//
// Input, double R, the right hand side of the ellipse equation.
//
// Input/output, int *SEED, a seed for the random number generator.
//
// Output, double ELLIPSE_SAMPLE[2*N], the points.
//
{
int i;
int j;
const int m = 2;
//
// Get the upper triangular Cholesky factor U of A.
//
double[] u = new double[m * m];
for (j = 0; j < m; j++)
{
for (i = 0; i < m; i++)
{
u[i + j * m] = a[i + j * m];
}
}
int info = typeMethods.r8po_fa(ref u, m, m);
if (info != 0)
{
Console.WriteLine("");
Console.WriteLine("ELLIPSE_SAMPLE - Fatal error!");
Console.WriteLine(" R8PO_FA reports that the matrix A");
Console.WriteLine(" is not positive definite symmetric.");
return(null);
}
//
// Get the points Y that satisfy Y' * Y <= R * R.
//
double[] x = Uniform.Sphere.uniform_in_sphere01_map(m, n, ref data, ref seed);
for (j = 0; j < n; j++)
{
for (i = 0; i < m; i++)
{
x[i + j * m] = r * x[i + j * m];
}
}
//
// Solve U * X = Y.
//
for (j = 0; j < n; j++)
{
typeMethods.r8po_sl(u, m, m, ref x, bIndex: +j * m);
}
return(x);
}
public static double[] asset_path(double s0, double mu, double sigma, double t1, int n,
ref typeMethods.r8vecNormalData data, ref int seed)
//****************************************************************************80
//
// Purpose:
//
// ASSET_PATH simulates the behavior of an asset price over time.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 08 June 2016
//
// Author:
//
// Original MATLAB version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// Black-Scholes for Scientific Computing Students,
// Computing in Science and Engineering,
// November/December 2004, Volume 6, Number 6, pages 72-79.
//
// Parameters:
//
// Input, double S0, the asset price at time 0.
//
// Input, double MU, the expected growth rate.
//
// Input, double SIGMA, the volatility of the asset.
//
// Input, double T1, the expiry date.
//
// Input, integer N, the number of steps to take between 0 and T1.
//
// Input/output, int &SEED, a seed for the random number generator.
//
// Output, double ASSET_PATH[N+1], the option values from time 0 to T1
// in equal steps.
//
{
int i;
double dt = t1 / n;
double[] r = typeMethods.r8vec_normal_01_new(n, ref data, ref seed);
double[] s = new double[n + 1];
s[0] = s0;
double p = s0;
for (i = 1; i <= n; i++)
{
p *= Math.Exp((mu - sigma * sigma) * dt + sigma * Math.Sqrt(dt) * r[i - 1]);
s[i] = p;
}
return(s);
}
public static void stab_asymptotic(ref typeMethods.r8vecNormalData vdata, ref typeMethods.r8NormalData data, ref int seed, int n, int p_max)
//****************************************************************************80
//
// Purpose:
//
// STAB_ASYMPTOTIC examines asymptotic stability.
//
// Discussion:
//
// The function tests the asymptotic stability
// of the Euler-Maruyama method applied to a stochastic differential
// equation (SDE).
//
// The SDE is
//
// dX = lambda*X dt + mu*X dW,
// X(0) = Xzero,
//
// where
//
// lambda is a constant,
// mu is a constant,
// Xzero = 1.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 28 September 2012
//
// Author:
//
// Original Matlab version by Desmond Higham.
// C++ version by John Burkardt.
//
// Reference:
//
// Desmond Higham,
// An Algorithmic Introduction to Numerical Simulation of
// Stochastic Differential Equations,
// SIAM Review,
// Volume 43, Number 3, September 2001, pages 525-546.
//
// Parameters:
//
// Input/output, int &SEED, a seed for the random
// number generator.
//
// Input, int N, the number of time steps for the
// first solution.
//
// Input, int P_MAX, the number of time step sizes.
//
{
const string command_filename = "stab_asymptotic_commands.txt";
List <string> command = new();
const string data_filename0 = "stab_asymptotic0_data.txt";
List <string> out_data = new();
int p;
Console.WriteLine("");
Console.WriteLine("STAB_ASYMPTOTIC:");
Console.WriteLine(" Investigate asymptotic stability of Euler-Maruyama");
Console.WriteLine(" solution with stepsize DT and MU.");
Console.WriteLine("");
Console.WriteLine(" SDE is asymptotically stable if");
Console.WriteLine(" Real ( lambda - 1/2 mu^2 ) < 0.");
Console.WriteLine("");
Console.WriteLine(" EM with DT is asymptotically stable if");
Console.WriteLine(" E log ( | 1 + lambda dt - sqrt(dt) mu n(0,1) | ) < 0.");
Console.WriteLine(" where n(0,1) is a normal random value.");
//
// Problem parameters.
//
const double lambda = 0.5;
double mu = Math.Sqrt(6.0);
const double xzero = 1.0;
//
// Test the SDE.
//
Console.WriteLine("");
Console.WriteLine(" Lambda = " + lambda + "");
Console.WriteLine(" Mu = " + mu + "");
double test = lambda - 0.5 * mu * mu;
Console.WriteLine(" SDE asymptotic stability test = " + test + "");
//
// Step parameters.
//
const double tmax = 500.0;
//
// For each stepsize, compute the Euler-Maruyama solution.
//
string data_filename = data_filename0;
double[] dtvals = new double[p_max];
for (p = 0; p < p_max; p++)
{
int nval = n * (int)Math.Pow(2, p);
double dt = tmax / nval;
dtvals[p] = dt;
//
// Test the EM for this DT.
//
Console.WriteLine("");
Console.WriteLine(" dt = " + dt + "");
double[] u = typeMethods.r8vec_normal_01_new(1000, ref vdata, ref seed);
int i;
for (i = 0; i < 1000; i++)
{
u[i] = Math.Log(Math.Abs(1.0 + lambda * dt - Math.Sqrt(dt) * mu * u[i]));
}
test = typeMethods.r8vec_mean(1000, u);
Console.WriteLine(" EM asymptotic test = " + test + "");
double xtemp = xzero;
double[] xemabs = new double[nval + 1];
xemabs[0] = xtemp;
int j;
for (j = 1; j <= nval; j++)
{
double winc = Math.Sqrt(dt) * typeMethods.r8_normal_01(ref data, ref seed);
xtemp = xtemp + dt * lambda * xtemp + mu * xtemp * winc;
xemabs[j] = Math.Abs(xtemp);
}
//
// Write this data to a file.
//
Files.filename_inc(ref data_filename);
//
// We have to impose a tiny lower bound on the values because we
// will end up plotting their logs.
//
double xmin = Math.Exp(-200.0);
for (i = 0; i <= nval; i++)
{
double t = tmax * i / nval;
out_data.Add(" " + t
+ " " + Math.Max(xemabs[i], xmin) + "");
}
File.WriteAllLines(data_filename, out_data);
Console.WriteLine("");
Console.WriteLine(" Data for DT = " + dt + " stored in \"" + data_filename + "\"");
}
command.Add("# stab_asymptotic_commands.txt");
command.Add("# created by sde::stab_asymptotic.");
command.Add("#");
command.Add("# Usage:");
command.Add("# gnuplot < stab_asymptotic_commands.txt");
command.Add("#");
command.Add("set term png");
command.Add("set output 'stab_asymptotic.png'");
command.Add("set xlabel 't'");
command.Add("set ylabel '|X(t)|'");
command.Add("set title 'Absolute value of EM Solution'");
command.Add("set grid");
command.Add("set logscale y 10");
command.Add("set style data lines");
data_filename = data_filename0;
Files.filename_inc(ref data_filename);
command.Add("plot '" + data_filename + "' using 1:2, \\");
for (p = 1; p < p_max - 1; p++)
{
Files.filename_inc(ref data_filename);
command.Add(" '" + data_filename + "' using 1:2, \\");
}
Files.filename_inc(ref data_filename);
command.Add(" '" + data_filename + "' using 1:2");
command.Add("quit");
File.WriteAllLines(command_filename, command);
Console.WriteLine(" STAB_ASYMPTOTIC plot stored in \"" + command_filename + "\".");
}
