public double integratedCovariance(int i, int j, double t, Vector x)
{
if (corrModel_.isTimeIndependent())
{
try {
// if all objects support these methods
// thats by far the fastest way to get the
// integrated covariance
return(corrModel_.correlation(i, j, 0.0, x)
* volaModel_.integratedVariance(j, i, t, x));
//verifier la methode integratedVariance, qui bdoit etre implémenté
}
catch (System.Exception) {
// okay proceed with the
// slow numerical integration routine
}
}
try {
Utils.QL_REQUIRE(x.empty() != false, () => "can not handle given x here");
}
catch { //OK x empty
}
double tmp = 0.0;
VarProxy_Helper helper = new VarProxy_Helper(this, i, j);
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(1e-10, 10000);
for (int k = 0; k < 64; ++k)
{
tmp += integrator.value(helper.value, k * t / 64.0, (k + 1) * t / 64.0);
}
return(tmp);
}
public virtual Matrix integratedCovariance(double t, Vector x)
{
// this implementation is not intended for production.
// because it is too slow and too inefficient.
// This method is useful for testing and R&D.
// Please overload the method within derived classes.
//QL_REQUIRE(x.empty(), "can not handle given x here");
try {
if (!(x.empty()))
{
throw new ApplicationException("can not handle given x here");
}
}
catch { } //x is empty or null
Matrix tmp = new Matrix(size_, size_, 0.0);
for (int i = 0; i < size_; ++i)
{
for (int j = 0; j <= i; ++j)
{
Var_Helper helper = new Var_Helper(this, i, j);
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(1e-10, 10000);
for (int k = 0; k < 64; ++k)
{
tmp[i, j] += integrator.value(helper.value, k * t / 64.0, (k + 1) * t / 64.0);
}
tmp[j, i] = tmp[i, j];
}
}
return(tmp);
}
public double integrate(double a, double b, ConundrumIntegrand integrand)
{
double result = .0;
//double abserr =.0;
//double alpha = 1.0;
//double epsabs = precision_;
//double epsrel = 1.0; // we are interested only in absolute precision
//size_t neval =0;
// we use the non adaptive algorithm only for semi infinite interval
if (a > 0)
{
// we estimate the actual boudary by testing integrand values
double upperBoundary = 2 * a;
while (integrand.value(upperBoundary) > precision_)
{
upperBoundary *= 2.0;
}
// sometimes b < a because of a wrong estimation of b based on stdev
if (b > a)
{
upperBoundary = Math.Min(upperBoundary, b);
}
GaussKronrodNonAdaptive gaussKronrodNonAdaptive = new GaussKronrodNonAdaptive(precision_, 1000000, 1.0);
// if the integration intervall is wide enough we use the
// following change variable x -> a + (b-a)*(t/(a-b))^3
upperBoundary = Math.Max(a, Math.Min(upperBoundary, hardUpperLimit_));
if (upperBoundary > 2 * a)
{
VariableChange variableChange = new VariableChange(integrand.value, a, upperBoundary, 3);
result = gaussKronrodNonAdaptive.value(variableChange.value, .0, 1.0);
}
else
{
result = gaussKronrodNonAdaptive.value(integrand.value, a, upperBoundary);
}
// if the expected precision has not been reached we use the old algorithm
if (!gaussKronrodNonAdaptive.integrationSuccess())
{
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(precision_, 100000);
b = Math.Max(a, Math.Min(b, hardUpperLimit_));
result = integrator.value(integrand.value, a, b);
}
} // if a < b we use the old algorithm
else
{
b = Math.Max(a, Math.Min(b, hardUpperLimit_));
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(precision_, 100000);
result = integrator.value(integrand.value, a, b);
}
return(result);
}
public virtual Matrix integratedCovariance(double t, Vector x = null)
{
// this implementation is not intended for production.
// because it is too slow and too inefficient.
// This method is useful for testing and R&D.
// Please overload the method within derived classes.
Utils.QL_REQUIRE(x == null ,()=> "can not handle given x here");
Matrix tmp= new Matrix(size_, size_,0.0);
for (int i=0; i<size_; ++i) {
for (int j=0; j<=i;++j) {
Var_Helper helper = new Var_Helper(this, i, j);
GaussKronrodAdaptive integrator=new GaussKronrodAdaptive(1e-10, 10000);
for (int k=0; k < 64; ++k) {
tmp[i,j] +=integrator.value(helper.value, k * t / 64.0, (k + 1) * t / 64.0);
}
tmp[j,i]=tmp[i,j];
}
}
return tmp;
}
public double integrate(double a, double b, ConundrumIntegrand integrand) {
double result = .0;
//double abserr =.0;
//double alpha = 1.0;
//double epsabs = precision_;
//double epsrel = 1.0; // we are interested only in absolute precision
//size_t neval =0;
// we use the non adaptive algorithm only for semi infinite interval
if (a > 0) {
// we estimate the actual boudary by testing integrand values
double upperBoundary = 2 * a;
while (integrand.value(upperBoundary) > precision_)
upperBoundary *= 2.0;
// sometimes b < a because of a wrong estimation of b based on stdev
if (b > a)
upperBoundary = Math.Min(upperBoundary, b);
GaussKronrodNonAdaptive gaussKronrodNonAdaptive = new GaussKronrodNonAdaptive(precision_, 1000000, 1.0);
// if the integration intervall is wide enough we use the
// following change variable x -> a + (b-a)*(t/(a-b))^3
if (upperBoundary > 2 * a) {
VariableChange variableChange = new VariableChange(integrand.value, a, upperBoundary, 3);
result = gaussKronrodNonAdaptive.value(variableChange.value, .0, 1.0);
} else {
result = gaussKronrodNonAdaptive.value(integrand.value, a, upperBoundary);
}
// if the expected precision has not been reached we use the old algorithm
if (!gaussKronrodNonAdaptive.integrationSuccess()) {
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(precision_, 1000000);
result = integrator.value(integrand.value, a, b);
}
} // if a < b we use the old algorithm
else {
GaussKronrodAdaptive integrator = new GaussKronrodAdaptive(precision_, 1000000);
result = integrator.value(integrand.value, a, b);
}
return result;
}
public double integratedCovariance(int i, int j, double t,Vector x)
{
if (corrModel_.isTimeIndependent()) {
try {
// if all objects support these methods
// thats by far the fastest way to get the
// integrated covariance
return corrModel_.correlation(i, j, 0.0, x)
*volaModel_.integratedVariance(j, i, t, x);
//verifier la methode integratedVariance, qui bdoit etre implémenté
}
catch (System.Exception) {
// okay proceed with the
// slow numerical integration routine
}
}
//QL_REQUIRE(x.empty(), "can not handle given x here");
try {
if (x.empty() == false)
throw new ApplicationException("can not handle given x here");
}
catch { //OK x empty
}
double tmp=0.0;
VarProxy_Helper helper = new VarProxy_Helper(this, i, j);
GaussKronrodAdaptive integrator=new GaussKronrodAdaptive(1e-10, 10000);
for (int k=0; k<64; ++k) {
tmp+=integrator.value(helper.value, k*t/64.0, (k+1)*t/64.0);
}
return tmp;
}
