RNG¶

Defined in <xf_fintech/rng.hpp>¶

class MT19937
class MT2203
class MT19937IcnRng
class MT19937BoxMullerNomralRng
class MT2203IcnRng
class MultiVariateNormalRng

SobolRsg¶

Defined in <xf_fintech/sobol_rsg.hpp>¶

class SobolRsg
class SobolRsg1D

BrownianBridge¶

Defined in <xf_fintech/brownian_bridge.hpp>¶

class BrownianBridge

TrinomialTree¶

Defined in <xf_fintech/trinomial_tree.hpp>¶

class TrinomialTree

TreeLattice¶

Defined in <xf_fintech/tree_lattice.hpp>¶

class TreeLattice

1DMesher¶

Defined in <xf_fintech/fdmmesher.hpp>¶

class 1DMesher

OrnsteinUhlenbeckProcess¶

Defined in <xf_fintech/ornstein_uhlenbeck_process.hpp>¶

class OrnsteinUhlenbeckProcess

StochasticProcess1D¶

Defined in <xf_fintech/stochastic_process.hpp>¶

class StochasticProcess1D

HWModel¶

Defined in <xf_fintech/hw_model.hpp>¶

class HWModel

G2Model¶

Defined in <xf_fintech/g2_model.hpp>¶

class G2Model

ECIRModel¶

Defined in <xf_fintech/ecir_model.hpp>¶

class ECIRModel

CIRModel¶

Defined in <xf_fintech/cir_model.hpp>¶

class CIRModel

VModel¶

Defined in <xf_fintech/v_model.hpp>¶

class VModel

HestonModel¶

Defined in <xf_fintech/heston_model.hpp>¶

class HestonModel

BKModel¶

Defined in <xf_fintech/bk_model.hpp>¶

class BKModel

BSModel¶

Defined in <xf_fintech/bs_model.hpp>¶

class BSModel

svd¶

#include "xf_fintech/jacobi_svd.hpp"

template <
    typename dataType,
    int diagSize
    >
void svd (
    dataType dataA [diagSize][diagSize],
    dataType sigma2 [diagSize][diagSize],
    dataType dataU_out2 [diagSize][diagSize],
    dataType dataV_out2 [diagSize][diagSize]
    )

Jacobi Singular Value Decomposition (SVD).

Parameters:

dataType	data type.
diagSize	matrix size.
dataA	diagSize x diagSize matrix
sigma2	the decomposed diagonal matrix of dataA
dataU_out2	the left U matrix of dataA
dataV_out2	the right V matrix of dataA

mcSimulation¶

#include "xf_fintech/mc_simulation.hpp"

template <
    typename DT,
    typename RNG,
    typename PathGeneratorT,
    typename PathPricerT,
    typename RNGSeqT,
    int UN,
    int VariateNum,
    int SampNum
    >
DT mcSimulation (
    ap_uint <16> timeSteps,
    ap_uint <27> maxSamples,
    ap_uint <27> requiredSamples,
    DT requiredTolerance,
    PathGeneratorT pathGenInst [UN][1],
    PathPricerT pathPriInst [UN][1],
    RNGSeqT rngSeqInst [UN][1]
    )

Monte Carlo Framework implementation.

Parameters:

DT	supported data type including double and float data type, which decides the precision of result, default double-precision data type.
RNG	random number generator type.
PathGeneratorT	path generator type which simulates the dynamics of the asset price.
PathPricerT	path pricer type which calcualtes the option price based on asset price.
RNGSeqT	random number sequence generator type.
UN	number of Monte Carlo Module in parallel, which affects the latency and resources utilization.
VariateNum	number of variate.
SampNum	the total samples are divided into several steps, SampNum is the number for each step.
timeSteps	number of the steps for each path.
maxSamples	the maximum sample number. When reaching it, the simulation will stop.
requiredTolerance	the tolerance required. If requiredSamples is not set, when reaching the required tolerance, simulation will stop.
requiredSamples	the samples number required. When reaching the required number, simulation will stop.
pathGenInst	instance of path generator.
pathPriInst	instance of path pricer.
rngSeqInst	instance of random number sequence.

pentadiagCr¶

#include "xf_fintech/pentadiag_cr.hpp"

template <
    typename T,
    unsigned int P_SIZE,
    unsigned int logN
    >
void pentadiagCr (
    T a [P_SIZE],
    T b [P_SIZE],
    T c [P_SIZE],
    T d [P_SIZE],
    T e [P_SIZE],
    T v [P_SIZE],
    T u [P_SIZE]
    )

Solves for u in linear system Pu = r

It calls function pentadiag_step for each step until all diagonals instead of main are zeros (or very close to zero). Result U is made by dividing Each of elements of right hand vactor ( v ) by main diagonal ( c )

Structure of input matrix:

c d e 0 0 |

b c d e 0 |

a b c d e |

0 a b c d |

0 0 a b c |.

Parameters:

T	data type used in whole function (double by default)
P_SIZE	Size of the operating matrix
logN	Number of steps for algorithm
c	Main diagonal
b	First lower
a	Second lower
d	First upper
e	Second upper
v	Right hand side vector of length n
u	Vectors of unknows to solve for

trap_integrate¶

#include "xf_fintech/quadrature.hpp"

template <typename DT>
static int trap_integrate (
    DT a,
    DT b,
    DT tol,
    DT* res,
    XF_USER_DATA_TYPE* p
    )

integration function using the adaptive trapezoidal technique

Parameters:

DT	float or DT, determines the precision of the result
a	lower limit of integration
b	upper limit of integration
tol	required tolerance of the result
res	the result of the integration
p	pointer to structure containing parameters for integrating function

Returns:

1 - success, 0 - fail (due to stack limitation)

simp_integrate¶

#include "xf_fintech/quadrature.hpp"

template <typename DT>
static int simp_integrate (
    DT a,
    DT b,
    DT tol,
    DT* res,
    XF_USER_DATA_TYPE* p
    )

integration function using the adaptive simpson 1/3 technique

Parameters:

DT	float or DT, determines the precision of the result
a	lower limit of integration
b	upper limit of integration
tol	required tolerance of the result
res	the result of the integration
p	pointer to structure containing parameters for integrating function

Returns:

1 - success, 0 - fail (due to stack limitation)

romberg_integrate¶

#include "xf_fintech/quadrature.hpp"

template <typename DT>
static int romberg_integrate (
    DT a,
    DT b,
    DT tol,
    DT* res,
    XF_USER_DATA_TYPE* p
    )

integration function using the romberg technique Based on https://en.wikipedia.org/wiki/Romberg%27s_method

Parameters:

DT	float or DT, determines the precision of the result
a	lower limit of integration
b	upper limit of integration
tol	required tolerance of the result
res	the result of the integration
p	pointer to structure containing parameters for integrating function

Returns:

1 - success, 0 - fail (due to stack limitation)

boxMullerTransform¶

#include "xf_fintech/rng.hpp"

template <typename mType>
void boxMullerTransform (
    mType u1,
    mType u2,
    mType& z1,
    mType& z2
    )

Box-Muller transform from uniform random number to normal random number.

Parameters:

mType	data type
u1	first uniform random number input. Notice that it should not be zero.
u2	second uniform random number input
z1	first normal random number output
z2	second normal random number output

inverseCumulativeNormalPPND7¶

#include "xf_fintech/rng.hpp"

template <typename mType>
mType inverseCumulativeNormalPPND7 (mType input)

Inverse Cumulative transform from random number to normal random number.

Reference: Algorithm AS 241, The Percentage Points of the Normal Distribution by Michael J. Wichura.

Parameters:

mType	data type.
input	random number input.

Returns:

normal random number.

inverseCumulativeNormalAcklam¶

#include "xf_fintech/rng.hpp"

template <typename mType>
mType inverseCumulativeNormalAcklam (mType input)

Inverse CumulativeNormal using Acklam’s approximation to transform uniform random number to normal random number.

Reference: Acklam’s approximation: by Peter J. Acklam, University of Oslo, Statistics Division.

Parameters:

mType	data type.
input	input uniform random number

Returns:

normal random number

trsvCore¶

#include "xf_fintech/trsv.hpp"

template <
    class T,
    unsigned int N,
    unsigned int logN,
    unsigned int NCU
    >
void trsvCore (
    T inlow [N],
    T indiag [N],
    T inup [N],
    T inrhs [N]
    )

Tridiagonal linear solver It solves tridiagonal linear system of equations by eliminating upper and lower diagonals To get result (U) divide each element of inrhs by coresponding element of main diagonal indiag .

Parameters:

T	data type
N	matrix size
logN	log2(N)(TOREMOVE)
NCU	number of compute units
inlow	lower diagonal
indiag	diagonal
inup	upper diagonal
inrhs	right-hand side