Inherits TNamed.
|
| | Chebyshev3D () |
| |
| | Chebyshev3D (const Chebyshev3D &src) |
| |
| | Chebyshev3D (const char *inpFile) |
| |
| | Chebyshev3D (FILE *stream) |
| |
| | ~Chebyshev3D () override |
| |
| Chebyshev3D & | operator= (const Chebyshev3D &rhs) |
| |
| void | Eval (const Float_t *par, Float_t *res) |
| | Evaluates Chebyshev parameterization for 3d->DimOut function.
|
| |
| Float_t | Eval (const Float_t *par, int idim) |
| | Evaluates Chebyshev parameterization for idim-th output dimension of 3d->DimOut function.
|
| |
| void | Eval (const Double_t *par, Double_t *res) |
| | Evaluates Chebyshev parameterization for 3d->DimOut function.
|
| |
| Double_t | Eval (const Double_t *par, int idim) |
| | Evaluates Chebyshev parameterization for idim-th output dimension of 3d->DimOut function.
|
| |
| void | evaluateDerivative (int dimd, const Float_t *par, Float_t *res) |
| |
| void | evaluateDerivative2 (int dimd1, int dimd2, const Float_t *par, Float_t *res) |
| |
| Float_t | evaluateDerivative (int dimd, const Float_t *par, int idim) |
| |
| Float_t | evaluateDerivative2 (int dimd1, int dimd2, const Float_t *par, int idim) |
| |
| void | evaluateDerivative3D (const Float_t *par, Float_t dbdr[3][3]) |
| | Returns the gradient matrix.
|
| |
| void | evaluateDerivative3D2 (const Float_t *par, Float_t dbdrdr[3][3][3]) |
| | Returns the gradient matrix.
|
| |
| void | Print (const Option_t *opt="") const override |
| |
| Bool_t | isInside (const Float_t *par) const |
| | Checks if the point is inside of the fitted box.
|
| |
| Bool_t | isInside (const Double_t *par) const |
| | Checks if the point is inside of the fitted box.
|
| |
| Chebyshev3DCalc * | getChebyshevCalc (int i) const |
| |
| Float_t | getBoundMin (int i) const |
| |
| Float_t | getBoundMax (int i) const |
| |
| Float_t * | getBoundMin () const |
| |
| Float_t * | getBoundMax () const |
| |
| Float_t * | getBoundaryMappingScale () const |
| |
| Float_t * | getBoundaryMappingOffset () const |
| |
| Float_t | getPrecision () const |
| |
| void | shiftBound (int id, float dif) |
| |
| void | loadData (const char *inpFile) |
| |
| void | loadData (FILE *stream) |
| |
Chebyshev3D produces the interpolation of the user 3D->NDimOut arbitrary function supplied in "void (*fcn)(float* inp,float* out)" format either in a separate macro file or as a function pointer. Only coefficients needed to guarantee the requested precision are kept. The user-callable methods are: To create the interpolation use: Cheb3D(const char* funName, // name of the file with user function or Cheb3D(void (ptr)(float,float*) // pointer on the user function
- Parameters
-
| Int_t | DimOut dimensionality of the function's output |
| Float_t | *bmin lower 3D bounds of interpolation domain |
| Float_t | *bmax upper 3D bounds of interpolation domain |
| Int_t | *npoints number of points in each of 3 input dimension, defining the interpolation grid |
| Float_t | prec=1E-6); requested max.absolute difference between the interpolation and any point on grid To test obtained parameterization use the method TH1* TestRMS(int idim,int npoints = 1000,TH1* histo=0); it will compare the user output of the user function and interpolation for idim-th output dimension and fill the difference in the supplied histogram. If no histogram is supplied, it will be created. To save the interpolation data: saveData(const char* filename, Bool_t append ) write text file with data. If append is kTRUE and the output file already exists, data will be added in the end of the file. Alternatively, saveData(FILE* stream) will write the data to already existing stream. To read back already stored interpolation use either the constructor Chebyshev3D(const char* inpFile); or the default constructor Chebyshev3D() followed by Chebyshev3D::loadData(const char* inpFile); To compute the interpolation use Eval(float* par,float *res) method, with par being 3D vector of arguments (inside the validity region) and res is the array of DimOut elements for the output. If only one component (say, idim-th) of the output is needed, use faster Float_t Eval(Float_t *par,int idim) method void Print(option="") will print the name, the ranges of validity and the absolute precision of the parameterization. Option "l" will also print the information about the number of coefficients for each output dimension. NOTE: during the evaluation no check is done for parameter vector being outside the interpolation region. If there is such a risk, use Bool_t isInside(float *par) method. Chebyshev parameterization is not good for extrapolation! For the properties of Chebyshev parameterization see: H.Wind, CERN EP Internal Report, 81-12/Rev. |
Definition at line 64 of file Chebyshev3D.h.
1.9.8