RooPolynomial.cxx

Go to the documentation of this file.

/*****************************************************************************
 * Project: RooFit                                                           *
 * Package: RooFitModels                                                     *
 * @(#)root/roofit:$Id$
 * Authors:                                                                  *
 *   WV, Wouter Verkerke, UC Santa Barbara, verkerke@slac.stanford.edu       *
 *   DK, David Kirkby,    UC Irvine,         dkirkby@uci.edu                 *
 *                                                                           *
 * Copyright (c) 2000-2005, Regents of the University of California          *
 *                          and Stanford University. All rights reserved.    *
 *                                                                           *
 * Redistribution and use in source and binary forms,                        *
 * with or without modification, are permitted according to the terms        *
 * listed in LICENSE (http://roofit.sourceforge.net/license.txt)             *
 *****************************************************************************/
 
/** \class RooPolynomial
    \ingroup Roofit
 
RooPolynomial implements a polynomial p.d.f of the form
\f[ f(x) = \mathcal{N} \cdot \sum_{i} a_{i} * x^i \f]
By default, the coefficient \f$ a_0 \f$ is chosen to be 1, as polynomial
probability density functions have one degree of freedom
less than polynomial functions due to the normalisation condition. \f$ \mathcal{N} \f$
is a normalisation constant that is automatically calculated when the polynomial is used
in computations.
 
The sum can be truncated at the low end. See the main constructor
RooPolynomial::RooPolynomial(const char*, const char*, RooAbsReal&, const RooArgList&, Int_t)
**/
 
#include "RooPolynomial.h"
#include "RooArgList.h"
#include "RooMsgService.h"
#include "RooBatchCompute.h"
 
#include "TError.h"
#include <vector>
using namespace std;
 
ClassImp(RooPolynomial);
 
////////////////////////////////////////////////////////////////////////////////
/// coverity[UNINIT_CTOR]
 
RooPolynomial::RooPolynomial()
{
}
 
////////////////////////////////////////////////////////////////////////////////
/// Create a polynomial in the variable `x`.
/// \param[in] name Name of the PDF
/// \param[in] title Title for plotting the PDF
/// \param[in] x The variable of the polynomial
/// \param[in] coefList The coefficients \f$ a_i \f$
/// \param[in] lowestOrder [optional] Truncate the sum such that it skips the lower orders:
/// \f[
///     1. + \sum_{i=0}^{\mathrm{coefList.size()}} a_{i} * x^{(i + \mathrm{lowestOrder})}
/// \f]
///
/// This means that
/// \code{.cpp}
/// RooPolynomial pol("pol", "pol", x, RooArgList(a, b), lowestOrder = 2)
/// \endcode
/// computes
/// \f[
///   \mathrm{pol}(x) = 1 * x^0 + (0 * x^{\ldots}) + a * x^2 + b * x^3.
/// \f]
 
 
RooPolynomial::RooPolynomial(const char* name, const char* title,
              RooAbsReal& x, const RooArgList& coefList, Int_t lowestOrder) :
  RooAbsPdf(name, title),
  _x("x", "Dependent", this, x),
  _coefList("coefList","List of coefficients",this),
  _lowestOrder(lowestOrder)
{
  // Check lowest order
  if (_lowestOrder<0) {
    coutE(InputArguments) << "RooPolynomial::ctor(" << GetName()
           << ") WARNING: lowestOrder must be >=0, setting value to 0" << endl ;
    _lowestOrder=0 ;
  }
 
  for (auto *coef : coefList) {
    if (!dynamic_cast<RooAbsReal*>(coef)) {
      coutE(InputArguments) << "RooPolynomial::ctor(" << GetName() << ") ERROR: coefficient " << coef->GetName()
             << " is not of type RooAbsReal" << endl ;
      R__ASSERT(0) ;
    }
    _coefList.add(*coef) ;
  }
}
 
////////////////////////////////////////////////////////////////////////////////
 
RooPolynomial::RooPolynomial(const char* name, const char* title,
                           RooAbsReal& x) :
  RooAbsPdf(name, title),
  _x("x", "Dependent", this, x),
  _coefList("coefList","List of coefficients",this),
  _lowestOrder(1)
{ }
 
////////////////////////////////////////////////////////////////////////////////
/// Copy constructor
 
RooPolynomial::RooPolynomial(const RooPolynomial& other, const char* name) :
  RooAbsPdf(other, name),
  _x("x", this, other._x),
  _coefList("coefList",this,other._coefList),
  _lowestOrder(other._lowestOrder)
{ }
 
////////////////////////////////////////////////////////////////////////////////
/// Destructor
 
RooPolynomial::~RooPolynomial()
{ }
 
////////////////////////////////////////////////////////////////////////////////
 
double RooPolynomial::evaluate() const
{
  // Calculate and return value of polynomial
 
  const unsigned sz = _coefList.getSize();
  const int lowestOrder = _lowestOrder;
  if (!sz) return lowestOrder ? 1. : 0.;
  _wksp.clear();
  _wksp.reserve(sz);
  {
    const RooArgSet* nset = _coefList.nset();
    for (auto *c : static_range_cast<RooAbsReal *>(_coefList)) {
       _wksp.push_back(c->getVal(nset));
    }
  }
  const double x = _x;
  double retVal = _wksp[sz - 1];
  for (unsigned i = sz - 1; i--; ) retVal = _wksp[i] + x * retVal;
  return retVal * std::pow(x, lowestOrder) + (lowestOrder ? 1.0 : 0.0);
}
 
// The batch mode support for RooPolynomial was commented out, because that
// implementation can't deal with observables used as polynomial coefficients
// yet.
 
//////////////////////////////////////////////////////////////////////////////////
///// Compute multiple values of Polynomial.
//void RooPolynomial::computeBatch(cudaStream_t* stream, double* output, size_t nEvents, RooFit::DataMap& dataMap) const
//{
  //RooBatchCompute::ArgVector extraArgs;
  //for (auto* coef:_coefList)
    //extraArgs.push_back( static_cast<const RooAbsReal*>(coef)->getVal() );
  //extraArgs.push_back(_lowestOrder);
  //auto dispatch = stream ? RooBatchCompute::dispatchCUDA : RooBatchCompute::dispatchCPU;
  //dispatch->compute(stream, RooBatchCompute::Polynomial, output, nEvents, dataMap, {&*_x,&*_norm}, extraArgs);
//}
 
////////////////////////////////////////////////////////////////////////////////
/// Advertise to RooFit that this function can be analytically integrated.
Int_t RooPolynomial::getAnalyticalIntegral(RooArgSet& allVars, RooArgSet& analVars, const char* /*rangeName*/) const
{
  if (matchArgs(allVars, analVars, _x)) return 1;
  return 0;
}
 
////////////////////////////////////////////////////////////////////////////////
/// Do the analytical integral according to the code that was returned by getAnalyticalIntegral().
double RooPolynomial::analyticalIntegral(Int_t code, const char* rangeName) const
{
  R__ASSERT(code==1) ;
 
  const double xmin = _x.min(rangeName), xmax = _x.max(rangeName);
  const int lowestOrder = _lowestOrder;
  const unsigned sz = _coefList.getSize();
  if (!sz) return xmax - xmin;
  _wksp.clear();
  _wksp.reserve(sz);
  {
    const RooArgSet* nset = _coefList.nset();
    unsigned i = 1 + lowestOrder;
    for (auto *c : static_range_cast<RooAbsReal *>(_coefList)) {
      _wksp.push_back(c->getVal(nset) / double(i));
      ++i;
    }
  }
  double min = _wksp[sz - 1], max = _wksp[sz - 1];
  for (unsigned i = sz - 1; i--; )
    min = _wksp[i] + xmin * min, max = _wksp[i] + xmax * max;
  return max * std::pow(xmax, 1 + lowestOrder) - min * std::pow(xmin, 1 + lowestOrder) +
      (lowestOrder ? (xmax - xmin) : 0.);
}