ROOT::Math::MinimizerOptions Class Reference

Minimizer options.

Class defining the options for the minimizer. It contains also static methods for setting the default Minimizer option values that will be used by default by all Minimizer instances. To see the current default options do:

 ROOT::Math::MinimizerOptions::PrintDefault();

Definition at line 40 of file MinimizerOptions.h.

Public Member Functions
	MinimizerOptions ()
	MinimizerOptions (const MinimizerOptions &opt)
	~MinimizerOptions ()
double	ErrorDef () const
	error definition
const IOptions *	ExtraOptions () const
	return extra options (NULL pointer if they are not present)
unsigned int	MaxFunctionCalls () const
	max number of function calls
unsigned int	MaxIterations () const
	max iterations
const std::string &	MinimizerAlgorithm () const
	type of algorithm
const std::string &	MinimizerType () const
	type of minimizer
MinimizerOptions &	operator= (const MinimizerOptions &opt)
	assignment operators
double	Precision () const
	precision in the objective function calculation (value <=0 means left to default)
void	Print (std::ostream &os=std::cout) const
	print all the options
int	PrintLevel () const
	non-static methods for retrieving options
void	ResetToDefaultOptions ()
	non-static methods for setting options
void	SetErrorDef (double err)
	set error def
void	SetExtraOptions (const IOptions &opt)
	set extra options (in this case pointer is cloned)
void	SetMaxFunctionCalls (unsigned int maxfcn)
	set maximum of function calls
void	SetMaxIterations (unsigned int maxiter)
	set maximum iterations (one iteration can have many function calls)
void	SetMinimizerAlgorithm (const char *type)
	set minimizer algorithm
void	SetMinimizerType (const char *type)
	set minimizer type
void	SetPrecision (double prec)
	set the precision
void	SetPrintLevel (int level)
	set print level
void	SetStrategy (int stra)
	set the strategy
void	SetTolerance (double tol)
	set the tolerance
int	Strategy () const
	strategy
double	Tolerance () const
	absolute tolerance

Static Public Member Functions
static ROOT::Math::IOptions &	Default (const char *name)
	Retrieve extra options for a given minimizer name.
static double	DefaultErrorDef ()
static IOptions *	DefaultExtraOptions ()
static int	DefaultMaxFunctionCalls ()
static int	DefaultMaxIterations ()
static const std::string &	DefaultMinimizerAlgo ()
static const std::string &	DefaultMinimizerType ()
static double	DefaultPrecision ()
static int	DefaultPrintLevel ()
static int	DefaultStrategy ()
static double	DefaultTolerance ()
static ROOT::Math::IOptions *	FindDefault (const char *name)
	Find an extra options and return a nullptr if it is not existing.
static void	PrintDefault (const char *name=nullptr, std::ostream &os=std::cout)
	Print all the default options including the extra one specific for a given minimizer name.
static void	SetDefaultErrorDef (double up)
	Set the default level for computing the parameter errors.
static void	SetDefaultExtraOptions (const IOptions *extraoptions)
	Set additional minimizer options as pair of (string,value).
static void	SetDefaultMaxFunctionCalls (int maxcall)
	Set the maximum number of function calls.
static void	SetDefaultMaxIterations (int maxiter)
	Set the maximum number of iterations.
static void	SetDefaultMinimizer (const char *type, const char *algo=nullptr)
	Set the default Minimizer type and corresponding algorithms.
static void	SetDefaultPrecision (double prec)
	Set the default Minimizer precision.
static void	SetDefaultPrintLevel (int level)
	Set the default Print Level.
static void	SetDefaultStrategy (int strat)
	Set the default strategy.
static void	SetDefaultTolerance (double tol)
	Set the Minimization tolerance.

Private Attributes
std::string	fAlgoType
	Minimizer algorithmic specification (Migrad, Minimize, ...)
double	fErrorDef
	error definition (=1. for getting 1 sigma error for chi2 fits)
ROOT::Math::IOptions *	fExtraOptions
int	fLevel
	debug print level
int	fMaxCalls
	maximum number of function calls
int	fMaxIter
	maximum number of iterations
std::string	fMinimType
	Minimizer type (Minuit, Minuit2, etc..
double	fPrecision
	precision of the objective function evaluation (value <=0 means left to default)
int	fStrategy
	minimizer strategy (used by Minuit)
double	fTolerance
	minimize tolerance to reach solution

#include <Math/MinimizerOptions.h>

Constructor & Destructor Documentation

MinimizerOptions() [1/2]

ROOT::Math::MinimizerOptions::MinimizerOptions

(

)

Definition at line 133 of file MinimizerOptions.cxx.

~MinimizerOptions()

ROOT::Math::MinimizerOptions::~MinimizerOptions

(

)

Definition at line 166 of file MinimizerOptions.cxx.

MinimizerOptions() [2/2]

ROOT::Math::MinimizerOptions::MinimizerOptions

(

const MinimizerOptions &

opt

)

Definition at line 142 of file MinimizerOptions.cxx.

Member Function Documentation

Default()

IOptions & ROOT::Math::MinimizerOptions::Default ( const char *  name )

static

Retrieve extra options for a given minimizer name.

If the extra options already exist in a global map of (Minimizer name, options) it will return a reference to that options, otherwise it will create a new one and return the corresponding reference

Definition at line 230 of file MinimizerOptions.cxx.

DefaultErrorDef()

double ROOT::Math::MinimizerOptions::DefaultErrorDef ( )

static

Definition at line 91 of file MinimizerOptions.cxx.

DefaultExtraOptions()

IOptions * ROOT::Math::MinimizerOptions::DefaultExtraOptions ( )

static

Definition at line 98 of file MinimizerOptions.cxx.

DefaultMaxFunctionCalls()

int ROOT::Math::MinimizerOptions::DefaultMaxFunctionCalls ( )

static

Definition at line 94 of file MinimizerOptions.cxx.

DefaultMaxIterations()

int ROOT::Math::MinimizerOptions::DefaultMaxIterations ( )

static

Definition at line 95 of file MinimizerOptions.cxx.

DefaultMinimizerAlgo()

const std::string & ROOT::Math::MinimizerOptions::DefaultMinimizerAlgo ( )

static

Definition at line 85 of file MinimizerOptions.cxx.

DefaultMinimizerType()

const std::string & ROOT::Math::MinimizerOptions::DefaultMinimizerType ( )

static

Definition at line 100 of file MinimizerOptions.cxx.

DefaultPrecision()

double ROOT::Math::MinimizerOptions::DefaultPrecision ( )

static

Definition at line 93 of file MinimizerOptions.cxx.

DefaultPrintLevel()

int ROOT::Math::MinimizerOptions::DefaultPrintLevel ( )

static

Definition at line 97 of file MinimizerOptions.cxx.

DefaultStrategy()

int ROOT::Math::MinimizerOptions::DefaultStrategy ( )

static

Definition at line 96 of file MinimizerOptions.cxx.

DefaultTolerance()

double ROOT::Math::MinimizerOptions::DefaultTolerance ( )

static

Definition at line 92 of file MinimizerOptions.cxx.

ErrorDef()

double ROOT::Math::MinimizerOptions::ErrorDef ( ) const

inline

error definition

Definition at line 192 of file MinimizerOptions.h.

ExtraOptions()

const IOptions * ROOT::Math::MinimizerOptions::ExtraOptions ( ) const

inline

return extra options (NULL pointer if they are not present)

Definition at line 195 of file MinimizerOptions.h.

FindDefault()

IOptions * ROOT::Math::MinimizerOptions::FindDefault ( const char *  name )

static

Find an extra options and return a nullptr if it is not existing.

Same as above but it will not create a new one

Definition at line 235 of file MinimizerOptions.cxx.

MaxFunctionCalls()

unsigned int ROOT::Math::MinimizerOptions::MaxFunctionCalls ( ) const

inline

max number of function calls

Definition at line 177 of file MinimizerOptions.h.

MaxIterations()

unsigned int ROOT::Math::MinimizerOptions::MaxIterations ( ) const

inline

max iterations

Definition at line 180 of file MinimizerOptions.h.

MinimizerAlgorithm()

const std::string & ROOT::Math::MinimizerOptions::MinimizerAlgorithm ( ) const

inline

type of algorithm

Definition at line 201 of file MinimizerOptions.h.

MinimizerType()

const std::string & ROOT::Math::MinimizerOptions::MinimizerType ( ) const

inline

type of minimizer

Definition at line 198 of file MinimizerOptions.h.

operator=()

MinimizerOptions & ROOT::Math::MinimizerOptions::operator=

(

const MinimizerOptions &

opt

)

assignment operators

Definition at line 147 of file MinimizerOptions.cxx.

Precision()

double ROOT::Math::MinimizerOptions::Precision ( ) const

inline

precision in the objective function calculation (value <=0 means left to default)

Definition at line 189 of file MinimizerOptions.h.

Print()

void ROOT::Math::MinimizerOptions::Print

(

std::ostream &

os = std::cout

)

const

print all the options

Definition at line 212 of file MinimizerOptions.cxx.

PrintDefault()

void ROOT::Math::MinimizerOptions::PrintDefault ( const char *  name = nullptr, std::ostream &  os = std::cout  )

static

Print all the default options including the extra one specific for a given minimizer name.

If no minimizer name is given, all the extra default options, which have been set and configured will be printed

Definition at line 240 of file MinimizerOptions.cxx.

PrintLevel()

int ROOT::Math::MinimizerOptions::PrintLevel ( ) const

inline

non-static methods for retrieving options

set print level

Definition at line 174 of file MinimizerOptions.h.

ResetToDefaultOptions()

void ROOT::Math::MinimizerOptions::ResetToDefaultOptions

(

)

non-static methods for setting options

Definition at line 170 of file MinimizerOptions.cxx.

SetDefaultErrorDef()

void ROOT::Math::MinimizerOptions::SetDefaultErrorDef ( double  up )

static

Set the default level for computing the parameter errors.

For example for 1-sigma parameter errors

• up = 1 for a chi-squared function
• up = 0.5 for a negative log-likelihood function

The value will be used also by Minos when computing the confidence interval

Definition at line 51 of file MinimizerOptions.cxx.

SetDefaultExtraOptions()

void ROOT::Math::MinimizerOptions::SetDefaultExtraOptions ( const IOptions *  extraoptions )

static

Set additional minimizer options as pair of (string,value).

Extra option defaults can be configured for a specific algorithm and then if a matching with the correct option name exists it will be used whenever creating a new minimizer instance. For example for changing the default number of steps of the Genetic minimizer from 100 to 500 do

 auto extraOpt = ROOT::Math::MinimizerOptions::Default("Genetic")
 extraOpts.SetValue("Steps",500);

and when creating the Genetic minimizer you will have the new value for the option:

auto gmin = ROOT::Math::Factory::CreateMinimizer("Genetic");
gmin->Options().Print();

Definition at line 79 of file MinimizerOptions.cxx.

SetDefaultMaxFunctionCalls()

void ROOT::Math::MinimizerOptions::SetDefaultMaxFunctionCalls ( int  maxcall )

static

Set the maximum number of function calls.

Definition at line 63 of file MinimizerOptions.cxx.

SetDefaultMaxIterations()

void ROOT::Math::MinimizerOptions::SetDefaultMaxIterations ( int  maxiter )

static

Set the maximum number of iterations.

Used by the GSL minimizers and Genetic. Not used by Minuit,Minuit2.

Definition at line 67 of file MinimizerOptions.cxx.

SetDefaultMinimizer()

void ROOT::Math::MinimizerOptions::SetDefaultMinimizer ( const char *  type, const char *  algo = nullptr  )

static

Set the default Minimizer type and corresponding algorithms.

Here is the list of the available minimizers and their corresponding algorithms. For some minimizers (e.g. Fumili) there are no specific algorithms available, then there is no need to specify it.

ROOT Minimizers

• Minuit Minimizer based on TMinuit, the legacy Minuit implementation. Here are the available algorithms:
  • Migrad default algorithm based on the variable metric minimizer
  • Minimize combination of Simplex and Migrad
  • Simplex minimization algorithm not using the gradient information
  • Scan brute function scan
• Minuit2 New C++ implementation of Minuit (the recommended one)
  • Migrad (default)
  • Minimize
  • Simplex
  • Fumili2 new implementation of Fumili integrated in Minuit2
• Fumili Minimizer using an approximation for the Hessian based on first derivatives of the model function (see TFumili). Works only for chi-squared and likelihood functions.
• Linear Linear minimizer (fitter) working only for linear functions (see TLinearFitter and TLinearMinimizer)
• GSLMultiMin Minimizer from GSL based on the ROOT::Math::GSLMinimizer. Available algorithms are:
  • BFGS2 (default)
  • BFGS
  • ConjugateFR
  • ConjugatePR
  • SteepestDescent
• GSLMultiFit Minimizer based on GSL for minimizing only non linear least-squared functions (using an approximation similar to Fumili). See ROOT::Math::GSLMultiFit.
• GSLSimAn Simulated annealing minimizer from GSL (see ROOT::Math::GSLSimAnMinimizer). It is a stochastic minimization algorithm using only function values and not the gradient.
• Genetic Genetic minimization algorithms (see TMVA::Genetic)

Definition at line 43 of file MinimizerOptions.cxx.

SetDefaultPrecision()

void ROOT::Math::MinimizerOptions::SetDefaultPrecision ( double  prec )

static

Set the default Minimizer precision.

(used only by MInuit and Minuit2) It is used to specify the numerical precision used for computing the objective function. It should be left to the default value found by the Minimizer (typically double precision)

Definition at line 59 of file MinimizerOptions.cxx.

SetDefaultPrintLevel()

void ROOT::Math::MinimizerOptions::SetDefaultPrintLevel ( int  level )

static

Set the default Print Level.

Possible levels are from 0 (minimal printing) to 3 (maximum printing)

Definition at line 75 of file MinimizerOptions.cxx.

SetDefaultStrategy()

void ROOT::Math::MinimizerOptions::SetDefaultStrategy ( int  strat )

static

Set the default strategy.

The strategy is a parameter used only by Minuit and Minuit2. Possible values are:

• strat = 0 : rough approximation of Hessian using the gradient. Avoid computing the full Hessian matrix
• strat = 1 (default and recommended one) - Use Hessian approximation but compute full Hessian at the end of minimization if needed.
• strat = 2 Perform several full Hessian computations during the minimization. Slower and not always working better than strat=1.

Definition at line 71 of file MinimizerOptions.cxx.

SetDefaultTolerance()

void ROOT::Math::MinimizerOptions::SetDefaultTolerance ( double  tol )

static

Set the Minimization tolerance.

The Default value for Minuit and Minuit2 is 0.01

Definition at line 55 of file MinimizerOptions.cxx.

SetErrorDef()

void ROOT::Math::MinimizerOptions::SetErrorDef ( double  err )

inline

set error def

Definition at line 228 of file MinimizerOptions.h.

SetExtraOptions()

void ROOT::Math::MinimizerOptions::SetExtraOptions

(

const IOptions &

opt

)

set extra options (in this case pointer is cloned)

Definition at line 206 of file MinimizerOptions.cxx.

SetMaxFunctionCalls()

void ROOT::Math::MinimizerOptions::SetMaxFunctionCalls ( unsigned int  maxfcn )

inline

set maximum of function calls

Definition at line 213 of file MinimizerOptions.h.

SetMaxIterations()

void ROOT::Math::MinimizerOptions::SetMaxIterations ( unsigned int  maxiter )

inline

set maximum iterations (one iteration can have many function calls)

Definition at line 216 of file MinimizerOptions.h.

SetMinimizerAlgorithm()

void ROOT::Math::MinimizerOptions::SetMinimizerAlgorithm ( const char *  type )

inline

set minimizer algorithm

Definition at line 234 of file MinimizerOptions.h.

SetMinimizerType()

void ROOT::Math::MinimizerOptions::SetMinimizerType ( const char *  type )

inline

set minimizer type

Definition at line 231 of file MinimizerOptions.h.

SetPrecision()

void ROOT::Math::MinimizerOptions::SetPrecision ( double  prec )

inline

set the precision

Definition at line 222 of file MinimizerOptions.h.

SetPrintLevel()

void ROOT::Math::MinimizerOptions::SetPrintLevel ( int  level )

inline

set print level

Definition at line 210 of file MinimizerOptions.h.

SetStrategy()

void ROOT::Math::MinimizerOptions::SetStrategy ( int  stra )

inline

set the strategy

Definition at line 225 of file MinimizerOptions.h.

SetTolerance()

void ROOT::Math::MinimizerOptions::SetTolerance ( double  tol )

inline

set the tolerance

Definition at line 219 of file MinimizerOptions.h.

Strategy()

int ROOT::Math::MinimizerOptions::Strategy ( ) const

inline

strategy

Definition at line 183 of file MinimizerOptions.h.

Tolerance()

double ROOT::Math::MinimizerOptions::Tolerance ( ) const

inline

absolute tolerance

Definition at line 186 of file MinimizerOptions.h.

Member Data Documentation

fAlgoType

std::string ROOT::Math::MinimizerOptions::fAlgoType

private

Minimizer algorithmic specification (Migrad, Minimize, ...)

Definition at line 250 of file MinimizerOptions.h.

fErrorDef

double ROOT::Math::MinimizerOptions::fErrorDef

private

error definition (=1. for getting 1 sigma error for chi2 fits)

Definition at line 246 of file MinimizerOptions.h.

fExtraOptions

ROOT::Math::IOptions* ROOT::Math::MinimizerOptions::fExtraOptions

private

Definition at line 253 of file MinimizerOptions.h.

fLevel

int ROOT::Math::MinimizerOptions::fLevel

private

debug print level

Definition at line 242 of file MinimizerOptions.h.

fMaxCalls

int ROOT::Math::MinimizerOptions::fMaxCalls

private

maximum number of function calls

Definition at line 243 of file MinimizerOptions.h.

fMaxIter

int ROOT::Math::MinimizerOptions::fMaxIter

private

maximum number of iterations

Definition at line 244 of file MinimizerOptions.h.

fMinimType

std::string ROOT::Math::MinimizerOptions::fMinimType

private

Minimizer type (Minuit, Minuit2, etc..

Definition at line 249 of file MinimizerOptions.h.

fPrecision

double ROOT::Math::MinimizerOptions::fPrecision

private

precision of the objective function evaluation (value <=0 means left to default)

Definition at line 248 of file MinimizerOptions.h.

fStrategy

int ROOT::Math::MinimizerOptions::fStrategy

private

minimizer strategy (used by Minuit)

Definition at line 245 of file MinimizerOptions.h.

fTolerance

double ROOT::Math::MinimizerOptions::fTolerance

private

minimize tolerance to reach solution

Definition at line 247 of file MinimizerOptions.h.

Libraries for ROOT::Math::MinimizerOptions:

---

The documentation for this class was generated from the following files:

• math/mathcore/inc/Math/MinimizerOptions.h
• math/mathcore/src/MinimizerOptions.cxx