BayesOpt
bayesopt Namespace Reference

Namespace of the library interface. More...

Namespaces

 utils
 Extra utils: math functions, ublas helpers, etc.
 

Classes

class  AnnealedExpectedImprovement
 Expected improvement criterion using Schonlau annealing. [Schonlau98]. More...
 
class  AnnealedLowerConfindenceBound
 Lower (upper) confidence bound using Srinivas annealing [Srinivas10]. More...
 
class  ARDkernel
 Abstract class for anisotropic kernel functors using ARD (Automatic Relevance Determination) More...
 
class  AtomicFunction
 Abstract class for an atomic kernel. More...
 
class  AtomicKernel
 Abstract class for an atomic kernel. More...
 
class  BayesOptBase
 Abstract module for Bayesian optimization. More...
 
class  BiasedExpectedImprovement
 Expected improvement criterion modification by Lizotte. More...
 
class  BOptState
 Class that represents the state of an optimization. More...
 
class  CombinedCriteria
 Abstract class for combined criteria functions. More...
 
class  CombinedFunction
 Abstract class for combined functions. More...
 
class  CombinedKernel
 Abstract class for combined kernel. More...
 
class  ConditionalBayesProcess
 Empirical Bayesian NonParametric process. More...
 
class  ConstantFunction
 Constant function. More...
 
class  ConstKernel
 Constant kernel. More...
 
class  ContinuousModel
 Bayesian optimization for functions in continuous input spaces. More...
 
class  CritCallback
 
class  Criteria
 Abstract interface for criteria functors. More...
 
class  CriteriaFactory
 Factory model for criterion functions This factory is based on the libgp library by Manuel Blum https://bitbucket.org/mblum/libgp which follows the squeme of GPML by Rasmussen and Nickisch http://www.gaussianprocess.org/gpml/code/matlab/doc/. More...
 
class  Dataset
 Dataset model to deal with the vector (real) based datasets. More...
 
class  DiscreteModel
 Bayesian optimization for functions in discrete spaces. More...
 
class  EmpiricalBayes
 Bayesian optimization using different non-parametric processes as distributions over surrogate functions. More...
 
class  ExpectedImprovement
 Expected improvement criterion by Mockus [Mockus78]. More...
 
class  ExpectedReturn
 Expected return criterion. More...
 
class  GaussianDistribution
 
class  GaussianProcess
 Standard zero mean gaussian process with noisy observations. More...
 
class  GaussianProcessML
 Gaussian process with ML parameters. More...
 
class  GaussianProcessNormal
 Gaussian process with normal prior on the parameters. More...
 
class  GP_Hedge
 GP_Hedge model as describen in Hoffman et al. More...
 
class  GP_Hedge_Random
 Modification of the GP_Hedge algorithm where the bandit gains are random outcomes (like Thompson sampling). More...
 
class  GreedyAOptimality
 Greedy A-Optimality criterion. More...
 
class  HammingKernel
 Kernel for categorical data. More...
 
class  HierarchicalGaussianProcess
 Virtual class for hierarchical Gaussian processes. More...
 
class  InputDistance
 Distance in input space. More...
 
class  ISOkernel
 Abstract class for isotropic kernel functors. More...
 
class  Kernel
 Interface for kernel functors. More...
 
class  KernelFactory
 Factory model for kernel functions This factory is based on the libgp library by Manuel Blum https://bitbucket.org/mblum/libgp which follows the squeme of GPML by Rasmussen and Nickisch http://www.gaussianprocess.org/gpml/code/matlab/doc/. More...
 
class  KernelModel
 
class  KernelParameters
 
class  KernelProd
 Product of two kernels. More...
 
class  KernelRegressor
 Abstract class to implement non-parametric processes. More...
 
class  KernelSum
 Sum of two kernels. More...
 
class  LinearFunction
 Linear combination function. More...
 
class  LinearPlusConstantFunction
 Linear combination plus constant function. More...
 
class  LinKernel
 Linear kernel. More...
 
class  LinKernelARD
 Linear kernel. More...
 
class  LowerConfidenceBound
 Lower (upper) confidence bound criterion by [Cox and John, 1992]. More...
 
class  MaternARD1
 Matern ARD kernel of 1st order. More...
 
class  MaternARD3
 Matern ARD kernel of 3rd order. More...
 
class  MaternARD5
 Matern ARD kernel of 5th order. More...
 
class  MaternIso1
 Matern isotropic kernel of 1st order. More...
 
class  MaternIso3
 Matern kernel of 3rd order. More...
 
class  MaternIso5
 Matern isotropic kernel of 5th order. More...
 
class  MCMCModel
 Posterior model of nonparametric processes/criteria based on MCMC samples. More...
 
class  MCMCSampler
 Markov Chain Monte Carlo sampler. More...
 
class  MeanFactory
 Factory model for parametric functions This factory is based on the libgp library by Manuel Blum https://bitbucket.org/mblum/libgp which follows the squeme of GPML by Rasmussen and Nickisch http://www.gaussianprocess.org/gpml/code/matlab/doc/. More...
 
class  MeanModel
 
class  MeanParameters
 
class  MutualInformation
 Mutual Information bound criterion b [Contal et al., 2014]. More...
 
class  NLOPT_Optimization
 
class  NonParametricProcess
 Abstract class to implement Bayesian regressors. More...
 
class  OneFunction
 Constant one function. More...
 
class  OptimisticSampling
 Optimistic sampling. More...
 
class  Parameters
 
class  ParametricFunction
 Interface for mean functors. More...
 
class  Polynomial
 Polynomial covariance function. More...
 
class  Polynomial2
 
class  Polynomial3
 
class  Polynomial4
 
class  Polynomial5
 
class  Polynomial6
 
class  Polynomial7
 
class  PosteriorFixed
 Bayesian optimization using different non-parametric processes as distributions over surrogate functions. More...
 
class  PosteriorModel
 Bayesian optimization using different non-parametric processes as distributions over surrogate functions. More...
 
class  ProbabilityDistribution
 
class  ProbabilityOfImprovement
 Probability of improvement criterion based on (Kushner). More...
 
class  ProdCriteria
 Product of criterion functions. More...
 
class  RBOptimizable
 
class  RBOptimizableWrapper
 
class  RGBOptimizable
 
class  RGBOptimizableWrapper
 
class  RQIso
 Rational quadratic (Student's t) kernel. More...
 
class  SEArd
 Square exponential (Gaussian) kernel. More...
 
class  SEIso
 Square exponential (Gaussian) kernel. More...
 
class  StudentTDistribution
 
class  StudentTProcessJeffreys
 Student T process with Jeffreys prior. More...
 
class  StudentTProcessNIG
 Student's t process with Normal Inverse-Gamma hyperprior on mean and snigal variance parameters. More...
 
class  SumCriteria
 Wrapper class for linear combination of criterion functions. More...
 
class  SumFunction
 Sum of two kernels. More...
 
class  ThompsonSampling
 Thompson sampling. Picks a random sample of the surrogate model. More...
 
class  ZeroFunction
 Constant zero function. More...
 

Typedefs

typedef boost::numeric::ublas::vector< double > vectord
 
typedef boost::numeric::ublas::vector< int > vectori
 
typedef boost::numeric::ublas::matrix< double > matrixd
 
typedef std::vector< vectord > vecOfvec
 
typedef double(* eval_func) (unsigned int n, const double *x, double *gradient, void *func_data)
 

Enumerations

enum  innerOptAlgorithms { DIRECT, LBFGS, BOBYQA, COMBINED }
 
enum  McmcAlgorithms { SLICE_MCMC }
 

Functions

template<typename KernelType >
Kernelcreate_func ()
 
template<typename MeanType >
ParametricFunctioncreate_func ()
 
template<typename CriteriaType >
Criteriacreate_func ()
 
double softmax (double g, double eta)
 Softmax function. More...
 
void checkNLOPTerror (nlopt_result errortype)
 
double run_nlopt (nlopt::algorithm algo, eval_func fpointer, vectord &Xnext, int maxf, const std::vector< double > &vd, const std::vector< double > &vu, void *objPointer)
 

Variables

const size_t MAX_INNER_EVALUATIONS = 500
 Used per dimmension.
 

Detailed Description

Namespace of the library interface.

Enumeration Type Documentation

◆ innerOptAlgorithms

Enumerator
DIRECT 

Global optimization.

LBFGS 

Local, derivative based.

BOBYQA 

Local, derivative free.

COMBINED 

Global exploration, local refinement (hand tuned)

Definition at line 36 of file inneroptimization.hpp.

◆ McmcAlgorithms

Enumerator
SLICE_MCMC 

Slice sampling.

Definition at line 35 of file mcmc_sampler.hpp.

Function Documentation

◆ softmax()

double bayesopt::softmax ( double  g,
double  eta 
)
inline

Softmax function.

Parameters
ggain function
etasmoothness coefficient
Returns

Definition at line 38 of file criteria_hedge.cpp.