hops::Rosenbrock Class Reference

multi-dimensional extension of rosenbrock function to N dimensions. Only defined on spaces of even N! More...

#include <Rosenbrock.hpp>

Inheritance diagram for hops::Rosenbrock:

Collaboration diagram for hops::Rosenbrock:

Public Member Functions
	Rosenbrock (double scaleParameter, VectorType shiftParameter)
MatrixType::Scalar	computeNegativeLogLikelihood (const VectorType &x) const override
	Evaluates the negative log likelihood for input x. More...
MatrixType	computeHessian (const VectorType &x) const
std::optional< VectorType >	computeLogLikelihoodGradient (const VectorType &x) const override
std::optional< MatrixType >	computeExpectedFisherInformation (const VectorType &x) const override
	Actually this computes the softmax of the hessian instead of the the expected fisher information is intractable for this model. More...
Public Member Functions inherited from hops::Model
virtual	~Model ()=default
virtual std::optional< std::vector< std::string > >	getParameterNames () const

Detailed Description

multi-dimensional extension of rosenbrock function to N dimensions. Only defined on spaces of even N!

Reference: https://doi.org/10.1162/evco.2009.17.3.437
Definition: \( f(x_1, x_2,..., x_N) = \sum_{i=1}^{N/2} [s_i \cdot (x^2_{2i-1}-x_{2i})^2 + (x_{2i-1} -a_i)^2] \)
where \( \boldsymbol{a} \) is the shiftParameter and \( \boldsymbol{s} \) is the scaleParameter in the Constructor. Both vectors have dimensions ( \( \frac{N}{2} \)).

Constructor & Destructor Documentation

Rosenbrock()

hops::Rosenbrock::Rosenbrock	(	double	scaleParameter,
		VectorType	shiftParameter
	)

@Brief shiftParameter has half the dimensions of the state vector

Parameters

scaleParameter
shiftParameter

Member Function Documentation

computeExpectedFisherInformation()

std::optional< MatrixType > hops::Rosenbrock::computeExpectedFisherInformation ( const VectorType &  x ) const

overridevirtual

Actually this computes the softmax of the hessian instead of the the expected fisher information is intractable for this model.

See 10.1007/978-3-642-40020-9_35

Returns

Reimplemented from hops::Model.

computeHessian()

MatrixType hops::Rosenbrock::computeHessian

(

const VectorType &

x

)

const

computeLogLikelihoodGradient()

std::optional< VectorType > hops::Rosenbrock::computeLogLikelihoodGradient ( const VectorType &  x ) const

overridevirtual

Reimplemented from hops::Model.

computeNegativeLogLikelihood()

MatrixType::Scalar hops::Rosenbrock::computeNegativeLogLikelihood ( const VectorType &  x ) const

overridevirtual

Evaluates the negative log likelihood for input x.

Parameters

x

Returns

Implements hops::Model.

---

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

• Rosenbrock.hpp