Dakota
Version 6.20
Explore and Predict with Confidence
|
Dakota transition kernel that updates proposal covariance based on derivatives (for random walk case) More...
Public Member Functions | |
DerivInformedPropCovTK (const char *prefix, const QUESO::VectorSpace< V, M > &vectorSpace, const std::vector< double > &scales, const M &covMatrix, NonDQUESOBayesCalibration *queso_instance) | |
Constructor for derivative-informed proposal covariance. | |
virtual | ~DerivInformedPropCovTK () |
Destructor for derivative-informed proposal covariance. | |
virtual void | updateTK () override |
update the transition kernel with new covariance information | |
virtual bool | covMatrixIsDirty () override |
whether the covariance matrix has been updated | |
virtual void | cleanCovMatrix () override |
dependent algorithms have taken necessary cleanup actions | |
Private Attributes | |
const QUESO::VectorSpace< V, M > & | m_vectorSpace |
calibration parameter vector space (note: hides base class member) | |
bool | covIsDirty |
whether we've updated the proposal covariance | |
unsigned int | chainIndex |
index into current chain position | |
NonDQUESOBayesCalibration * | nonDQUESOInstance |
Dakota QUESO instance for callbacks. | |
Dakota transition kernel that updates proposal covariance based on derivatives (for random walk case)