Mapped XC Evaluators

Function evaluators operate on transformed descriptors and return both a value and its derivatives with respect to the descriptors. Mapped DFT kernels apply the stored feature transforms and energy baselines. The top-level mapped XC object combines all kernel components with their shared feature settings.

Original mapped interface

class ciderpress.dft.xc_evaluator.AntisymRBFEvaluator(kernel, X1ctrl, alpha)
class ciderpress.dft.xc_evaluator.FuncEvaluator
build(*args, **kwargs)
class ciderpress.dft.xc_evaluator.GlobalLinearEvaluator(consts)
class ciderpress.dft.xc_evaluator.KernelEvaluator(kernel, X1ctrl, alpha)
class ciderpress.dft.xc_evaluator.MappedDFTKernel(fevals, feature_list, mode, multiplicative_baseline, additive_baseline=None)

This class evaluates the XC term arising from a single DFTKernel object, using one or a list of FuncEvaluator objects.

property N1
classmethod from_dict(d)
to_dict()
class ciderpress.dft.xc_evaluator.MappedXC(mapped_kernels, settings, libxc_baseline=None)

Inference-time evaluator for a mapped CIDER functional.

A MappedXC bundles one or more mapped DFT kernels with the FeatureSettings that define their inputs. It is the object reconstructed from a serialized CIDER model file (for example the packaged CIDER23X exchange models) and evaluates the XC energy contribution and its functional derivatives from normalized features.

Parameters:

  • mapped_kernels (list of DFTKernel)

  • settings (FeatureSettings)

  • libxc_baseline (str)

property nfeat
property normalizer
class ciderpress.dft.xc_evaluator.ModelWithNormalizer(model, normalizer)

Initialize a ModelWithNormalizer.

Parameters:

  • model (MappedFunctional) – Model evaluator

  • normalizer (FeatNormalizerList) – Instructions for normalizing the features.

get_x0_usps()
property nfeat
class ciderpress.dft.xc_evaluator.NNEvaluator(model, device=None)
build(*args, **kwargs)
set_device(device=None)
class ciderpress.dft.xc_evaluator.RBFEvaluator(kernel, X1ctrl, alpha)
class ciderpress.dft.xc_evaluator.SpinRBFEvaluator(kernel, X1ctrl, alpha)
class ciderpress.dft.xc_evaluator.SplineSetEvaluator(scale, ind_sets, spline_grids, coeff_sets, const=0)
classmethod from_dict(d)
to_dict()
class ciderpress.dft.xc_evaluator.XCEvalSerializable
dump(fname)

Save the Evaluator to a file name fname as yaml format.

classmethod from_dict(d)
classmethod load(fname)

Load an instance of this class from yaml

to_dict()

Libxc-baseline/full-XC interface

class ciderpress.dft.xc_evaluator2.MappedDFTKernel2(fevals, feature_list, mode, multiplicative_baseline, additive_baseline=None, omega=0.0, component=None)

This class evaluates the XC term arising from a single DFTKernel object, using one or a list of FuncEvaluator objects.

fevals: FuncEvaluator

class ciderpress.dft.xc_evaluator2.MappedXC2(mapped_kernels, settings, libxc_baseline=None)

Inference-time evaluator for mapped density-dependent kernels.

MappedXC2 extends the mapped-model concept of MappedXC to kernels that take density arguments in addition to normalized features. This representation is used by both range-separated exchange models and the CIDER26XC full exchange-correlation models.

Parameters:

  • mapped_kernels (list of MappedDFTKernel2)

  • settings (FeatureSettings)

  • libxc_baseline (str)