romtools.rom.multifidelity#

Jointly conditioned autoregressive multifidelity Gaussian-process QoI surrogate.

Each fidelity has a different, potentially non-nested set of training inputs. The public interface supports any number of ordered fidelities (two or more).

Classes

AutoregressiveGPBuilder(config[, ...])

Build a jointly trained AR-GP from all accepted fidelity datasets.

AutoregressiveGPConfig(levels[, ...])

Configuration of a joint autoregressive GP hierarchy.

AutoregressiveGPQoiModel(fidelity_data, config)

Scalar-QoI Kennedy--O'Hagan autoregressive GP with joint conditioning.

FidelityData(parameters, qois[, ...])

Accepted observations at one fidelity, in matching row order.

GPLevelConfig(kernel[, rho, noise_variance])

Base kernel for level zero; discrepancy kernel and rho for later levels.

class romtools.rom.multifidelity.AutoregressiveGPBuilder(config, parameter_names=None)[source]#

Bases: object

Build a jointly trained AR-GP from all accepted fidelity datasets.

Parameters:
class romtools.rom.multifidelity.AutoregressiveGPConfig(levels, tune_hyperparameters=True, normalize_parameters=True, optimizer_restarts=4, length_scale_bounds=(0.01, 100.0), signal_variance_bounds=(1e-06, 100.0), rho_bounds=(-5.0, 5.0))[source]#

Bases: object

Configuration of a joint autoregressive GP hierarchy.

Parameters:
  • levels (Sequence[GPLevelConfig])

  • tune_hyperparameters (bool)

  • normalize_parameters (bool)

  • optimizer_restarts (int)

  • length_scale_bounds (tuple[float, float])

  • signal_variance_bounds (tuple[float, float])

  • rho_bounds (tuple[float, float])

class romtools.rom.multifidelity.AutoregressiveGPQoiModel(fidelity_data, config, parameter_names=None)[source]#

Bases: object

Scalar-QoI Kennedy–O’Hagan autoregressive GP with joint conditioning.

f_0 = g_0; f_l = rho_l * f_(l-1) + delta_l, where the base process and all discrepancies are independent zero-mean GPs. Hyperparameters are fitted by joint marginal likelihood; in particular, the high-fidelity observations also inform the base process posterior.

Parameters:
predict(parameters, fidelity=-1, return_std=False)[source]#

Predict latent QoI mean and optionally posterior standard deviation.

Parameters:
  • fidelity (int)

  • return_std (bool)

class romtools.rom.multifidelity.FidelityData(parameters, qois, training_dirs=(), name=None)[source]#

Bases: object

Accepted observations at one fidelity, in matching row order.

Parameters:
  • parameters (ndarray)

  • qois (ndarray)

  • training_dirs (tuple[str, ...])

  • name (str | None)

class romtools.rom.multifidelity.GPLevelConfig(kernel, rho=1.0, noise_variance=1e-08)[source]#

Bases: object

Base kernel for level zero; discrepancy kernel and rho for later levels.

Parameters: