BQState

class probnum.quad.solvers.BQState(measure, kernel, scale_sq=1.0, integral_belief=None, previous_integral_beliefs=(), nodes=None, fun_evals=None, gram=array([], shape=(1, 0), dtype=float64), kernel_means=array([], dtype=float64))

Bases: object

Container for the quantities defining the BQ problem and the BQ belief.

Parameters
  • measure (IntegrationMeasure) – The integration measure.

  • kernel (Kernel) – The kernel used for BQ.

  • scale_sq (FloatLike) – Square of the kernel scaling parameter.

  • integral_belief (Optional[Normal]) – Normal distribution over the value of the integral.

  • previous_integral_beliefs (Tuple[Normal]) – Integral beliefs on computed on previous iterations.

  • nodes (Optional[np.ndarray]) – All locations at which function evaluations are available.

  • fun_evals (Optional[np.ndarray]) – Function evaluations at nodes.

  • gram (np.ndarray) – The kernel Gram matrix.

  • kernel_means (np.ndarray) – All kernel mean evaluations at nodes.

See also

BQIterInfo

Container for quantities concerning the BQ loop iteration.

Methods Summary

from_new_data(kernel, scale_sq, nodes, ...)

Initialize state from updated data.

Methods Documentation

classmethod from_new_data(kernel, scale_sq, nodes, fun_evals, integral_belief, prev_state, gram, kernel_means)[source]

Initialize state from updated data.

Parameters
  • kernel (Kernel) – The kernel used for BQ.

  • scale_sq (FloatLike) – Square of the kernel scaling parameter.

  • nodes (ndarray) – All locations at which function evaluations are available.

  • fun_evals (ndarray) – Function evaluations at nodes.

  • integral_belief (Normal) – Normal distribution over the value of the integral.

  • prev_state (BQState) – Previous state of the BQ loop.

  • gram (ndarray) – The Gram matrix of the given nodes.

  • kernel_means (ndarray) – The kernel means at the given nodes.

Returns

An instance of this class.

Return type

bq_state