netket.vqs

This module defines the variational states, the heart of NetKet itself.

Abstract Interface

VariationalState	Abstract class for variational states representing either pure states or mixed quantum states.
VariationalMixedState

Concrete Variational States

FullSumState	Variational State for a variational quantum state computed on the whole Hilbert space without Monte Carlo sampling by summing over the whole Hilbert space.
MCState	Variational State for a Variational Neural Quantum State.
MCMixedState	Variational State for a Mixed Variational Neural Quantum State.

and the experimental Variational state for a single slater determinant state (which does not use Monte-Carlo sampling)

DeterminantVariationalState

Variational State for fermionic mean-field states (Hartree-Fock ansatz).

Functions

apply_operator	Apply an operator to a variational state.
local_estimators	Compute per-sample local estimator data for operator op on vstate.
get_local_kernel	Returns the function computing the local estimator for the given variational state and operator.
get_local_kernel_arguments	Returns the samples of vstate used to compute the expectation value of the operator O, and the connected elements and matrix elements.

Freezing parameters

The following functions return a new variational state in which a subset of the parameters has been frozen (moved from parameters into model_state), so they are automatically excluded from gradient computation and optimizer updates. See the freezing parameters example for a worked example.

freeze_parameters	Freeze a subset of model parameters in a variational state.
unfreeze_parameters	Restore all frozen parameters in vstate to the trainable set.