deeperwin.optimization.build_value_and_grad_func

deeperwin.optimization.build_value_and_grad_func(log_psi_sqr_func, clipping_config: deeperwin.configuration.ClippingConfig)[source]

Returns a callable that computes the gradient of the mean local energy for a given set of MCMC walkers with respect to the model defined by log_psi_func.

Parameters

log_psi_sqr_func (callable) – A function representing the wavefunction model
clipping_config (ClippingConfig) – Clipping hyperparameters
use_fwd_fwd_hessian (bool) – If true, the second partial derivatives required for computing the local energy are obtained with a forward-forward scheme.