def _util_newton_updates()

in botorch/models/pairwise_gp.py [0:0]

• 32 lines of code
• 5 McCabe index (conditional complexity)

    def _util_newton_updates(self, dp, x0, max_iter=1, xtol=None) -> Tensor:
        r"""Make `max_iter` newton updates on utility.

        This is used in `forward` to calculate and fill in gradient into tensors.
        Instead of doing utility -= H^-1 @ g, use substition method.
        See more explanation in _update_utility_derived_values.dd
        By default only need to run one iteration just to fill the the gradients.

        Args:
            dp: (Transformed) datapoints.
            x0: A `batch_size x n` dimension tensor, initial values.
            max_iter: Max number of iterations.
            xtol: Stop creteria. If `None`, do not stop until
                finishing `max_iter` updates.
        """
        xtol = float("-Inf") if xtol is None else xtol
        D, DT, ch, ci = (
            self.D,
            self.DT,
            self.covar_chol,
            self.covar_inv,
        )
        covar = self.covar
        diff = float("Inf")
        i = 0
        x = x0
        eye = None
        while i < max_iter and diff > xtol:
            hl = self._hess_likelihood_f_sum(x, D, DT)
            cov_hl = covar @ hl
            if eye is None:
                eye = torch.eye(
                    cov_hl.size(-1),
                    dtype=dp.dtype,
                    device=dp.device,
                ).expand(cov_hl.shape)
            cov_hl = cov_hl + eye  # add 1 to cov_hl
            g = self._grad_posterior_f(x, dp, D, DT, ch, ci)
            cov_g = covar @ g.unsqueeze(-1)
            x_update = torch.linalg.solve(cov_hl, cov_g).squeeze(-1)
            x_next = x - x_update
            diff = torch.norm(x - x_next)
            x = x_next
            i += 1
        return x