def unwarp()

in bayesmark/space.py [0:0]

• 10 lines of code
• 7 McCabe index (conditional complexity)

    def unwarp(self, X_w, fixed_vals={}):
        """Inverse of :func:`.warp`.

        Parameters
        ----------
        X_w : :class:`numpy:numpy.ndarray` of shape (n, m)
            Warped version of input space. Must be 2D `float` :class:`numpy:numpy.ndarray`. `n` is the number of
            separate points in the warped joint space. `m` is the size of the joint warped space, which can be inferred
            in advance by calling :func:`.get_bounds`.
        fixed_vals : dict
            Subset of variables we want to keep fixed in X. Unwarp checks that the unwarped version of `X_w` matches
            `fixed_vals` up to numerical error. Otherwise, an error is raised.

        Returns
        -------
        X : list(dict(str, object)) of shape (n,)
            List of `n` points in the joint space to warp. Each list element is a dictionary where each key corresponds
            to a variable in the joint space.
        """
        X_w = np.asarray(X_w)
        check_array(X_w, "X", ndim=2, shape_endswith=(self.blocks[-1],), dtype=WARPED_DTYPE)
        N = X_w.shape[0]

        # Use snap_to to make sure we get exact value (no-round off) for cases where we know expected answer
        X = {
            param: snap_to(self.spaces[param].unwarp(xx), fixed_vals.get(param, None))
            for param, xx in zip(self.param_list, np.hsplit(X_w, self.blocks[:-1]))
        }
        # Convert dict of arrays to list of dicts, this would not be needed if
        # we used pandas but we do not want to add it as a dep. np.asscalar and
        # .item() appear to be the same thing but asscalar seems more readable.
        X = [{param: X[param][ii].item() for param in self.param_list} for ii in range(N)]
        return X