@@ -300,7 +300,13 @@ def _tensormap_to_matrix(mol, tensor, fast=False):
300300 mol (pyscf.gto.Mole): pyscf Mole object.
301301 tensor (metatensor.TensorMap): tensor to transform.
302302 fast (bool): Whether to use a faster approach using assumptions
303- on the internal ordering of the TensorMap. Default is False.
303+ on the internal ordering of the TensorMap. Default is False.
304+ Assumptions:
305+ * tensor keys are ordered as (o3_lambda1, o3_lambda2, center_type1, center_type2),
306+ * for each element, the basis functions are ordered wrt L,
307+ * the individual samples are ordered as nested itemizations of atom_i1,atom_i2,
308+ the components as that of m1,m2, the properties as that of n1,n2 in this order.
309+ This structure correstonds to `_matrix_to_tensormap()` output.
304310
305311 Returns:
306312 numpy.ndarray: 2D array (matrix) representation.
@@ -404,7 +410,8 @@ def tensormap_to_array(mol, tensor, dest='pyscf', fast=False):
404410 tensor (metatensor.TensorMap): Tensor to transform.
405411 dest (str): Destination AO ordering of the output array. Default is 'pyscf'.
406412 fast (bool): Whether to use a faster approach for matrix conversion
407- using assumptions on the internal ordering of the TensorMap. Default is False.
413+ using assumptions on the internal ordering of the TensorMap
414+ (see `_tensormap_to_matrix()`). Default is False.
408415
409416 Returns:
410417 numpy.ndarray: Array representation (1D vector or 2D matrix).
0 commit comments