deepmd.tf.nvnmd.utils.weight
============================

.. py:module:: deepmd.tf.nvnmd.utils.weight


Attributes
----------

.. autoapisummary::

   deepmd.tf.nvnmd.utils.weight.log


Functions
---------

.. autoapisummary::

   deepmd.tf.nvnmd.utils.weight.get_weight
   deepmd.tf.nvnmd.utils.weight.get_normalize
   deepmd.tf.nvnmd.utils.weight.get_type_embedding_weight
   deepmd.tf.nvnmd.utils.weight.get_filter_weight
   deepmd.tf.nvnmd.utils.weight.get_filter_type_weight
   deepmd.tf.nvnmd.utils.weight.get_fitnet_weight
   deepmd.tf.nvnmd.utils.weight.get_type_weight
   deepmd.tf.nvnmd.utils.weight.get_constant_initializer


Module Contents
---------------

.. py:data:: log

.. py:function:: get_weight(weights: dict, key: str) -> Any | None

   
   Get weight value according to key.
















   ..
       !! processed by numpydoc !!

.. py:function:: get_normalize(weights: dict) -> tuple

   
   Get normalize parameter (avg and std) of :math:`s_{ji}`.
















   ..
       !! processed by numpydoc !!

.. py:function:: get_type_embedding_weight(weights: dict, layer_l: int) -> tuple

   
   Get weight and bias of type_embedding network.


   :Parameters:

       **weights** : :class:`python:dict`
           weights

       **layer_l**
           layer order in embedding network
           1~nlayer














   ..
       !! processed by numpydoc !!

.. py:function:: get_filter_weight(weights: int, spe_j: int, layer_l: int) -> tuple

   
   Get weight and bias of embedding network.


   :Parameters:

       **weights** : :class:`python:dict`
           weights

       **spe_j** : :class:`python:int`
           special order of neighbor atom j
           0~ntype-1

       **layer_l**
           layer order in embedding network
           1~nlayer














   ..
       !! processed by numpydoc !!

.. py:function:: get_filter_type_weight(weights: dict, layer_l: int) -> tuple

   
   Get weight and bias of two_side_type_embedding network.


   :Parameters:

       **weights** : :class:`python:dict`
           weights

       **layer_l**
           layer order in embedding network
           1~nlayer














   ..
       !! processed by numpydoc !!

.. py:function:: get_fitnet_weight(weights: dict, spe_i: int, layer_l: int, nlayer: int = 10) -> tuple

   
   Get weight and bias of fitting network.


   :Parameters:

       **weights** : :class:`python:dict`
           weights

       **spe_i** : :class:`python:int`
           special order of central atom i
           0~ntype-1

       **layer_l** : :class:`python:int`
           layer order in embedding network
           0~nlayer-1

       **nlayer** : :class:`python:int`
           number of layers














   ..
       !! processed by numpydoc !!

.. py:function:: get_type_weight(weights: dict, layer_l: int) -> Any

   
   Get weight and bias of fitting network.


   :Parameters:

       **weights** : :class:`python:dict`
           weights

       **layer_l** : :class:`python:int`
           layer order in embedding network
           0~nlayer-1














   ..
       !! processed by numpydoc !!

.. py:function:: get_constant_initializer(weights: dict, name: str) -> Any

   
   Get initial value by name and create a initializer.
















   ..
       !! processed by numpydoc !!