LocalModel
==========

.. toctree::
   :hidden:

   /autoapi/t3toolbox/backend/optimizers/LocalModel.jacobian
   /autoapi/t3toolbox/backend/optimizers/LocalModel.gn_quadratic
   /autoapi/t3toolbox/backend/optimizers/LocalModel.hvp
   /autoapi/t3toolbox/backend/optimizers/LocalModel.retract

.. py:class:: t3toolbox.backend.optimizers.LocalModel

   The Gauss-Newton model linearized at a point -- the backend twin of ``fitting.GaussNewtonModel``.
   Built by ``Problem.local_model``; the frame sweep is computed once and reused by every method below.


   .. py:attribute:: geom
      :type:  GeometryOps


   .. py:attribute:: kind
      :type:  t3toolbox.backend.common.typ.Any


   .. py:attribute:: sample
      :type:  t3toolbox.backend.common.typ.Any


   .. py:attribute:: frame
      :type:  t3toolbox.backend.common.typ.Tuple


   .. py:attribute:: sweep
      :type:  t3toolbox.backend.common.typ.Tuple


   .. py:attribute:: residual
      :type:  t3toolbox.backend.common.typ.Any


   .. py:attribute:: n_w
      :type:  int


   .. py:attribute:: regularizer
      :type:  t3toolbox.backend.common.typ.Any
      :value: None



   .. py:property:: misfit


   .. py:property:: regularization


   .. py:property:: objective


   .. py:property:: gradient
      :type: Tangent


Methods
-------

.. autoapisummary::

   t3toolbox.backend.optimizers.LocalModel.jacobian
   t3toolbox.backend.optimizers.LocalModel.gn_quadratic
   t3toolbox.backend.optimizers.LocalModel.hvp
   t3toolbox.backend.optimizers.LocalModel.retract