.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "user/_auto_howtos/image_2d.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download_user__auto_howtos_image_2d.py>`
        to download the full example code.

.. rst-class:: sphx-glr-example-title

.. _sphx_glr_user__auto_howtos_image_2d.py:


The use of Image2D
==================

This script demonstrates the use of :class:`~cuqi.geometry.Image2D` for domain and range geometries in CUQIpy forward models. The settings of this geometry, in particular the so-called `visual_only` flag, depend on how the forward model input is operated on in the forward model. `visual_only` flag tells the geometry to interpret the variable as a 2D image only when visualizing the variable but keep it as a vector when being operated on in the forward model. Generally, if the forward operator expects an image input, set `visual_only` to `False`, which is the default in CUQIpy. Otherwise, if it expects a vector input, set `visual_only` to `True` in the domain geometry. In the same way, the range geometry `visual_only` setting should be consistent with your forward operator output.

.. GENERATED FROM PYTHON SOURCE LINES 9-10

Import necessary libraries

.. GENERATED FROM PYTHON SOURCE LINES 10-14

.. code-block:: Python

    import cuqi
    import numpy as np
    import matplotlib.pyplot as plt








.. GENERATED FROM PYTHON SOURCE LINES 15-16

Define dimensions

.. GENERATED FROM PYTHON SOURCE LINES 16-18

.. code-block:: Python

    dim_x, dim_y = 2, 4








.. GENERATED FROM PYTHON SOURCE LINES 19-25

Example illustrating the default behavior: `visual_only=False`
 With `visual_only=False`, the underlying structure of a CUQI array or sample will be reshaped to the
 shape of the specified geometry, i.e., an image of size (dim_x, dim_y), before being passed to the model.
 Here we intend to define a forward operator that downsamples the input image by a factor of 2. It's easier
 to define such a forward operator in terms of an image. We do not need to explicitly set `visual_only` to 
`False` because it is the default behavior.

.. GENERATED FROM PYTHON SOURCE LINES 25-50

.. code-block:: Python


    def forward_func_a(image):
        assert image.shape == (dim_x, dim_y)
        return image[::2, ::2]

    model_A = cuqi.model.LinearModel(forward_func_a,
                                     domain_geometry=cuqi.geometry.Image2D((dim_x, dim_y)),
                                     range_geometry=cuqi.geometry.Image2D((dim_x // 2, dim_y // 2)))

    print(model_A)

    x_sample = cuqi.array.CUQIarray(np.linspace(0, 1, dim_x * dim_y),
                                    geometry=cuqi.geometry.Image2D((dim_x, dim_y)))

    # Plot the original sample
    plt.figure()
    x_sample.plot(cmap='Greens')
    plt.title("input image x")

    # Apply the model and plot the result
    plt.figure()
    y_sample = model_A @ x_sample
    y_sample.plot(cmap='Greens')
    plt.title("output y=model_A(x)")




.. rst-class:: sphx-glr-horizontal


    *

      .. image-sg:: /user/_auto_howtos/images/sphx_glr_image_2d_001.png
         :alt: input image x
         :srcset: /user/_auto_howtos/images/sphx_glr_image_2d_001.png
         :class: sphx-glr-multi-img

    *

      .. image-sg:: /user/_auto_howtos/images/sphx_glr_image_2d_002.png
         :alt: output y=model_A(x)
         :srcset: /user/_auto_howtos/images/sphx_glr_image_2d_002.png
         :class: sphx-glr-multi-img


.. rst-class:: sphx-glr-script-out

 .. code-block:: none

    CUQI LinearModel: Image2D[8: (2, 4)] -> Image2D[2: (1, 2)].
        Forward parameters: ['image'].

    Text(0.5, 1.0, 'output y=model_A(x)')



.. GENERATED FROM PYTHON SOURCE LINES 51-55

Example with `visual_only=True`
With `visual_only=True`, the underlying structure of a cuqi array or sample will not be changed before
being passed to the model. This is useful when the model expects a vector input. Here we define a 
forward operator that reverses the input vector, which can be easily defined in terms of a vector.

.. GENERATED FROM PYTHON SOURCE LINES 55-70

.. code-block:: Python

    def forward_func_b(x):
        assert x.shape == (dim_x * dim_y, )
        return x[::-1]

    model_B = cuqi.model.LinearModel(forward_func_b,
                                     domain_geometry=cuqi.geometry.Image2D((dim_x, dim_y), visual_only=True),
                                     range_geometry=cuqi.geometry.Image2D((dim_x, dim_y), visual_only=True))

    # Note we explicitely set the `visual_only` flag to be `True` here.
    print(model_B)

    # Apply the model and plot the result
    plt.figure()
    y_sample = model_B @ x_sample
    y_sample.plot(cmap='Greens')
    plt.title("output y=model_B(x)")


.. image-sg:: /user/_auto_howtos/images/sphx_glr_image_2d_003.png
   :alt: output y=model_B(x)
   :srcset: /user/_auto_howtos/images/sphx_glr_image_2d_003.png
   :class: sphx-glr-single-img


.. rst-class:: sphx-glr-script-out

 .. code-block:: none

    CUQI LinearModel: Image2D[8] -> Image2D[8].
        Forward parameters: ['x'].

    Text(0.5, 1.0, 'output y=model_B(x)')




.. rst-class:: sphx-glr-timing

   **Total running time of the script:** (0 minutes 0.139 seconds)


.. _sphx_glr_download_user__auto_howtos_image_2d.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: image_2d.ipynb <image_2d.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: image_2d.py <image_2d.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: image_2d.zip <image_2d.zip>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_