Github action: auto-update.

Author: github-actions[bot]

dev/_downloads/02a7a82499cf14d2702c9234035c44c4/plot_diffusion_advection_solver.zip

@@ -4,14 +4,14 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "\n\n# A simple finite-difference solver\nAn intro to our loss module's finite difference utility demonstrating\nits use to create a simple numerical solver for the diffusion-advection equation.\n"
+        "\n\n# A simple finite-difference solver for the diffusion-advection equation\nAn intro to our loss module's finite difference utility demonstrating\nits use to create a simple numerical solver for the diffusion-advection equation.\n"
       ]
     },
     {
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## Import the library\nWe first import our `neuralop` library and required dependencies.\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Import the library\nWe first import our `neuralop` library and required dependencies.\n\n"
       ]
     },
     {
@@ -29,7 +29,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## Defining our problem\nWe aim to solve the 2D diffusion advection equation:\n\n$u_t + cx \\cdot u_x + cy \\cdot u_y = \\nu (u_xx + u_yy) + f(x,y,t)$,\n\nWhere $f(x,y,t)$ is a source term and $cx$ and $cy$ are advection speeds in x and y.\nWe set simulation parameters below:\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Defining our problem\nWe aim to solve the 2D diffusion advection equation:\n\n\\begin{align}\\frac{\\partial u}{\\partial t} + c_x \\frac{\\partial u}{\\partial x} + c_y \\frac{\\partial u}{\\partial y} = \\nu \\left(\\frac{\\partial^2 u}{\\partial x^2} + \\frac{\\partial^2 u}{\\partial y^2}\\right) + f(x,y,t)\\end{align}\n\nWhere $f(x,y,t)$ is a source term and $c_x$ and $c_y$ are advection speeds in x and y.\nWe set simulation parameters below:\n\n"
       ]
     },
     {
@@ -47,7 +47,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## Simulate evolution using numerical solver\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Simulate evolution using numerical solver\n\n"
       ]
     },
     {
@@ -65,7 +65,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## Animate our solution\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Animate our solution\n\n"
       ]
     },
     {

dev/_downloads/07795e165bda9881ff980becb31905ab/plot_diffusion_advection_solver.ipynb

@@ -7,6 +7,13 @@
         "\n# Checkpointing and loading training states\n\nDemonstrating the ``Trainer``'s saving and loading functionality, \nwhich makes it easy to checkpoint and resume training states.\n"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Import dependencies\n\n"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -22,7 +29,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Loading the Navier-Stokes dataset in 128x128 resolution\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Loading the Darcy-Flow dataset\n\n"
       ]
     },
     {
@@ -40,7 +47,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "We create an FNO model\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Creating the FNO model\n\n"
       ]
     },
     {
@@ -58,7 +65,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Create the optimizer\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Creating the optimizer and scheduler\n\n"
       ]
     },
     {
@@ -76,7 +83,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Creating the losses\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Creating the losses\n\n"
       ]
     },
     {
@@ -90,6 +97,13 @@
         "l2loss = LpLoss(d=2, p=2)\nh1loss = H1Loss(d=2)\n\ntrain_loss = h1loss\neval_losses={'h1': h1loss, 'l2': l2loss}"
       ]
     },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Displaying configuration\n\n"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -105,7 +119,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Create the trainer\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Creating the trainer\n\n"
       ]
     },
     {
@@ -123,7 +137,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Actually train the model on our small Darcy-Flow dataset\n\n"
+        ".. raw:: html\n\n   <div style=\"margin-top: 3em;\"></div>\n\n## Training the model\nWe train and save checkpoints\n\n"
       ]
     },
     {

dev/_downloads/07fcc19ba03226cd3d83d4e40ec44385/auto_examples_python.zip

@@ -3,10 +3,19 @@
 
 A simple Darcy-Flow dataset
 ===========================
-An intro to the small Darcy-Flow example dataset we ship with the package.
+An introduction to the small Darcy-Flow example dataset we ship with the package.
+
+The Darcy-Flow problem is a fundamental partial differential equation (PDE) in fluid mechanics
+that describes the flow of a fluid through a porous medium. In this tutorial, we explore the
+dataset structure and visualize how the data is processed for neural operator training.
+
 """
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # Import the library
 # ------------------
 # We first import our `neuralop` library and required dependencies.
@@ -16,6 +25,10 @@
 from neuralop.layers.embeddings import GridEmbedding2D
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # Load the dataset
 # ----------------
 # Training samples are 16x16 and we load testing samples at both 
@@ -29,17 +42,22 @@
 train_dataset = train_loader.dataset
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # Visualizing the data
 # --------------------
+# Let's examine the shape and structure of our dataset at different resolutions.
 
 for res, test_loader in test_loaders.items():
-    print(res)
+    print(f"Resolution: {res}")
     # Get first batch
     batch = next(iter(test_loader))
-    x = batch['x']
-    y = batch['y']
+    x = batch['x']  # Input
+    y = batch['y']  # Output
 
-    print(f'Testing samples for res {res} have shape {x.shape[1:]}')
+    print(f'Testing samples for resolution {res} have shape {x.shape[1:]}')
 
 
 data = train_dataset[0]
@@ -48,7 +66,6 @@
 
 print(f'Training samples have shape {x.shape[1:]}')
 
-
 # Which sample to view
 index = 0
 
@@ -59,23 +76,34 @@
 # positional embedding. We will add it manually here to
 # visualize the channels appended by this embedding.
 positional_embedding = GridEmbedding2D(in_channels=1)
-# at train time, data will be collated with a batch dim.
-# we create a batch dim to pass into the embedding, then re-squeeze
+# At train time, data will be collated with a batch dimension.
+# We create a batch dimension to pass into the embedding, then re-squeeze
 x = positional_embedding(data['x'].unsqueeze(0)).squeeze(0)
 y = data['y']
+
+# %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
+# Visualizing the processed data
+# ------------------------------
+# We can see how the positional embedding adds coordinate information to our input data.
+# This helps the neural operator understand spatial relationships in the data.
+
 fig = plt.figure(figsize=(7, 7))
 ax = fig.add_subplot(2, 2, 1)
 ax.imshow(x[0], cmap='gray')
-ax.set_title('input x')
+ax.set_title('Input x')
 ax = fig.add_subplot(2, 2, 2)
 ax.imshow(y.squeeze())
-ax.set_title('input y')
+ax.set_title('Output y')
 ax = fig.add_subplot(2, 2, 3)
 ax.imshow(x[1])
-ax.set_title('x: 1st pos embedding')
+ax.set_title('Positional embedding: x-coordinates')
 ax = fig.add_subplot(2, 2, 4)
 ax.imshow(x[2])
-ax.set_title('x: 2nd pos embedding')
-fig.suptitle('Visualizing one input sample', y=0.98)
+ax.set_title('Positional embedding: y-coordinates')
+fig.suptitle('Visualizing one input sample with positional embeddings', y=0.98)
 plt.tight_layout()
 fig.show()

dev/_downloads/09a1631d14c6eb51e552aa53b9f6d3e8/checkpoint_FNO_darcy.ipynb

@@ -29,6 +29,10 @@
 from neuralop.layers.resample import resample
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # First, let's generate a data input. We create a high-resolution Gaussian Random Field (GRF), which 
 # is a smooth, continuous signal, making it ideal for visualizing the effects of resampling.
 device = 'cpu'
@@ -80,6 +84,10 @@ def generate_grf(shape, alpha=2.5, device='cpu'):
 low_res = 32
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # Now, let's use the ``resample`` function to simulate downsampling and upsampling operations.
 # This could for instance be used in the encoder and decoder of a U-Net architecture.
 # The function takes an input tensor, a `scale_factor`, and a list of 
@@ -95,6 +103,10 @@ def generate_grf(shape, alpha=2.5, device='cpu'):
 
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # Finally, let's visualize the results to see the effect of the ``resample`` function.
 
 fig, axs = plt.subplots(1, 3, figsize=(14, 6))
@@ -128,6 +140,10 @@ def generate_grf(shape, alpha=2.5, device='cpu'):
 plt.show()
 
 # %%
+# .. raw:: html
+# 
+#    <div style="margin-top: 3em;"></div>
+# 
 # The ``resample`` function effectively changes the resolution of the data.
 # Notice that the upsampled image on the right is a faithful, if slightly blurrier,
 # reconstruction of the original. This is because the downsampling step is lossy;