polish docs and update github action

.github/workflows/docs.yml

@@ -1,12 +1,11 @@
 name: Build and Deploy
 
 on:
-  push:
-    branches:
-      - master
+  release:
+    types: [published]
   pull_request:
-    branches:
-      - "*"
+    types:
+      - opened
 permissions:
   contents: write
 
@@ -29,14 +28,14 @@ jobs:
           sphinx-build -j 4 -T -b html docs/source docs/build/html
 
       - name: Deploy Stable 🚀
-        if: startsWith(github.ref, 'refs/tags/')
+        if: github.event_name == 'release'
         uses: JamesIves/github-pages-deploy-action@v4
         with:
           folder: docs/build/html
           branch: gh-pages
 
       - name: Deploy Dev 🚀
-        if: ${{ ! startsWith(github.ref, 'refs/tags/') }}
+        if: github.event_name == 'pull_request'
         uses: JamesIves/github-pages-deploy-action@v4
         with:
           folder: docs/build/html

docs/source/learn.md

@@ -195,7 +195,7 @@ A generalisation of the binary Hierarchical Gaussian Filter to multiarmed bandit
 Hand-on exercises for theComputational Psychiatry Course (Zurich) to build intuition around the generalised Hierarchical Gaussian Filter, how to create and manipulate probabilistic networks, design an agent to perform a reinforcement learning task and use MCMC sampling for parameter inference and model comparison—about 4 hours.
 
 
-::::{grid} 1 1 2 3
+::::{grid} 1 1 2 2
 :gutter: 1
 
 :::{grid-item-card} Introduction to the Generalised Hierarchical Gaussian Filter

docs/source/notebooks/0.1-Theory.ipynb

@@ -42,14 +42,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": null,
    "id": "6430ec8b-1b42-450b-9ed5-a9866384042f",
    "metadata": {},
    "outputs": [],
    "source": [
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
-    "import seaborn as sns"
+    "import seaborn as sns\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -416,8 +418,6 @@
     "\n",
     "axs[1, 2].plot(x_1_3, \"-\", linewidth=1, color=\"#55a868\")\n",
     "axs[1, 2].set(xlabel=\"Time steps\", ylabel=r\"$x_1$\")\n",
-    "\n",
-    "plt.tight_layout()\n",
     "sns.despine()"
    ]
   },
@@ -679,6 +679,7 @@
     "\n",
     "Gaussian Random Walk\n",
     "    A Gaussian Random Walk is a [Random Walk](https://en.wikipedia.org/wiki/Random_walk) with normally distributed increments. This is a mathematical model used to describe the behaviour of a variable that changes randomly over time. The Generalized Hierarchical Gaussian Filter implies that the behaviour of continuous state nodes is guided by this kind of process. This process can be further complexified by adding a constant drift or an autoregression component.\n",
+    "\n",
     "Node\n",
     "    HGF models are defined as networks of probabilistic nodes. A node is defined by its parameters (e.g. sufficient statistics, coupling values...) and by its connection with other nodes. The dependencies structure can have more than one dimension (i.e. there are many kinds of dependencies between nodes, especially here the volatility coupling and the value coupling).\n",
     "\n",

docs/source/notebooks/0.2-Creating_networks.ipynb

@@ -51,7 +51,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": null,
    "id": "81a63672",
    "metadata": {
     "editable": true,
@@ -66,7 +66,9 @@
     "import numpy as np\n",
     "import seaborn as sns\n",
     "from pyhgf.model import Network, HGF\n",
-    "from pyhgf.typing import AdjacencyLists"
+    "from pyhgf.typing import AdjacencyLists\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -881,7 +883,6 @@
     "many_value_children_hgf.input_data(input_data=input_data)\n",
     "\n",
     "many_value_children_hgf.plot_nodes([3, 2], figsize=(12, 5), show_observations=True)\n",
-    "plt.tight_layout()\n",
     "sns.despine()"
    ]
   },
@@ -1089,7 +1090,6 @@
     "    figsize=(12, 8),\n",
     "    show_observations=False,\n",
     ")\n",
-    "plt.tight_layout()\n",
     "sns.despine()"
    ]
   },
@@ -1317,7 +1317,6 @@
     "axs[-2].legend()\n",
     "axs[-3].plot(u_1_prob, label=\"Contingencies - Input Node 1\", linestyle=\":\")\n",
     "axs[-3].legend()\n",
-    "plt.tight_layout()\n",
     "sns.despine()"
    ]
   },

docs/source/notebooks/0.3-Generalised_filtering.ipynb

@@ -39,18 +39,17 @@
    ],
    "source": [
     "import jax.numpy as jnp\n",
-    "import matplotlib.animation as animation\n",
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
     "import seaborn as sns\n",
     "from matplotlib.ticker import MultipleLocator\n",
     "from pyhgf.math import MultivariateNormal, Normal, gaussian_predictive_distribution\n",
     "from pyhgf.model import HGF, Network\n",
     "from pyhgf.utils import beliefs_propagation\n",
-    "from scipy.special import gammaln\n",
     "from scipy.stats import norm, t\n",
     "\n",
-    "np.random.seed(123)"
+    "np.random.seed(123)\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -903,7 +902,6 @@
     "    ax[i].set_ylabel(rf\"$\\nu_{i}$\")\n",
     "    ax[i].set_xlabel(\"Observations\")\n",
     "    ax[i].legend()\n",
-    "plt.tight_layout()\n",
     "sns.despine()"
    ]
   },

docs/source/notebooks/1.1-Binary_HGF.ipynb

@@ -72,11 +72,12 @@
     "import jax.numpy as jnp\n",
     "import matplotlib.pyplot as plt\n",
     "import pymc as pm\n",
-    "import seaborn as sns\n",
     "from pyhgf import load_data\n",
     "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.model import HGF\n",
-    "from pyhgf.response import first_level_binary_surprise"
+    "from pyhgf.response import first_level_binary_surprise\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -890,8 +891,7 @@
     }
    ],
    "source": [
-    "az.plot_trace(two_level_hgf_idata, var_names=[\"tonic_volatility_2\"])\n",
-    "plt.tight_layout()"
+    "az.plot_trace(two_level_hgf_idata, var_names=[\"tonic_volatility_2\"]);"
    ]
   },
   {
@@ -1322,8 +1322,7 @@
    "source": [
     "az.plot_trace(\n",
     "    three_level_hgf_idata, var_names=[\"tonic_volatility_2\", \"tonic_volatility_3\"]\n",
-    ")\n",
-    "plt.tight_layout()"
+    ");"
    ]
   },
   {

docs/source/notebooks/1.2-Categorical_HGF.ipynb

@@ -58,17 +58,13 @@
    },
    "outputs": [],
    "source": [
-    "import arviz as az\n",
     "import jax.numpy as jnp\n",
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
-    "import pymc as pm\n",
-    "import pytensor\n",
     "import pytensor.tensor as pt\n",
     "import seaborn as sns\n",
-    "from jax import grad, jit, vjp\n",
+    "from jax import jit, vjp\n",
     "from jax.tree_util import Partial\n",
-    "from pyhgf import load_data\n",
     "from pyhgf.model import HGF\n",
     "from pyhgf.plots import plot_nodes\n",
     "from pytensor.graph import Apply, Op"
@@ -786,8 +782,7 @@
     "axs[4].set_title(\"Sum of binary surprises\", loc=\"left\")\n",
     "axs[2].set_ylabel(\"Surprises\")\n",
     "\n",
-    "sns.despine()\n",
-    "plt.tight_layout()"
+    "sns.despine()"
    ]
   },
   {

docs/source/notebooks/1.3-Continuous_HGF.ipynb

@@ -77,12 +77,13 @@
     "import jax.numpy as jnp\n",
     "import matplotlib.pyplot as plt\n",
     "import pymc as pm\n",
-    "import seaborn as sns\n",
     "\n",
     "from pyhgf import load_data\n",
     "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.model import HGF\n",
-    "from pyhgf.response import first_level_gaussian_surprise"
+    "from pyhgf.response import first_level_gaussian_surprise\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -1373,8 +1374,7 @@
     }
    ],
    "source": [
-    "az.plot_trace(three_level_hgf_idata);\n",
-    "plt.tight_layout()"
+    "az.plot_trace(three_level_hgf_idata);"
    ]
   },
   {

docs/source/notebooks/2-Using_custom_response_functions.ipynb

@@ -79,7 +79,9 @@
     "\n",
     "from pyhgf import load_data\n",
     "from pyhgf.distribution import HGFDistribution\n",
-    "from pyhgf.model import HGF"
+    "from pyhgf.model import HGF\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -988,7 +990,6 @@
    ],
    "source": [
     "az.plot_trace(sigmoid_hgf_idata, var_names=[\"tonic_volatility_2\"]);\n",
-    "plt.tight_layout()\n",
     "az.summary(sigmoid_hgf_idata, var_names=[\"tonic_volatility_2\"])"
    ]
   },

docs/source/notebooks/3-Multilevel_HGF.ipynb

@@ -85,13 +85,13 @@
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
     "import pymc as pm\n",
-    "import pytensor.tensor as tt\n",
     "import seaborn as sns\n",
-    "from numpy import loadtxt\n",
     "from pyhgf import load_data\n",
-    "from pyhgf.distribution import HGFDistribution, hgf_logp\n",
+    "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.model import HGF\n",
-    "from pyhgf.response import binary_softmax_inverse_temperature"
+    "from pyhgf.response import binary_softmax_inverse_temperature\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -625,8 +625,7 @@
     "    two_level_hgf_idata,\n",
     "    var_names=[\"mu_temperature\", \"mu_volatility\"],\n",
     "    ref_val=[0.5, -4.0],\n",
-    ")\n",
-    "plt.tight_layout()"
+    ")"
    ]
   },
   {

docs/source/notebooks/4-Parameter_recovery.ipynb

@@ -78,9 +78,8 @@
     "import numpy as np\n",
     "import pymc as pm\n",
     "import seaborn as sns\n",
-    "from numpy import loadtxt\n",
     "from pyhgf import load_data\n",
-    "from pyhgf.distribution import HGFDistribution, hgf_logp\n",
+    "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.math import sigmoid_inverse_temperature\n",
     "from pyhgf.model import HGF\n",
     "from pyhgf.response import binary_softmax_inverse_temperature"

docs/source/notebooks/Example_1_Heart_rate_variability.ipynb

@@ -75,15 +75,16 @@
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
     "import pymc as pm\n",
-    "import seaborn as sns\n",
     "from systole import import_dataset1\n",
     "from systole.detection import ecg_peaks\n",
     "from systole.plots import plot_raw\n",
     "from systole.utils import input_conversion\n",
     "\n",
     "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.model import HGF\n",
-    "from pyhgf.response import total_gaussian_surprise"
+    "from pyhgf.response import total_gaussian_surprise\n",
+    "\n",
+    "plt.rcParams[\"figure.constrained_layout.use\"] = True"
    ]
   },
   {
@@ -410,8 +411,7 @@
     }
    ],
    "source": [
-    "az.plot_trace(idata)\n",
-    "plt.tight_layout()"
+    "az.plot_trace(idata);"
    ]
   },
   {

docs/source/notebooks/Example_2_Input_node_volatility_coupling.ipynb

@@ -63,14 +63,10 @@
    },
    "outputs": [],
    "source": [
-    "import arviz as az\n",
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
-    "import pymc as pm\n",
     "import seaborn as sns\n",
     "from scipy.stats import norm\n",
-    "\n",
-    "from pyhgf.distribution import HGFDistribution\n",
     "from pyhgf.model import Network"
    ]
   },