update doc

docs/examples/index.rst

@@ -19,14 +19,14 @@ Below is a gallery of examples on how to use DeepHyper.
 
 .. only:: html
 
-  .. image:: /examples/images/thumb/sphx_glr_plot_from_serial_to_parallel_hyperparameter_search_thumb.png
+  .. image:: /examples/images/thumb/sphx_glr_plot_experimental_design_thumb.png
     :alt:
 
-  :ref:`sphx_glr_examples_plot_from_serial_to_parallel_hyperparameter_search.py`
+  :ref:`sphx_glr_examples_plot_experimental_design.py`
 
 .. raw:: html
 
-      <div class="sphx-glr-thumbnail-title">From Serial to Parallel Evaluations</div>
+      <div class="sphx-glr-thumbnail-title">Profile the Worker Utilization</div>
     </div>
 
 
@@ -64,6 +64,23 @@ Below is a gallery of examples on how to use DeepHyper.
     </div>
 
 
+.. raw:: html
+
+    <div class="sphx-glr-thumbcontainer" tooltip="Author(s): Romain Egele.">
+
+.. only:: html
+
+  .. image:: /examples/images/thumb/sphx_glr_plot_from_serial_to_parallel_hyperparameter_search_thumb.png
+    :alt:
+
+  :ref:`sphx_glr_examples_plot_from_serial_to_parallel_hyperparameter_search.py`
+
+.. raw:: html
+
+      <div class="sphx-glr-thumbnail-title">From Serial to Parallel Evaluations</div>
+    </div>
+
+
 .. raw:: html
 
     <div class="sphx-glr-thumbcontainer" tooltip="Author(s): Romain Egele.">
@@ -89,9 +106,10 @@ Below is a gallery of examples on how to use DeepHyper.
 .. toctree::
    :hidden:
 
-   /examples/plot_from_serial_to_parallel_hyperparameter_search
+   /examples/plot_experimental_design
    /examples/plot_notify_failures_hyperparameter_search
    /examples/plot_transfer_learning_for_hps
+   /examples/plot_from_serial_to_parallel_hyperparameter_search
    /examples/plot_profile_worker_utilization
 
 

docs/examples/plot_experimental_design.ipynb

@@ -0,0 +1,115 @@
+{
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "\n# Profile the Worker Utilization\n\n**Author(s)**: Romain Egele.\n\nThis example demonstrates how to generate points from standard experimental designs (e.g., random, grid, lhs).\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "from deephyper.analysis._matplotlib import update_matplotlib_rc\n\nupdate_matplotlib_rc()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "First we define the hyperparameter search space.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "from deephyper.problem import HpProblem\n\n\nproblem = HpProblem()\nproblem.add_hyperparameter((0.0001, 100.0, \"log-uniform\"), \"x\")\nproblem.add_hyperparameter((0.0, 100.0), \"y\")\nproblem.add_hyperparameter([1, 2, 3], \"z\")\nproblem\nproblem"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Then we define the black-box function to optimize.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "def run(job):\n    config = job.parameters\n    objective = config[\"x\"] + config[\"y\"]\n    return objective"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Then we define the search. In this example, we use the `ExperimentalDesignSearch` class to generate points from a grid design. The `Evaluator` can also be used with this class to parallelize evalutions.\nNote that `n_points` and `max_evals` take the same value here.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "from deephyper.search.hps import ExperimentalDesignSearch\n\n\nmax_evals = 200\nsearch = ExperimentalDesignSearch(problem, run, n_points=max_evals, design=\"grid\")\nresults = search.search(max_evals)"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "Finally, we plot the results from the collected DataFrame.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "import matplotlib.pyplot as plt\n\n\nfig, ax = plt.subplots()\nax.scatter(results[\"p:x\"], results[\"p:y\"], c=results[\"p:z\"], alpha=0.3)\nax.set_xscale(\"log\")\nplt.xlabel(\"x\")\nplt.ylabel(\"y\")\nplt.show()"
+      ]
+    }
+  ],
+  "metadata": {
+    "kernelspec": {
+      "display_name": "Python 3",
+      "language": "python",
+      "name": "python3"
+    },
+    "language_info": {
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
+      "name": "python",
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.11.6"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}

docs/examples/plot_experimental_design.py

@@ -0,0 +1,57 @@
+# -*- coding: utf-8 -*-
+"""
+Profile the Worker Utilization
+==============================
+
+**Author(s)**: Romain Egele.
+
+This example demonstrates how to generate points from standard experimental designs (e.g., random, grid, lhs).
+"""
+
+
+from deephyper.analysis._matplotlib import update_matplotlib_rc
+
+update_matplotlib_rc()
+
+# %%
+# First we define the hyperparameter search space.
+from deephyper.problem import HpProblem
+
+
+problem = HpProblem()
+problem.add_hyperparameter((0.0001, 100.0, "log-uniform"), "x")
+problem.add_hyperparameter((0.0, 100.0), "y")
+problem.add_hyperparameter([1, 2, 3], "z")
+problem
+problem
+
+
+# %%
+# Then we define the black-box function to optimize.
+def run(job):
+    config = job.parameters
+    objective = config["x"] + config["y"]
+    return objective
+
+
+# %%
+# Then we define the search. In this example, we use the `ExperimentalDesignSearch` class to generate points from a grid design. The `Evaluator` can also be used with this class to parallelize evalutions.
+# Note that `n_points` and `max_evals` take the same value here.
+from deephyper.search.hps import ExperimentalDesignSearch
+
+
+max_evals = 200
+search = ExperimentalDesignSearch(problem, run, n_points=max_evals, design="grid")
+results = search.search(max_evals)
+
+# %%
+# Finally, we plot the results from the collected DataFrame.
+import matplotlib.pyplot as plt
+
+
+fig, ax = plt.subplots()
+ax.scatter(results["p:x"], results["p:y"], c=results["p:z"], alpha=0.3)
+ax.set_xscale("log")
+plt.xlabel("x")
+plt.ylabel("y")
+plt.show()

docs/examples/plot_experimental_design.py.md5

@@ -0,0 +1 @@
+7f69999a456e564a57a3f22168030ff9