From e226494c5349b53181a0d50bb21bc323b5a5a0ab Mon Sep 17 00:00:00 2001
From: Robert Lee <84771576+R-M-Lee@users.noreply.github.com>
Date: Tue, 1 Apr 2025 12:20:48 +0200
Subject: [PATCH 1/3] naive random sampling pfront estimation
---
.../Pareto_front_estimation_quick.ipynb | 607 ++++++++++++++++++
1 file changed, 607 insertions(+)
create mode 100644 tutorials/basic_examples/Pareto_front_estimation_quick.ipynb
diff --git a/tutorials/basic_examples/Pareto_front_estimation_quick.ipynb b/tutorials/basic_examples/Pareto_front_estimation_quick.ipynb
new file mode 100644
index 000000000..cbe0ec5a3
--- /dev/null
+++ b/tutorials/basic_examples/Pareto_front_estimation_quick.ipynb
@@ -0,0 +1,607 @@
+{
+ "cells": [
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "id": "0",
+ "metadata": {
+ "papermill": {
+ "duration": 0.005215,
+ "end_time": "2024-10-10T20:35:34.138342",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:34.133127",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "source": [
+ "# Estimating Pareto fronts using random samples\n",
+ "\n",
+ "## Imports"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "1",
+ "metadata": {
+ "papermill": {
+ "duration": 3.323246,
+ "end_time": "2024-10-10T20:35:37.465135",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:34.141889",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "outputs": [],
+ "source": [
+ "import os\n",
+ "\n",
+ "import matplotlib.pyplot as plt\n",
+ "import pandas as pd\n",
+ "\n",
+ "import bofire.strategies.api as strategies\n",
+ "from bofire.benchmarks.api import BNH, Benchmark\n",
+ "from bofire.data_models.strategies.api import MoboStrategy, RandomStrategy\n",
+ "from bofire.utils.multiobjective import get_pareto_front\n",
+ "\n",
+ "\n",
+ "SMOKE_TEST = os.environ.get(\"SMOKE_TEST\")\n",
+ "NUM_INIT_SAMPLES = 12 if not SMOKE_TEST else 20\n",
+ "NUM_SAMPLES = 10000 if not SMOKE_TEST else 100"
+ ]
+ },
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "id": "2",
+ "metadata": {
+ "papermill": {
+ "duration": 0.001315,
+ "end_time": "2024-10-10T20:35:37.468031",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:37.466716",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "source": [
+ "## Introduction\n",
+ "\n",
+ "In this notebook, we will take a simple multiobjective example and train surrogate models for each output. Then we will use a simple random sampling approach to estimate the Pareto front, which is the set of best possible compromises between the objectives. Here we will not do any Bayesian optimization iterations; we merely demonstrate the random sampling approach for Pareto fronts, which can provide a baseline for Pareto-front estimation methods.\n",
+ "\n",
+ "You can compare this naive approach with the more sophisticated method in, e.g., the [tutorial for the ZDT1 benchmark](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/011-ZDT1.ipynb). The random-sampling approach in this notebook will not work well for higher-dimensional problems because we are just hoping that some random samples will fall near the Pareto front and need to be extremely lucky for this to be the case when there are many dimensions.\n",
+ "\n",
+ "## Generating the initial data for modeling\n",
+ "\n",
+ "This function gives us a random dataset (including the 'measured' outputs) for the benchmark BNH, which has its own [tutorial](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/012-BNH.ipynb)."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "3",
+ "metadata": {
+ "papermill": {
+ "duration": 0.18909,
+ "end_time": "2024-10-10T20:35:37.658331",
+ "exception": true,
+ "start_time": "2024-10-10T20:35:37.469241",
+ "status": "failed"
+ },
+ "tags": []
+ },
+ "outputs": [],
+ "source": [
+ "def sample(benchmark: Benchmark, Nsamples: int = 10, inputs_only=False):\n",
+ " datamodel = RandomStrategy(domain=benchmark.domain)\n",
+ " sampler = strategies.map(data_model=datamodel)\n",
+ " samples = sampler.ask(Nsamples)\n",
+ " return samples if inputs_only else benchmark.f(samples, return_complete=True)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "4",
+ "metadata": {},
+ "source": [
+ "Let's use the function to generate the dataset that we will use for modeling:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "5",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/html": [
+ "\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " x1 \n",
+ " x2 \n",
+ " f1 \n",
+ " f2 \n",
+ " c1 \n",
+ " c2 \n",
+ " valid_c1 \n",
+ " valid_c2 \n",
+ " valid_f1 \n",
+ " valid_f2 \n",
+ " \n",
+ " \n",
+ " \n",
+ " \n",
+ " 0 \n",
+ " 3.235181 \n",
+ " 1.897152 \n",
+ " 56.262340 \n",
+ " 12.742250 \n",
+ " 6.713771 \n",
+ " 46.685594 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 1 \n",
+ " 0.944766 \n",
+ " 2.871801 \n",
+ " 36.559290 \n",
+ " 20.974156 \n",
+ " 24.692165 \n",
+ " 84.254376 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 2 \n",
+ " 4.620137 \n",
+ " 2.464522 \n",
+ " 109.678133 \n",
+ " 6.572943 \n",
+ " 6.218167 \n",
+ " 41.284482 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 3 \n",
+ " 2.993973 \n",
+ " 1.165587 \n",
+ " 41.289861 \n",
+ " 18.726870 \n",
+ " 5.382738 \n",
+ " 42.412421 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 4 \n",
+ " 1.476776 \n",
+ " 0.990304 \n",
+ " 12.646279 \n",
+ " 28.490769 \n",
+ " 13.393807 \n",
+ " 58.474973 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " x1 \n",
+ " x2 \n",
+ " c1_pred \n",
+ " c2_pred \n",
+ " f1 \n",
+ " f2 \n",
+ " c1_sd \n",
+ " c2_sd \n",
+ " f1_sd \n",
+ " f2_sd \n",
+ " f1_des \n",
+ " f2_des \n",
+ " c1_des \n",
+ " c2_des \n",
+ " valid_f1 \n",
+ " valid_f2 \n",
+ " \n",
+ " \n",
+ " \n",
+ " \n",
+ " 0 \n",
+ " 4.824193 \n",
+ " 1.418243 \n",
+ " 2.936403 \n",
+ " 32.164223 \n",
+ " 96.224058 \n",
+ " 12.813320 \n",
+ " 0.878153 \n",
+ " 1.507216 \n",
+ " 3.744860 \n",
+ " 1.108512 \n",
+ " -96.224058 \n",
+ " -12.813320 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 1 \n",
+ " 2.406195 \n",
+ " 1.414166 \n",
+ " 8.614710 \n",
+ " 50.625390 \n",
+ " 30.881127 \n",
+ " 19.762983 \n",
+ " 0.433521 \n",
+ " 0.821362 \n",
+ " 1.621519 \n",
+ " 0.591357 \n",
+ " -30.881127 \n",
+ " -19.762983 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 2 \n",
+ " 1.286647 \n",
+ " 1.204312 \n",
+ " 15.428966 \n",
+ " 63.005427 \n",
+ " 12.103810 \n",
+ " 28.668753 \n",
+ " 0.456731 \n",
+ " 0.867202 \n",
+ " 1.669497 \n",
+ " 0.625886 \n",
+ " -12.103810 \n",
+ " -28.668753 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 3 \n",
+ " 2.429053 \n",
+ " 2.916580 \n",
+ " 14.891103 \n",
+ " 65.707672 \n",
+ " 55.993033 \n",
+ " 10.877458 \n",
+ " 0.818691 \n",
+ " 1.394782 \n",
+ " 3.607390 \n",
+ " 1.067341 \n",
+ " -55.993033 \n",
+ " -10.877458 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 4 \n",
+ " 2.967768 \n",
+ " 1.533018 \n",
+ " 6.255027 \n",
+ " 45.460447 \n",
+ " 44.682814 \n",
+ " 16.057206 \n",
+ " 0.421102 \n",
+ " 0.801401 \n",
+ " 1.540146 \n",
+ " 0.574769 \n",
+ " -44.682814 \n",
+ " -16.057206 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ "
\n",
+ "
"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
+ "source": [
+ "true_front = get_pareto_front(\n",
+ " domain=bnh.domain, experiments=sample(bnh, Nsamples=NUM_SAMPLES)\n",
+ ")\n",
+ "\n",
+ "fig, ax = plt.subplots()\n",
+ "ax.plot(\n",
+ " dense_samples.f1,\n",
+ " dense_samples.f2,\n",
+ " \"o\",\n",
+ " alpha=0.2,\n",
+ " label=\"MOBO Predictions (random inputs)\",\n",
+ " color=\"green\",\n",
+ ")\n",
+ "ax.plot(\n",
+ " estimated_front.f1, estimated_front.f2, \"o\", label=\"MOBO Pareto front\", color=\"blue\"\n",
+ ")\n",
+ "ax.plot(\n",
+ " true_front.f1,\n",
+ " true_front.f2,\n",
+ " \"o\",\n",
+ " markersize=3,\n",
+ " label=\"True front (non-dominated random samples)\",\n",
+ " color=\"red\",\n",
+ ")\n",
+ "ax.plot(\n",
+ " init_data.f1, init_data.f2, \"o\", label=\"Initial samples for MOBO\", color=\"black\"\n",
+ ")\n",
+ "ax.set_xlabel(\"f1\")\n",
+ "ax.set_ylabel(\"f2\")\n",
+ "ax.legend()\n",
+ "plt.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": "3.10.12"
+ },
+ "papermill": {
+ "default_parameters": {},
+ "duration": 5.249527,
+ "end_time": "2024-10-10T20:35:38.481859",
+ "environment_variables": {},
+ "exception": true,
+ "parameters": {},
+ "start_time": "2024-10-10T20:35:33.232332",
+ "version": "2.6.0"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
From f4059131e37f2ccfaf6632c7683c32251862e524 Mon Sep 17 00:00:00 2001
From: Robert Lee <84771576+R-M-Lee@users.noreply.github.com>
Date: Tue, 1 Apr 2025 12:42:32 +0200
Subject: [PATCH 2/3] symlink to notebook for docs
---
docs/pareto_front_naive_estimation.ipynb | 607 +++++++++++++++++++++++
1 file changed, 607 insertions(+)
create mode 100644 docs/pareto_front_naive_estimation.ipynb
diff --git a/docs/pareto_front_naive_estimation.ipynb b/docs/pareto_front_naive_estimation.ipynb
new file mode 100644
index 000000000..cbe0ec5a3
--- /dev/null
+++ b/docs/pareto_front_naive_estimation.ipynb
@@ -0,0 +1,607 @@
+{
+ "cells": [
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "id": "0",
+ "metadata": {
+ "papermill": {
+ "duration": 0.005215,
+ "end_time": "2024-10-10T20:35:34.138342",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:34.133127",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "source": [
+ "# Estimating Pareto fronts using random samples\n",
+ "\n",
+ "## Imports"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "1",
+ "metadata": {
+ "papermill": {
+ "duration": 3.323246,
+ "end_time": "2024-10-10T20:35:37.465135",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:34.141889",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "outputs": [],
+ "source": [
+ "import os\n",
+ "\n",
+ "import matplotlib.pyplot as plt\n",
+ "import pandas as pd\n",
+ "\n",
+ "import bofire.strategies.api as strategies\n",
+ "from bofire.benchmarks.api import BNH, Benchmark\n",
+ "from bofire.data_models.strategies.api import MoboStrategy, RandomStrategy\n",
+ "from bofire.utils.multiobjective import get_pareto_front\n",
+ "\n",
+ "\n",
+ "SMOKE_TEST = os.environ.get(\"SMOKE_TEST\")\n",
+ "NUM_INIT_SAMPLES = 12 if not SMOKE_TEST else 20\n",
+ "NUM_SAMPLES = 10000 if not SMOKE_TEST else 100"
+ ]
+ },
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "id": "2",
+ "metadata": {
+ "papermill": {
+ "duration": 0.001315,
+ "end_time": "2024-10-10T20:35:37.468031",
+ "exception": false,
+ "start_time": "2024-10-10T20:35:37.466716",
+ "status": "completed"
+ },
+ "tags": []
+ },
+ "source": [
+ "## Introduction\n",
+ "\n",
+ "In this notebook, we will take a simple multiobjective example and train surrogate models for each output. Then we will use a simple random sampling approach to estimate the Pareto front, which is the set of best possible compromises between the objectives. Here we will not do any Bayesian optimization iterations; we merely demonstrate the random sampling approach for Pareto fronts, which can provide a baseline for Pareto-front estimation methods.\n",
+ "\n",
+ "You can compare this naive approach with the more sophisticated method in, e.g., the [tutorial for the ZDT1 benchmark](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/011-ZDT1.ipynb). The random-sampling approach in this notebook will not work well for higher-dimensional problems because we are just hoping that some random samples will fall near the Pareto front and need to be extremely lucky for this to be the case when there are many dimensions.\n",
+ "\n",
+ "## Generating the initial data for modeling\n",
+ "\n",
+ "This function gives us a random dataset (including the 'measured' outputs) for the benchmark BNH, which has its own [tutorial](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/012-BNH.ipynb)."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "3",
+ "metadata": {
+ "papermill": {
+ "duration": 0.18909,
+ "end_time": "2024-10-10T20:35:37.658331",
+ "exception": true,
+ "start_time": "2024-10-10T20:35:37.469241",
+ "status": "failed"
+ },
+ "tags": []
+ },
+ "outputs": [],
+ "source": [
+ "def sample(benchmark: Benchmark, Nsamples: int = 10, inputs_only=False):\n",
+ " datamodel = RandomStrategy(domain=benchmark.domain)\n",
+ " sampler = strategies.map(data_model=datamodel)\n",
+ " samples = sampler.ask(Nsamples)\n",
+ " return samples if inputs_only else benchmark.f(samples, return_complete=True)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "id": "4",
+ "metadata": {},
+ "source": [
+ "Let's use the function to generate the dataset that we will use for modeling:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "5",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/html": [
+ "\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " x1 \n",
+ " x2 \n",
+ " f1 \n",
+ " f2 \n",
+ " c1 \n",
+ " c2 \n",
+ " valid_c1 \n",
+ " valid_c2 \n",
+ " valid_f1 \n",
+ " valid_f2 \n",
+ " \n",
+ " \n",
+ " \n",
+ " \n",
+ " 0 \n",
+ " 3.235181 \n",
+ " 1.897152 \n",
+ " 56.262340 \n",
+ " 12.742250 \n",
+ " 6.713771 \n",
+ " 46.685594 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 1 \n",
+ " 0.944766 \n",
+ " 2.871801 \n",
+ " 36.559290 \n",
+ " 20.974156 \n",
+ " 24.692165 \n",
+ " 84.254376 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 2 \n",
+ " 4.620137 \n",
+ " 2.464522 \n",
+ " 109.678133 \n",
+ " 6.572943 \n",
+ " 6.218167 \n",
+ " 41.284482 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 3 \n",
+ " 2.993973 \n",
+ " 1.165587 \n",
+ " 41.289861 \n",
+ " 18.726870 \n",
+ " 5.382738 \n",
+ " 42.412421 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 4 \n",
+ " 1.476776 \n",
+ " 0.990304 \n",
+ " 12.646279 \n",
+ " 28.490769 \n",
+ " 13.393807 \n",
+ " 58.474973 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ "
\n",
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " x1 \n",
+ " x2 \n",
+ " c1_pred \n",
+ " c2_pred \n",
+ " f1 \n",
+ " f2 \n",
+ " c1_sd \n",
+ " c2_sd \n",
+ " f1_sd \n",
+ " f2_sd \n",
+ " f1_des \n",
+ " f2_des \n",
+ " c1_des \n",
+ " c2_des \n",
+ " valid_f1 \n",
+ " valid_f2 \n",
+ " \n",
+ " \n",
+ " \n",
+ " \n",
+ " 0 \n",
+ " 4.824193 \n",
+ " 1.418243 \n",
+ " 2.936403 \n",
+ " 32.164223 \n",
+ " 96.224058 \n",
+ " 12.813320 \n",
+ " 0.878153 \n",
+ " 1.507216 \n",
+ " 3.744860 \n",
+ " 1.108512 \n",
+ " -96.224058 \n",
+ " -12.813320 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 1 \n",
+ " 2.406195 \n",
+ " 1.414166 \n",
+ " 8.614710 \n",
+ " 50.625390 \n",
+ " 30.881127 \n",
+ " 19.762983 \n",
+ " 0.433521 \n",
+ " 0.821362 \n",
+ " 1.621519 \n",
+ " 0.591357 \n",
+ " -30.881127 \n",
+ " -19.762983 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 2 \n",
+ " 1.286647 \n",
+ " 1.204312 \n",
+ " 15.428966 \n",
+ " 63.005427 \n",
+ " 12.103810 \n",
+ " 28.668753 \n",
+ " 0.456731 \n",
+ " 0.867202 \n",
+ " 1.669497 \n",
+ " 0.625886 \n",
+ " -12.103810 \n",
+ " -28.668753 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 3 \n",
+ " 2.429053 \n",
+ " 2.916580 \n",
+ " 14.891103 \n",
+ " 65.707672 \n",
+ " 55.993033 \n",
+ " 10.877458 \n",
+ " 0.818691 \n",
+ " 1.394782 \n",
+ " 3.607390 \n",
+ " 1.067341 \n",
+ " -55.993033 \n",
+ " -10.877458 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ " 4 \n",
+ " 2.967768 \n",
+ " 1.533018 \n",
+ " 6.255027 \n",
+ " 45.460447 \n",
+ " 44.682814 \n",
+ " 16.057206 \n",
+ " 0.421102 \n",
+ " 0.801401 \n",
+ " 1.540146 \n",
+ " 0.574769 \n",
+ " -44.682814 \n",
+ " -16.057206 \n",
+ " 1.0 \n",
+ " 1.0 \n",
+ " 1 \n",
+ " 1 \n",
+ " \n",
+ " \n",
+ "
\n",
+ "
"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
+ "source": [
+ "true_front = get_pareto_front(\n",
+ " domain=bnh.domain, experiments=sample(bnh, Nsamples=NUM_SAMPLES)\n",
+ ")\n",
+ "\n",
+ "fig, ax = plt.subplots()\n",
+ "ax.plot(\n",
+ " dense_samples.f1,\n",
+ " dense_samples.f2,\n",
+ " \"o\",\n",
+ " alpha=0.2,\n",
+ " label=\"MOBO Predictions (random inputs)\",\n",
+ " color=\"green\",\n",
+ ")\n",
+ "ax.plot(\n",
+ " estimated_front.f1, estimated_front.f2, \"o\", label=\"MOBO Pareto front\", color=\"blue\"\n",
+ ")\n",
+ "ax.plot(\n",
+ " true_front.f1,\n",
+ " true_front.f2,\n",
+ " \"o\",\n",
+ " markersize=3,\n",
+ " label=\"True front (non-dominated random samples)\",\n",
+ " color=\"red\",\n",
+ ")\n",
+ "ax.plot(\n",
+ " init_data.f1, init_data.f2, \"o\", label=\"Initial samples for MOBO\", color=\"black\"\n",
+ ")\n",
+ "ax.set_xlabel(\"f1\")\n",
+ "ax.set_ylabel(\"f2\")\n",
+ "ax.legend()\n",
+ "plt.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": "3.10.12"
+ },
+ "papermill": {
+ "default_parameters": {},
+ "duration": 5.249527,
+ "end_time": "2024-10-10T20:35:38.481859",
+ "environment_variables": {},
+ "exception": true,
+ "parameters": {},
+ "start_time": "2024-10-10T20:35:33.232332",
+ "version": "2.6.0"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
From 7b7f7e3db8db8a9d10aebcf5f883d5e9bddc70f4 Mon Sep 17 00:00:00 2001
From: Robert Lee <84771576+R-M-Lee@users.noreply.github.com>
Date: Tue, 1 Apr 2025 12:45:14 +0200
Subject: [PATCH 3/3] remove bad symlink
---
docs/pareto_front_naive_estimation.ipynb | 607 -----------------------
1 file changed, 607 deletions(-)
delete mode 100644 docs/pareto_front_naive_estimation.ipynb
diff --git a/docs/pareto_front_naive_estimation.ipynb b/docs/pareto_front_naive_estimation.ipynb
deleted file mode 100644
index cbe0ec5a3..000000000
--- a/docs/pareto_front_naive_estimation.ipynb
+++ /dev/null
@@ -1,607 +0,0 @@
-{
- "cells": [
- {
- "attachments": {},
- "cell_type": "markdown",
- "id": "0",
- "metadata": {
- "papermill": {
- "duration": 0.005215,
- "end_time": "2024-10-10T20:35:34.138342",
- "exception": false,
- "start_time": "2024-10-10T20:35:34.133127",
- "status": "completed"
- },
- "tags": []
- },
- "source": [
- "# Estimating Pareto fronts using random samples\n",
- "\n",
- "## Imports"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "id": "1",
- "metadata": {
- "papermill": {
- "duration": 3.323246,
- "end_time": "2024-10-10T20:35:37.465135",
- "exception": false,
- "start_time": "2024-10-10T20:35:34.141889",
- "status": "completed"
- },
- "tags": []
- },
- "outputs": [],
- "source": [
- "import os\n",
- "\n",
- "import matplotlib.pyplot as plt\n",
- "import pandas as pd\n",
- "\n",
- "import bofire.strategies.api as strategies\n",
- "from bofire.benchmarks.api import BNH, Benchmark\n",
- "from bofire.data_models.strategies.api import MoboStrategy, RandomStrategy\n",
- "from bofire.utils.multiobjective import get_pareto_front\n",
- "\n",
- "\n",
- "SMOKE_TEST = os.environ.get(\"SMOKE_TEST\")\n",
- "NUM_INIT_SAMPLES = 12 if not SMOKE_TEST else 20\n",
- "NUM_SAMPLES = 10000 if not SMOKE_TEST else 100"
- ]
- },
- {
- "attachments": {},
- "cell_type": "markdown",
- "id": "2",
- "metadata": {
- "papermill": {
- "duration": 0.001315,
- "end_time": "2024-10-10T20:35:37.468031",
- "exception": false,
- "start_time": "2024-10-10T20:35:37.466716",
- "status": "completed"
- },
- "tags": []
- },
- "source": [
- "## Introduction\n",
- "\n",
- "In this notebook, we will take a simple multiobjective example and train surrogate models for each output. Then we will use a simple random sampling approach to estimate the Pareto front, which is the set of best possible compromises between the objectives. Here we will not do any Bayesian optimization iterations; we merely demonstrate the random sampling approach for Pareto fronts, which can provide a baseline for Pareto-front estimation methods.\n",
- "\n",
- "You can compare this naive approach with the more sophisticated method in, e.g., the [tutorial for the ZDT1 benchmark](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/011-ZDT1.ipynb). The random-sampling approach in this notebook will not work well for higher-dimensional problems because we are just hoping that some random samples will fall near the Pareto front and need to be extremely lucky for this to be the case when there are many dimensions.\n",
- "\n",
- "## Generating the initial data for modeling\n",
- "\n",
- "This function gives us a random dataset (including the 'measured' outputs) for the benchmark BNH, which has its own [tutorial](https://github.com/experimental-design/bofire/blob/main/tutorials/benchmarks/012-BNH.ipynb)."
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "id": "3",
- "metadata": {
- "papermill": {
- "duration": 0.18909,
- "end_time": "2024-10-10T20:35:37.658331",
- "exception": true,
- "start_time": "2024-10-10T20:35:37.469241",
- "status": "failed"
- },
- "tags": []
- },
- "outputs": [],
- "source": [
- "def sample(benchmark: Benchmark, Nsamples: int = 10, inputs_only=False):\n",
- " datamodel = RandomStrategy(domain=benchmark.domain)\n",
- " sampler = strategies.map(data_model=datamodel)\n",
- " samples = sampler.ask(Nsamples)\n",
- " return samples if inputs_only else benchmark.f(samples, return_complete=True)"
- ]
- },
- {
- "cell_type": "markdown",
- "id": "4",
- "metadata": {},
- "source": [
- "Let's use the function to generate the dataset that we will use for modeling:"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "id": "5",
- "metadata": {},
- "outputs": [
- {
- "data": {
- "text/html": [
- "\n",
- "\n",
- "
\n",
- " \n",
- " \n",
- " x1 \n",
- " x2 \n",
- " f1 \n",
- " f2 \n",
- " c1 \n",
- " c2 \n",
- " valid_c1 \n",
- " valid_c2 \n",
- " valid_f1 \n",
- " valid_f2 \n",
- " \n",
- " \n",
- " \n",
- " \n",
- " 0 \n",
- " 3.235181 \n",
- " 1.897152 \n",
- " 56.262340 \n",
- " 12.742250 \n",
- " 6.713771 \n",
- " 46.685594 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 1 \n",
- " 0.944766 \n",
- " 2.871801 \n",
- " 36.559290 \n",
- " 20.974156 \n",
- " 24.692165 \n",
- " 84.254376 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 2 \n",
- " 4.620137 \n",
- " 2.464522 \n",
- " 109.678133 \n",
- " 6.572943 \n",
- " 6.218167 \n",
- " 41.284482 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 3 \n",
- " 2.993973 \n",
- " 1.165587 \n",
- " 41.289861 \n",
- " 18.726870 \n",
- " 5.382738 \n",
- " 42.412421 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 4 \n",
- " 1.476776 \n",
- " 0.990304 \n",
- " 12.646279 \n",
- " 28.490769 \n",
- " 13.393807 \n",
- " 58.474973 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- "
\n",
- "
\n",
- "\n",
- "
\n",
- " \n",
- " \n",
- " x1 \n",
- " x2 \n",
- " c1_pred \n",
- " c2_pred \n",
- " f1 \n",
- " f2 \n",
- " c1_sd \n",
- " c2_sd \n",
- " f1_sd \n",
- " f2_sd \n",
- " f1_des \n",
- " f2_des \n",
- " c1_des \n",
- " c2_des \n",
- " valid_f1 \n",
- " valid_f2 \n",
- " \n",
- " \n",
- " \n",
- " \n",
- " 0 \n",
- " 4.824193 \n",
- " 1.418243 \n",
- " 2.936403 \n",
- " 32.164223 \n",
- " 96.224058 \n",
- " 12.813320 \n",
- " 0.878153 \n",
- " 1.507216 \n",
- " 3.744860 \n",
- " 1.108512 \n",
- " -96.224058 \n",
- " -12.813320 \n",
- " 1.0 \n",
- " 1.0 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 1 \n",
- " 2.406195 \n",
- " 1.414166 \n",
- " 8.614710 \n",
- " 50.625390 \n",
- " 30.881127 \n",
- " 19.762983 \n",
- " 0.433521 \n",
- " 0.821362 \n",
- " 1.621519 \n",
- " 0.591357 \n",
- " -30.881127 \n",
- " -19.762983 \n",
- " 1.0 \n",
- " 1.0 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 2 \n",
- " 1.286647 \n",
- " 1.204312 \n",
- " 15.428966 \n",
- " 63.005427 \n",
- " 12.103810 \n",
- " 28.668753 \n",
- " 0.456731 \n",
- " 0.867202 \n",
- " 1.669497 \n",
- " 0.625886 \n",
- " -12.103810 \n",
- " -28.668753 \n",
- " 1.0 \n",
- " 1.0 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 3 \n",
- " 2.429053 \n",
- " 2.916580 \n",
- " 14.891103 \n",
- " 65.707672 \n",
- " 55.993033 \n",
- " 10.877458 \n",
- " 0.818691 \n",
- " 1.394782 \n",
- " 3.607390 \n",
- " 1.067341 \n",
- " -55.993033 \n",
- " -10.877458 \n",
- " 1.0 \n",
- " 1.0 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- " 4 \n",
- " 2.967768 \n",
- " 1.533018 \n",
- " 6.255027 \n",
- " 45.460447 \n",
- " 44.682814 \n",
- " 16.057206 \n",
- " 0.421102 \n",
- " 0.801401 \n",
- " 1.540146 \n",
- " 0.574769 \n",
- " -44.682814 \n",
- " -16.057206 \n",
- " 1.0 \n",
- " 1.0 \n",
- " 1 \n",
- " 1 \n",
- " \n",
- " \n",
- "
\n",
- "
"
- ]
- },
- "metadata": {},
- "output_type": "display_data"
- }
- ],
- "source": [
- "true_front = get_pareto_front(\n",
- " domain=bnh.domain, experiments=sample(bnh, Nsamples=NUM_SAMPLES)\n",
- ")\n",
- "\n",
- "fig, ax = plt.subplots()\n",
- "ax.plot(\n",
- " dense_samples.f1,\n",
- " dense_samples.f2,\n",
- " \"o\",\n",
- " alpha=0.2,\n",
- " label=\"MOBO Predictions (random inputs)\",\n",
- " color=\"green\",\n",
- ")\n",
- "ax.plot(\n",
- " estimated_front.f1, estimated_front.f2, \"o\", label=\"MOBO Pareto front\", color=\"blue\"\n",
- ")\n",
- "ax.plot(\n",
- " true_front.f1,\n",
- " true_front.f2,\n",
- " \"o\",\n",
- " markersize=3,\n",
- " label=\"True front (non-dominated random samples)\",\n",
- " color=\"red\",\n",
- ")\n",
- "ax.plot(\n",
- " init_data.f1, init_data.f2, \"o\", label=\"Initial samples for MOBO\", color=\"black\"\n",
- ")\n",
- "ax.set_xlabel(\"f1\")\n",
- "ax.set_ylabel(\"f2\")\n",
- "ax.legend()\n",
- "plt.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": "3.10.12"
- },
- "papermill": {
- "default_parameters": {},
- "duration": 5.249527,
- "end_time": "2024-10-10T20:35:38.481859",
- "environment_variables": {},
- "exception": true,
- "parameters": {},
- "start_time": "2024-10-10T20:35:33.232332",
- "version": "2.6.0"
- }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}