Skip to content

Commit

Permalink
tests updtaed and examples added
Browse files Browse the repository at this point in the history
  • Loading branch information
@MarcusMNoack
MarcusMNoack committed Jan 28, 2022
1 parent cbf7376 commit 5a31c33
Show file tree
Hide file tree
Showing 6 changed files with 727 additions and 31 deletions.
240 changes: 240 additions & 0 deletions examples/.ipynb_checkpoints/multi_task_test_notebook-checkpoint.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,240 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "ff0ce27e",
"metadata": {},
"source": [
"# FVGP MULTI Task Notebook\n",
"In this notebook we will go through many features of FVGP. We will be primarily concerned with regression over a single dimension output and multiple tasks."
]
},
{
"cell_type": "markdown",
"id": "346f9628",
"metadata": {},
"source": [
"## This first cell has nothing to do with gpCAM, it's just a function to plot some results later"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "f17458ce",
"metadata": {},
"outputs": [],
"source": [
"import plotly.graph_objects as go\n",
"import numpy as np\n",
"def plot(x,y,z,data = None):\n",
" fig = go.Figure()\n",
" fig.add_trace(go.Surface(x = x, y = y,z=z))\n",
" if data is not None: \n",
" fig.add_trace(go.Scatter3d(x=data[:,0], y=data[:,1], z=data[:,2],\n",
" mode='markers'))\n",
"\n",
" fig.update_layout(title='Posterior Mean', autosize=True,\n",
" width=800, height=800,\n",
" margin=dict(l=65, r=50, b=65, t=90))\n",
"\n",
"\n",
" fig.show()"
]
},
{
"cell_type": "markdown",
"id": "82296522",
"metadata": {},
"source": [
"## Import fvgp and relevant libraries"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3cc21dc8",
"metadata": {},
"outputs": [],
"source": [
"import fvgp\n",
"from fvgp import gp, fvgp\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "markdown",
"id": "87ae086d",
"metadata": {},
"source": [
"## Defining some input data and testing points"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "31d4da07",
"metadata": {},
"outputs": [],
"source": [
"def test_data_function(x):\n",
" data_1 = 100*np.sin(x)+np.cos(x)\n",
" data_2 = 5*np.ones(x.shape)\n",
" data_3 = 1*np.cos(x/10 + 5)\n",
" data_4 = 5*np.sin(x/200)\n",
" data_5 = 10*np.cos(x)\n",
" \n",
" return np.column_stack((data_1, data_2, data_3, data_4, data_5))\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d3b9ede4",
"metadata": {},
"outputs": [],
"source": [
"x_input = np.linspace(-2*np.pi, 10*np.pi,100)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "4dda18dd",
"metadata": {},
"outputs": [],
"source": [
"y_output = test_data_function(x_input)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "e7ff3403",
"metadata": {},
"outputs": [],
"source": [
"x_input_test = np.linspace(3*np.pi, 4*np.pi, 100)"
]
},
{
"cell_type": "markdown",
"id": "2b0cc16d",
"metadata": {},
"source": [
"## Setting up the fvgp multi task object\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a2874db0",
"metadata": {},
"outputs": [],
"source": [
"obj = fvgp.fvGP(input_space_dim = 1, \n",
" output_space_dim = 1, \n",
" output_number = 5, \n",
" points = x_input.reshape(-1,1),\n",
" values = y_output,\n",
" init_hyperparameters = np.array([10,10,10]))"
]
},
{
"cell_type": "markdown",
"id": "ca3686f5",
"metadata": {},
"source": [
"## Training our gaussian process regression on given data\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "75dcee32",
"metadata": {
"scrolled": false
},
"outputs": [],
"source": [
"hyper_param_bounds = np.array([[0.0001, 1000],[ 0.0001, 1000],[ 0.0001, 1000]])\n",
"obj.train(hyper_param_bounds)"
]
},
{
"cell_type": "markdown",
"id": "3b143d85",
"metadata": {},
"source": [
"## Looking at the posterior mean at the test points (remember that we did not define a particularly good kernel)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1acce22c",
"metadata": {},
"outputs": [],
"source": [
"task_idx = 1"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "6a2d2781",
"metadata": {},
"outputs": [],
"source": [
"x_linspace = np.linspace(3*np.pi, 4*np.pi,100)\n",
"y_linspace = np.linspace(0,4,100)\n",
"x_grid, y_grid = np.meshgrid(x_linspace, y_linspace)\n",
"post_mean= obj.posterior_mean(np.column_stack((x_input_test, task_idx*np.ones(x_input_test.shape))))\n",
"posterior_mean_data = obj.posterior_mean(np.column_stack((x_grid.flatten(), y_grid.flatten())))\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9e7f6b42",
"metadata": {},
"outputs": [],
"source": [
"fig = plt.figure(figsize = (10,10))\n",
"ax = fig.add_subplot(projection='3d')\n",
"\n",
"\n",
"ax.scatter(post_mean['x'][:,0], np.ones(100)*0, y_output[:,0])\n",
"ax.scatter(post_mean['x'][:,0], np.ones(100)*1,y_output[:,1])\n",
"ax.scatter(post_mean['x'][:,0], np.ones(100)*2,y_output[:,2])\n",
"ax.scatter(post_mean['x'][:,0], np.ones(100)*3,y_output[:,3])\n",
"ax.scatter(post_mean['x'][:,0], np.ones(100)*4,y_output[:,4])\n",
"ax.plot_surface(x_grid, y_grid, posterior_mean_data['f(x)'].reshape(100,100), rstride=1, cstride=1,cmap='inferno')\n",
"# ax.scatter(x_grid.flatten(), y_grid.flatten(), posterior_mean_data['f(x)'])\n",
"\n",
"# pyplot.pcolormesh(x_grid, y_grid, posterior_mean_data['f(x)'].reshape(100,100))\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "fvgp_env",
"language": "python",
"name": "fvgp_env"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.10"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

0 comments on commit 5a31c33

Please sign in to comment.