-
Notifications
You must be signed in to change notification settings - Fork 0
/
paper_1.html
381 lines (276 loc) · 16.6 KB
/
paper_1.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
<!DOCTYPE html>
<html lang="en">
<head>
<!-- Required meta tags -->
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no" />
<!-- External Javascript libs_ext -->
<script src="https://cdn.jsdelivr.net/npm/d3@5/dist/d3.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/handlebars@4.7.3/dist/handlebars.min.js"
integrity="sha256-/PJBs6QWvXijOFIX04kZpLb6ZtSQckdOIavLWKKOgXU=" crossorigin="anonymous"></script>
<script src="https://code.jquery.com/jquery-3.6.0.min.js"
integrity="sha256-/xUj+3OJU5yExlq6GSYGSHk7tPXikynS7ogEvDej/m4=" crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/npm/popper.js@1.16.0/dist/umd/popper.min.js"
integrity="sha384-Q6E9RHvbIyZFJoft+2mJbHaEWldlvI9IOYy5n3zV9zzTtmI3UksdQRVvoxMfooAo"
crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/npm/bootstrap@4.4.1/dist/js/bootstrap.min.js"
integrity="sha256-WqU1JavFxSAMcLP2WIOI+GB2zWmShMI82mTpLDcqFUg=" crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/npm/moment@2.24.0/min/moment.min.js"
integrity="sha256-4iQZ6BVL4qNKlQ27TExEhBN1HFPvAvAMbFavKKosSWQ=" crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/npm/moment-timezone@0.5.28/builds/moment-timezone-with-data.min.js"
integrity="sha256-IWYg4uIC8/erItNXYvLtyYHioRi2zT1TFva8qaAU/ww=" crossorigin="anonymous"></script>
<!-- Library libs_ext -->
<script src="static/js/libs_ext/typeahead.bundle.js"></script>
<script src="static/js/tagEditor/jquery.caret.min.js"></script>
<script src="static/js/tagEditor/jquery.tag-editor.min.js"></script>
<!-- Internal Libs -->
<script src="static/js/data/api.js"></script>
<!-- External CSS -->
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bootstrap@4.3.1/dist/css/bootstrap.min.css"
integrity="sha256-YLGeXaapI0/5IgZopewRJcFXomhRMlYYjugPLSyNjTY=" crossorigin="anonymous">
<!-- External Fonts (no google for china) -->
<link href="static/css/Lato.css" rel="stylesheet" />
<link href="static/css/Exo.css" rel="stylesheet" />
<link href="static/css/Cuprum.css" rel="stylesheet" />
<link rel="stylesheet" href="static/css/main.css" />
<!-- <link rel="stylesheet" href="static/css/fa_regular.css"/>-->
<link rel="stylesheet" href="static/css/fa_solid.css" />
<link rel="stylesheet" href="static/css/lazy_load.css" />
<link rel="stylesheet" href="static/css/typeahead.css" />
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/daterangepicker/daterangepicker.min.js"></script>
<link rel="stylesheet" type="text/css" href="https://cdn.jsdelivr.net/npm/daterangepicker/daterangepicker.css" />
<title>Machine Learning Simulation: Learning to Simulate Complex Physics with Graph Networks</title>
<link rel="icon" type="image/png" href="favicon.ico">
<!-- Google Analytics -->
<!-- <script async src="https://www.googletagmanager.com/gtag/js?id=G-GT7LEVW06Z"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() { dataLayer.push(arguments); }
gtag('js', new Date());
gtag('config', 'G-GT7LEVW06Z');
</script> -->
<!-- head block ends here-->
<meta name="citation_title" content="Learning to Simulate Complex Physics with Graph Networks" />
<meta name="citation_author" content="Alvaro Sanchez-Gonzalez" />
<meta name="citation_author" content="Jonathan Godwin" />
<meta name="citation_author" content="Tobias Pfaff" />
<meta name="citation_author" content="Rex Ying" />
<meta name="citation_author" content="Jure Leskovec" />
<meta name="citation_author" content="Peter W. Battaglia" />
<meta name="citation_abstract" content="Here we present a machine learning framework and model implementation that can learn to simulate a wide variety of challenging physical domains, involving fluids, rigid solids, and deformable materials interacting with one another. Our framework---which we term Graph Network-based Simulators (GNS)---represents the state of a physical system with particles, expressed as nodes in a graph, and computes dynamics via learned message-passing. Our results show that our model can generalize from single-timestep predictions with thousands of particles during training, to different initial conditions, thousands of timesteps, and at least an order of magnitude more particles at test time. Our model was robust to hyperparameter choices across various evaluation metrics: the main determinants of long-term performance were the number of message-passing steps, and mitigating the accumulation of error by corrupting the training data with noise. Our GNS framework advances the state-of-the-art in learned physical simulation, and holds promise for solving a wide range of complex forward and inverse problems." />
<meta name="citation_keywords" content="Graph Neural Networks" />
<meta name="citation_keywords" content="forward simulation" />
<meta name="citation_keywords" content="GNS" />
<meta name="citation_keywords" content="GNNs" />
<meta name="citation_keywords" content="message passing" />
<meta name="citation_keywords" content="encoder-processor-decoder" />
<meta name="citation_keywords" content="physical simulation" />
<meta name="citation_pdf_url" content="https://arxiv.org/pdf/2002.09405.pdf" />
<script type="text/javascript" src="static/js/vis/vis-network.min.js"></script>
<style>
.red-hyper-link {
color: #ED1C24 !important;
}
</style>
</head>
<body>
<!-- NAV -->
<!-- add the below items to the "navigation_bar" list if those pages are needed; can't comment those out b/c jinja doesn't support comment
('index.html', 'Home'),
('papers.html', 'Advanced Search'),
('add_paper.html', 'Add Paper'),
('join_us.html', 'Join Us'),
('faq.html', 'FAQ')
-->
<nav class="navbar sticky-top navbar-expand-lg navbar-light" style="background-color: #EFECE5;" id="main-nav">
<div class="container">
<a class="navbar-brand" href="index.html">
<img class="logo" src="static/images/mls_logo_withfont.svg" style="width: 250px;"/>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-target="#navbarNav"
aria-controls="navbarNav" aria-expanded="false" aria-label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>
<div class="collapse navbar-collapse text-right flex-grow-1" id="navbarNav">
<ul class="navbar-nav ml-auto">
<li class="nav-item ">
<a class="nav-link" href="index.html" style="font-family: 'Ubuntu', sans-serif;">Home</a>
</li>
<li class="nav-item ">
<a class="nav-link" href="add_paper.html" style="font-family: 'Ubuntu', sans-serif;">Add Paper</a>
</li>
</ul>
</div>
</div>
</nav>
<!-- User Overrides -->
<!-- top block ends here-->
<div class="container">
<!-- Tabs -->
<div class="tabs">
</div>
<!-- Content -->
<div class="content">
<!-- Title -->
<div class="pp-card m-3">
<div class="card-header">
<h2 class="card-title main-title text-center" style="color: black;">
Learning to Simulate Complex Physics with Graph Networks
</h2>
<h3 class="card-subtitle mb-2 text-muted text-center">
<a href="papers.html?author=Alvaro Sanchez-Gonzalez" target="_blank"
data-tippy-content="See all papers authored by Alvaro Sanchez-Gonzalez"
class="text-muted filterByAuthorLink">Alvaro Sanchez-Gonzalez</a>,
<a href="papers.html?author=Jonathan Godwin" target="_blank"
data-tippy-content="See all papers authored by Jonathan Godwin"
class="text-muted filterByAuthorLink">Jonathan Godwin</a>,
<a href="papers.html?author=Tobias Pfaff" target="_blank"
data-tippy-content="See all papers authored by Tobias Pfaff"
class="text-muted filterByAuthorLink">Tobias Pfaff</a>,
<a href="papers.html?author=Rex Ying" target="_blank"
data-tippy-content="See all papers authored by Rex Ying"
class="text-muted filterByAuthorLink">Rex Ying</a>,
<a href="papers.html?author=Jure Leskovec" target="_blank"
data-tippy-content="See all papers authored by Jure Leskovec"
class="text-muted filterByAuthorLink">Jure Leskovec</a>,
<a href="papers.html?author=Peter W. Battaglia" target="_blank"
data-tippy-content="See all papers authored by Peter W. Battaglia"
class="text-muted filterByAuthorLink">Peter W. Battaglia</a>
</h3>
<h3 class="card-subtitle mb-2 text-muted text-center">
14/9/2020
</h3>
<div id="citation-code-wrapper" class="text-muted text-center"
>
</div>
<p class="text-center" style="margin-bottom: 0px;">
<span>Keywords:</span>
<a href="papers.html?keyword=Graph Neural Networks" target="_blank"
data-tippy-content="See all papers with keyword 'Graph Neural Networks'"
class="text-secondary text-decoration-none filterByKeywordLink">Graph Neural Networks</a>,
<a href="papers.html?keyword=forward simulation" target="_blank"
data-tippy-content="See all papers with keyword 'forward simulation'"
class="text-secondary text-decoration-none filterByKeywordLink">forward simulation</a>,
<a href="papers.html?keyword=GNS" target="_blank"
data-tippy-content="See all papers with keyword 'GNS'"
class="text-secondary text-decoration-none filterByKeywordLink">GNS</a>,
<a href="papers.html?keyword=GNNs" target="_blank"
data-tippy-content="See all papers with keyword 'GNNs'"
class="text-secondary text-decoration-none filterByKeywordLink">GNNs</a>,
<a href="papers.html?keyword=message passing" target="_blank"
data-tippy-content="See all papers with keyword 'message passing'"
class="text-secondary text-decoration-none filterByKeywordLink">message passing</a>,
<a href="papers.html?keyword=encoder-processor-decoder" target="_blank"
data-tippy-content="See all papers with keyword 'encoder-processor-decoder'"
class="text-secondary text-decoration-none filterByKeywordLink">encoder-processor-decoder</a>,
<a href="papers.html?keyword=physical simulation" target="_blank"
data-tippy-content="See all papers with keyword 'physical simulation'"
class="text-secondary text-decoration-none filterByKeywordLink">physical simulation</a>
</p>
<p class="text-center" style="margin-bottom: 0px;">
<span>Venue: </span>
<a href="papers.html?venue=ICML" target="_blank" class="text-secondary text-decoration-none">ICML 2020</a>
</p>
<div class="text-center p-3">
<a class="card-link red-hyper-link" target="_blank" href="https://arxiv.org/pdf/2002.09405.pdf">
Paper
</a>
<a class="card-link red-hyper-link" data-toggle="collapse" role="button" href="#citation">
Citation
</a>
</div>
</div>
<span id="invisible-paper-id" style="display: none;">1</span>
</div>
<div id="citation" class="pp-card m-3 collapse">
<div class="card-body">
<div class="card-text">
<span style="font-size: large; font-weight: bold;">Bibtex:</span>
<span style="white-space: pre-line; position: relative; left: 20px;">
@inproceedings{DBLP:conf/icml/Sanchez-Gonzalez20,
author = {Alvaro Sanchez{-}Gonzalez and
Jonathan Godwin and
Tobias Pfaff and
Rex Ying and
Jure Leskovec and
Peter W. Battaglia},
title = {Learning to Simulate Complex Physics with Graph Networks},
booktitle = {Proceedings of the 37th International Conference on Machine Learning,
{ICML} 2020, 13-18 July 2020, Virtual Event},
series = {Proceedings of Machine Learning Research},
volume = {119},
pages = {8459--8468},
publisher = {{PMLR}},
year = {2020},
url = {http://proceedings.mlr.press/v119/sanchez-gonzalez20a.html},
timestamp = {Tue, 15 Dec 2020 17:40:19 +0100},
biburl = {https://dblp.org/rec/conf/icml/Sanchez-Gonzalez20.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
</span>
</div>
<p></p>
</div>
</div>
<div id="details" class="pp-card m-3">
<div class="card-body">
<div class="card-text">
<div id="abstractExample">
<p style="font-weight: bolder; font-size: 25px; text-align: center;">Abstract</p>
Here we present a machine learning framework and model implementation that can learn to simulate a wide variety of challenging physical domains, involving fluids, rigid solids, and deformable materials interacting with one another. Our framework---which we term Graph Network-based Simulators (GNS)---represents the state of a physical system with particles, expressed as nodes in a graph, and computes dynamics via learned message-passing. Our results show that our model can generalize from single-timestep predictions with thousands of particles during training, to different initial conditions, thousands of timesteps, and at least an order of magnitude more particles at test time. Our model was robust to hyperparameter choices across various evaluation metrics: the main determinants of long-term performance were the number of message-passing steps, and mitigating the accumulation of error by corrupting the training data with noise. Our GNS framework advances the state-of-the-art in learned physical simulation, and holds promise for solving a wide range of complex forward and inverse problems.
</div>
</div>
<p></p>
</div>
</div>
<!-- Youtube Video -->
<!-- Project Webpage -->
<!-- Citation Graph -->
<!-- <div class="border-top my-3"></div>
<div class="row p-4">
<div class="col-12" style="text-align: center;">
<span style="font-size: 25px;">Citation Graph</span> <br>
<span>(Double click on nodes to open corresponding papers' pages)</span>
</div>
</div>
<div id="citationGraph" style="width: 100%;
height: 600px;
border: 1px solid lightgray;" ondblclick="openPaperLink()"></div>
<p><span style="font-size: 10px;">*</span> Showing citation graph for papers within our database. Data retrieved from <a href="https://www.semanticscholar.org/search?q=Learning to Simulate Complex Physics with Graph Networks&sort=relevance">Semantic Scholar</a>. For full citation graphs, visit <a href="https://www.connectedpapers.com/search?q=Learning to Simulate Complex Physics with Graph Networks">ConnectedPapers</a>.</p>
<script src="https://cdn.jsdelivr.net/npm/@popperjs/core@2.4.0/dist/umd/popper.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/tippy.js@6/dist/tippy-bundle.umd.min.js"></script>
<script src="static/js/views/paper_detail.js"></script>
<script>
tippy(".filterByAuthorLink");
tippy(".filterByKeywordLink");
const paperID = document.getElementById("invisible-paper-id").innerHTML;
drawCitationGraphAndGenerateCitationCode(paperID);
</script>
-->
</div>
</div>
<!-- body block ends here-->
<!-- Footer -->
<footer class="footer"
style="background-color: #353535; width: 100%; height: 4vh; display: flex; justify-content: center; align-items: center; ">
<!-- no contents needed -->
</footer>
<!-- Code for hash tags -->
<script type="text/javascript">
$(document).ready(function () {
if (window.location.hash !== "") {
$(`a[href="${window.location.hash}"]`).tab("show");
}
$("a[data-toggle='tab']").on("shown.bs.tab", function (e) {
const hash = $(e.target).attr("href");
if (hash.substr(0, 1) === "#") {
const position = $(window).scrollTop();
window.location.replace(`#${hash.substr(1)}`);
$(window).scrollTop(position);
}
});
});
</script>
<!-- footer block ends here -->
</body>
</html>