CINXE.COM

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">     <html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en" xmlns:og="http://ogp.me/ns#" xmlns:fb="http://ogp.me/ns/fb#"> <head> <title>Using Machine Learning to Improve Dynamic Aperture Estimates - CERN Document Server</title> <link href='https://framework.web.cern.ch/framework/2.0/fonts/PTSansWeb/PTSansWeb.css' rel='stylesheet' type='text/css' /> <link rel="stylesheet" href="http://cds.cern.ch/img/invenio.css?v=20141127" type="text/css" /> <link rel="stylesheet" href="http://cds.cern.ch/img/cern_theme/css/cern_theme.css?v=20141127" type="text/css" /> <link rel="stylesheet"href="/css/font-awesome.min.css"> <meta http-equiv="X-UA-Compatible" content="IE=Edge"/> <link rel="stylesheet" href="http://cds.cern.ch/img/cern_toolbar/css/toolbar.css" type="text/css" />    <link rel="canonical" href="http://cds.cern.ch/record/2804876" /> <link rel="alternate" hreflang="el" href="http://cds.cern.ch/record/2804876?ln=el" /> <link rel="alternate" hreflang="fr" href="http://cds.cern.ch/record/2804876?ln=fr" /> <link rel="alternate" hreflang="bg" href="http://cds.cern.ch/record/2804876?ln=bg" /> <link rel="alternate" hreflang="zh-TW" href="http://cds.cern.ch/record/2804876?ln=zh_TW" /> <link rel="alternate" hreflang="pt" href="http://cds.cern.ch/record/2804876?ln=pt" /> <link rel="alternate" hreflang="no" href="http://cds.cern.ch/record/2804876?ln=no" /> <link rel="alternate" hreflang="hr" href="http://cds.cern.ch/record/2804876?ln=hr" /> <link rel="alternate" hreflang="ca" href="http://cds.cern.ch/record/2804876?ln=ca" /> <link rel="alternate" hreflang="de" href="http://cds.cern.ch/record/2804876?ln=de" /> <link rel="alternate" hreflang="it" href="http://cds.cern.ch/record/2804876?ln=it" /> <link rel="alternate" hreflang="zh-CN" href="http://cds.cern.ch/record/2804876?ln=zh_CN" /> <link rel="alternate" hreflang="sv" href="http://cds.cern.ch/record/2804876?ln=sv" /> <link rel="alternate" hreflang="sk" href="http://cds.cern.ch/record/2804876?ln=sk" /> <link rel="alternate" hreflang="en" href="http://cds.cern.ch/record/2804876?ln=en" /> <link rel="alternate" hreflang="pl" href="http://cds.cern.ch/record/2804876?ln=pl" /> <link rel="alternate" hreflang="ru" href="http://cds.cern.ch/record/2804876?ln=ru" /> <link rel="alternate" hreflang="ka" href="http://cds.cern.ch/record/2804876?ln=ka" /> <link rel="alternate" hreflang="ja" href="http://cds.cern.ch/record/2804876?ln=ja" /> <link rel="alternate" hreflang="es" href="http://cds.cern.ch/record/2804876?ln=es" /> <link rel="alternate" type="application/rss+xml" title="CERN Document Server RSS" href="/rss?ln=en" /> <link rel="search" type="application/opensearchdescription+xml" href="http://cds.cern.ch/opensearchdescription" title="CERN Document Server" /> <link rel="unapi-server" type="application/xml" title="unAPI" href="http://cds.cern.ch/unapi" /> <link rel="apple-touch-icon" href="/apple-touch-icon.png"/> <link rel="apple-touch-icon-precomposed" href="/apple-touch-icon-precomposed.png"/> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> <meta http-equiv="Content-Language" content="en" /> <meta name="description" content="The dynamic aperture (DA) is an important concept in the study of nonlinear beam dynamics. Several analytical models used to describe the evolution of DA as a function of time, and to extrapolate to realistic time scales that would not be reachable otherwise due to computational limitations, have been successfully developed. Even though these models have been quite successful in the past, the fitting procedure is rather sensitive to several details. Machine Learning (ML) techniques, which have been around for decades and have matured into powerful tools ever since, carry the potential to address some of these challenges. In this paper, two applications of ML approaches are presented and discussed in detail. Firstly, ML has been used to efficiently detect outliers in the DA computations. Secondly, ML techniques have been applied to improve the fitting procedures of the DA models, thus improving their predictive power. Van der Veken, Frederik F; Giovannozzi, Massimo; Maclean, Ewen H; Montanari, Carlo Emilio; Valentino, Gianluca" /> <meta name="keywords" content="simulation, dynamic-aperture, collider, hadron, operation" /> <script type="text/javascript" src="http://cds.cern.ch/js/jquery.min.js"></script>  <link rel="stylesheet" href="http://cds.cern.ch/img/webnews.css" type="text/css" />  <script type="text/javascript" src="http://cds.cern.ch/js/webnews.js?v=20131009"></script> <meta property="fb:app_id" content="137353533001720"/> <script type="text/x-mathjax-config"> MathJax.Hub.Config({ tex2jax: {inlineMath: [['$','$']], processEscapes: true}, showProcessingMessages: false, messageStyle: "none" }); </script> <script src="/MathJax/MathJax.js?config=TeX-AMS_CHTML" type="text/javascript"> </script>  <meta content="JACoW : Using Machine Learning to Improve Dynamic Aperture Estimates" name="citation_title" /> <meta content="Van der Veken, Frederik F" name="citation_author" /> <meta content="Giovannozzi, Massimo" name="citation_author" /> <meta content="Valentino, Gianluca" name="citation_author" /> <meta content="Montanari, Carlo Emilio" name="citation_author" /> <meta content="Maclean, Ewen H" name="citation_author" /> <meta content="JACoW" name="citation_publisher" /> <meta content="10.18429/JACoW-IPAC2021-MOPAB028" name="citation_doi" /> <meta content="JACoW IPAC" name="citation_journal_title" /> <meta content=" 2021" name="citation_volume" /> <meta content="134-137" name="citation_firstpage" /> <meta content="2021" name="citation_publication_date" /> <meta name="citation_online_date" content="2022/03/25"> <meta content="10.18429/JACoW-IPAC2021-MOPAB028" name="citation_doi" /> <meta name="citation_pdf_url" content="http://cds.cern.ch/record/2804876/files/document.pdf" />  <meta content="USING MACHINE LEARNING TO IMPROVE DYNAMIC APERTURE ESTIMATES" property="og:title" /> <meta content="Using Machine Learning to Improve Dynamic Aperture Estimates" property="og:title" /> <meta content="JACoW" property="og:title" /> <meta content="website" property="og:type" /> <meta content="website" property="og:type" /> <meta content="http://cds.cern.ch/record/2804876" property="og:url" /> <meta content="CERN Document Server" property="og:site_name" /> <meta content="The dynamic aperture (DA) is an important concept in the study of nonlinear beam dynamics. Several analytical models used to describe the evolution of DA as a function of time, and to extrapolate to realistic time scales that would not be reachable otherwise due to computational limitations, have been successfully developed. Even though these models have been quite successful in the past, the fitting procedure is rather sensitive to several details. Machine Learning (ML) techniques, which have been around for decades and have matured into powerful tools ever since, carry the potential to address some of these challenges. In this paper, two applications of ML approaches are presented and discussed in detail. Firstly, ML has been used to efficiently detect outliers in the DA computations. Secondly, ML techniques have been applied to improve the fitting procedures of the DA models, thus improving their predictive power." property="og:description" /> <meta content="JACoW" property="og:description" />  <meta content="summary" name="twitter:card" /> <meta content="summary" name="twitter:card" /> <style></style> </head> <body class="CERN32Document32Server search" lang="en">  <div id="cern-toolbar"> <h1><a href="http://cern.ch" title="CERN">CERN <span>Accelerating science</span></a></h1> <ul> <li class="cern-accountlinks"><a class="cern-account" href="https://cds.cern.ch/youraccount/login?ln=en&referer=http%3A//cds.cern.ch/record/2804876%3Fln%3Den" title="Sign in to your CERN account">Sign in</a></li> <li><a class="cern-directory" href="http://cern.ch/directory" title="Search CERN resources and browse the directory">Directory</a></li> </ul> </div>   <div role="banner" class="clearfix" id="header"> <div class="header-inner inner"> <hgroup class="clearfix"> <h2 id="site-name"> <a rel="home" title="Home" href="/"><span>CERN Document Server</span></a> </h2> <h3 id="site-slogan">Access articles, reports and multimedia content in HEP</h3> </hgroup> <div role="navigation" id="main-navigation" class="cdsmenu"> <h2 class="element-invisible">Main menu</h2><ul class="links inline clearfix"> <li class="menu-386 first active-trail"><a class="active-trail" href="http://cds.cern.ch/?ln=en">Search</a></li> <li class="menu-444 "><a class="" title="" href="http://cds.cern.ch/submit?ln=en">Submit</a></li> <li class="menu-426 "><a class="" href="http://cds.cern.ch/help/?ln=en">Help</a></li> <li class="leaf hassubcdsmenu"> <a hreflang="en" class="header" href="https://cds.cern.ch/youraccount/display?ln=en">Personalize</a> <ul class="subsubcdsmenu"><li><a href="https://cds.cern.ch/youralerts/list?ln=en">Your alerts</a></li><li><a href="https://cds.cern.ch/yourbaskets/display?ln=en">Your baskets</a></li><li><a href="https://cds.cern.ch/yourcomments?ln=en">Your comments</a></li><li><a href="https://cds.cern.ch/youralerts/display?ln=en">Your searches</a></li></ul></li> </ul> </div> </div> </div>  <table class="navtrailbox"> <tr> <td class="navtrailboxbody"> <a href="/?ln=en" class="navtrail">Home</a> > Using Machine Learning to Improve Dynamic Aperture Estimates </td> </tr> </table> </div> <div class="pagebody"><div class="pagebodystripemiddle"> <div class="detailedrecordbox"> <div class="detailedrecordtabs"> <div> <ul class="detailedrecordtabs"><li class="on first"><a href="/record/2804876/?ln=en">Information </a></li><li class=""><a href="/record/2804876/files?ln=en">Files </a></li></ul> <div id="tabsSpacer" style="clear:both;height:0px"> </div></div> </div> <div class="detailedrecordboxcontent"> <div class="top-left-folded"></div> <div class="top-right-folded"></div> <div class="inside">  <abbr class="unapi-id" title="2804876"></abbr> <style type="text/css">  </style> <table class="formatRecordTableFullWidth" > <tr> <td class="formatRecordHeader" style="background-image: url('http://cds.cern.ch/img/journals.jpg');" colspan="2">  Article </td> </tr> <script type="text/javascript"> $( document ).ready(function() { $('.showAuthor').on('click', function() { var author = '<p>' + $(this).data('name') + '</p>'; var affiliation = $(this).data('affiliation') + '</br>'; var contribution = $(this).data('contribution') + '</br>'; $.magnificPopup.open({ items: { src: '<div id="ovelary-mathjax" class="overlay-white oc-content overlay-white-500">' + author + affiliation + contribution + '</div>', type: 'inline' }, callbacks: { open: function() { var div = document.getElementById("overlay-mathjax") MathJax.Hub.Queue(["Typeset", MathJax.Hub, div]); }, } }) }) }); </script> <tr><td class="formatRecordLabel"> Title </td><td style="padding-left:5px;"><b>Using Machine Learning to Improve Dynamic Aperture Estimates</b></td></tr> <tr><td class="formatRecordLabel" style="vertical-align:top;">Related title</td><td style="padding-left:5px; ">USING MACHINE LEARNING TO IMPROVE DYNAMIC APERTURE ESTIMATES<br/></td></tr> <tr><td class="formatRecordLabel"><span style="white-space:nowrap;"> Author(s) </span> </td><td style="padding-left:5px;"><a href="http://cds.cern.ch/search?f=author&p=Van%20der%20Veken%2C%20Frederik%20F&ln=en">Van der Veken, Frederik F</a> (CERN ; Malta U.) ; <a href="http://cds.cern.ch/search?f=author&p=Giovannozzi%2C%20Massimo&ln=en">Giovannozzi, Massimo</a> (CERN) ; <a href="http://cds.cern.ch/search?f=author&p=Maclean%2C%20Ewen%20H&ln=en">Maclean, Ewen H</a> (CERN ; Malta U.) ; <a href="http://cds.cern.ch/search?f=author&p=Montanari%2C%20Carlo%20Emilio&ln=en">Montanari, Carlo Emilio</a> (Bologna U. ; CERN) ; <a href="http://cds.cern.ch/search?f=author&p=Valentino%2C%20Gianluca&ln=en">Valentino, Gianluca</a> (Malta U. ; CERN)</td></tr> <tr><td class="formatRecordLabel"> Publication </td><td style="padding-left:5px;">JACoW, 2021</td></tr> <tr><td class="formatRecordLabel"> Number of pages </td><td style="padding-left:5px;">4</td></tr> <tr><td class="formatRecordLabel"> In: </td><td style="padding-left:5px;"><a href="http://dx.doi.org/10.18429/JACoW-IPAC2021-MOPAB028"><i>JACoW IPAC</i> 2021 (2021) 134-137</a> </a></td></tr> <tr><td class="formatRecordLabel"> In: </td><td style="padding-left:5px;"><a href="http://cds.cern.ch/record/2760128">12th International Particle Accelerator Conference (IPAC 2021)</a>, Online, 24 - 28 May 2021, pp.134-137</td></tr> <tr><td class="formatRecordLabel"> DOI </td><td style="padding-left:5px;"><a href="http://dx.doi.org/10.18429/JACoW-IPAC2021-MOPAB028" title="DOI" target="_blank">10.18429/JACoW-IPAC2021-MOPAB028</a> (publication) <tr><td class="formatRecordLabel"> Subject category </td><td style="padding-left:5px;">Accelerators and Storage Rings</td></tr> <tr><td class="formatRecordLabel"> Abstract </td><td style="padding-left:5px;">The dynamic aperture (DA) is an important concept in the study of nonlinear beam dynamics. Several analytical models used to describe the evolution of DA as a function of time, and to extrapolate to realistic time scales that would not be reachable otherwise due to computational limitations, have been successfully developed. Even though these models have been quite successful in the past, the fitting procedure is rather sensitive to several details. Machine Learning (ML) techniques, which have been around for decades and have matured into powerful tools ever since, carry the potential to address some of these challenges. In this paper, two applications of ML approaches are presented and discussed in detail. Firstly, ML has been used to efficiently detect outliers in the DA computations. Secondly, ML techniques have been applied to improve the fitting procedures of the DA models, thus improving their predictive power.</td></tr> <tr><td class="formatRecordLabel"> Copyright/License </td><td style="padding-left:5px;">publication: © 2021 (License: <a href="http://creativecommons.org/licenses/by/3.0/">CC-BY-3.0</a>)</td></tr> </table> <br/>Corresponding record in: <a href="http://inspirehep.net/record/1926684">Inspire</a> <small> </small> <br/> <br/><br/><div align="right"><div style="padding-bottom:2px;padding-top:30px;"><span class="moreinfo" style="margin-right:10px;"> <a href="" class="moreinfo">Back to search</a> </span></div></div> <div class="bottom-left-folded"><div class="recordlastmodifiedbox" style="position:relative;margin-left:1px"> Record created 2022-03-25, last modified 2022-03-25</div></div> <div class="bottom-right-folded" style="text-align:right;padding-bottom:2px;"> <span class="moreinfo" style="margin-right:10px;"></span></div> </div> </div> </div> <br/> <br /> <div class="detailedrecordminipanel"> <div class="top-left"></div><div class="top-right"></div> <div class="inside"> <div id="detailedrecordminipanelfile" style="width:33%;float:left;text-align:center;margin-top:0"> <div><small class="detailedRecordActions">Fulltext:</small> <br /><a href="/record/2804876/files/document.pdf"><img style="border:none" src="/img/file-icon-text-34x48.gif" alt="Download fulltext" /><br />PDF</a><br /></div> </div> <div id="detailedrecordminipanelreview" style="width:30%;float:left;text-align:center"> </div> <div id="detailedrecordminipanelactions" style="width:36%;float:right;text-align:right;"> <ul class="detailedrecordactions"> <li><a href="/yourbaskets/add?ln=en&recid=2804876">Add to personal basket</a></li> <li>Export as <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/hx?ln=en">BibTeX</a>, <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/hm?ln=en">MARC</a>, <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/xm?ln=en">MARCXML</a>, <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/xd?ln=en">DC</a>, <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/xe?ln=en">EndNote</a>,  <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/xn?ln=en">NLM</a>, <a style="text-decoration:underline;font-weight:normal" href="/record/2804876/export/xw?ln=en">RefWorks</a> </li> </ul> <div style='padding-left: 13px;'>  <div id="bookmark"></div> <div id="bookmark_sciencewise"></div> <style type="text/css"> #bookmark_sciencewise, #bookmark {float: left;} #bookmark_sciencewise li {padding: 2px; width: 25px;} #bookmark_sciencewise ul, #bookmark ul {list-style-image: none;} </style> <script type="text/javascript" src="/js/jquery.bookmark.min.js"></script> <style type="text/css">@import "/css/jquery.bookmark.css";</style> <script type="text/javascript">// <![CDATA[ $.bookmark.addSite('sciencewise', 'ScienceWise.info', 'http://cds.cern.ch/img/sciencewise.png', 'en', 'bookmark', 'http://sciencewise.info/bookmarks/cds:2804876/add'); $('#bookmark_sciencewise').bookmark({sites: ['sciencewise']}); $('#bookmark').bookmark({ sites: ['facebook', 'twitter', 'linkedin', 'google_plusone'], icons: '/img/bookmarks.png', url: 'http://cds.cern.ch/record/2804876', addEmail: true, title: "Using Machine Learning to Improve Dynamic Aperture Estimates", description: "The dynamic aperture (DA) is an important concept in the study of nonlinear beam dynamics. Several analytical models used to describe the evolution of DA as a function of time, and to extrapolate to realistic time scales that would not be reachable otherwise due to computational limitations, have been successfully developed. Even though these models have been quite successful in the past, the fitting procedure is rather sensitive to several details. Machine Learning (ML) techniques, which have been around for decades and have matured into powerful tools ever since, carry the potential to address some of these challenges. In this paper, two applications of ML approaches are presented and discussed in detail. Firstly, ML has been used to efficiently detect outliers in the DA computations. Secondly, ML techniques have been applied to improve the fitting procedures of the DA models, thus improving their predictive power." }); // ]]> </script>  </div> </div> <div style="clear:both;margin-bottom: 0;"></div> </div> <div class="bottom-left"></div><div class="bottom-right"></div> </div> </div></div> <footer id="footer" class="pagefooter clearfix">  <div class="pagefooterstripeleft"> CERN Document Server :: <a class="footer" href="http://cds.cern.ch/?ln=en">Search</a> :: <a class="footer" href="http://cds.cern.ch/submit?ln=en">Submit</a> :: <a class="footer" href="https://cds.cern.ch/youraccount/display?ln=en">Personalize</a> :: <a class="footer" href="http://cds.cern.ch/help/?ln=en">Help</a> :: <a class="footer" href="https://cern.service-now.com/service-portal?id=privacy_policy&se=CDS-Service" target="_blank">Privacy Notice</a> :: <a class="footer" href="https://repository.cern/content-policy" target="_blank">Content Policy</a> :: <a class="footer" href="https://repository.cern/terms" target="_blank">Terms and Conditions</a> <br /> Powered by <a class="footer" href="http://invenio-software.org/">Invenio</a> <br /> Maintained by <a class="footer" href="https://cern.service-now.com/service-portal?id=service_element&name=CDS-Service">CDS Service</a> - Need help? Contact <a href="https://cern.service-now.com/service-portal?id=service_element&name=CDS-Service">CDS Support</a>. <br /> </div> <div class="pagefooterstriperight"> <div class="cern-logo"> <a id="logo" href="http://cern.ch" title="CERN" rel="CERN" ><img src="http://cds.cern.ch/img/cern_theme/img/cern-logo-large.png" alt="CERN" /></a> </div> <div class="cern-languagebox"> This site is also available in the following languages:<br /><a href="/record/2804876?ln=bg" class="langinfo">Български</a>  <a href="/record/2804876?ln=ca" class="langinfo">Català</a>  <a href="/record/2804876?ln=de" class="langinfo">Deutsch</a>  <a href="/record/2804876?ln=el" class="langinfo">Ελληνικά</a>  <span class="langinfo">English</span>  <a href="/record/2804876?ln=es" class="langinfo">Español</a>  <a href="/record/2804876?ln=fr" class="langinfo">Français</a>  <a href="/record/2804876?ln=hr" class="langinfo">Hrvatski</a>  <a href="/record/2804876?ln=it" class="langinfo">Italiano</a>  <a href="/record/2804876?ln=ja" class="langinfo">日本語</a>  <a href="/record/2804876?ln=ka" class="langinfo">ქართული</a>  <a href="/record/2804876?ln=no" class="langinfo">Norsk/Bokmål</a>  <a href="/record/2804876?ln=pl" class="langinfo">Polski</a>  <a href="/record/2804876?ln=pt" class="langinfo">Português</a>  <a href="/record/2804876?ln=ru" class="langinfo">Русский</a>  <a href="/record/2804876?ln=sk" class="langinfo">Slovensky</a>  <a href="/record/2804876?ln=sv" class="langinfo">Svenska</a>  <a href="/record/2804876?ln=zh_CN" class="langinfo">中文(简)</a>  <a href="/record/2804876?ln=zh_TW" class="langinfo">中文(繁)</a> </div> </div>  </footer> <script type="text/javascript"> var SyndeticsBookCovers = (function() { var SMALL_SIZE = "sc.gif", MEDIUM_SIZE = "mc.gif", RAW_URL = "https://secure.syndetics.com/index.aspx?isbn=THEISBN/THESIZE&client=cernlibrary"; replaceCover = function(imgElement, isbns, hdFormat) { var img = new Image(), size = hdFormat ? MEDIUM_SIZE : SMALL_SIZE; var _isbns = isbns.sort(function(a, b) { // sort from shortest to longest ISBN (more modern) return a.length > b.length ? 1 : -1; }); function next() { var isbn = _isbns.pop(); if (isbn) { var url = RAW_URL.replace("THEISBN", isbn).replace("THESIZE", size); img.src = url; } } function done() { imgElement.src = img.src; } img.onload = function() { if (this.width > 1) { done(); } else { next(); } }; next(); }; return { replaceCover: replaceCover }; })(); $(document).ready(function() { // get book covers $("img.book-cover").each(function() { var $this = $(this), strIsbns = $this.data("isbns") || "", isbnsArray = String(strIsbns).split(","), hdFormat = $this.hasClass("hd"); SyndeticsBookCovers.replaceCover(this, isbnsArray, hdFormat); }); // WebNews tooltips $.ajax({ url: "/news/tooltips", success: function(data) { create_tooltips(data); }, dataType: "json", cache: false }); }); </script>  <script src="//cds.cern.ch/js/feedback.js"></script>   <script> var _paq = window._paq = window._paq || []; /* tracker methods like "setCustomDimension" should be called before "trackPageView" */ _paq.push(['trackPageView']); _paq.push(['enableLinkTracking']); (function() { var u="https://webanalytics.web.cern.ch/"; _paq.push(['setTrackerUrl', u+'matomo.php']); _paq.push(['setSiteId', '756']); var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; g.async=true; g.src=u+'matomo.js'; s.parentNode.insertBefore(g,s); })(); </script>  </body> </html>