[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118291-en":3,"doc-seo-118291-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},118291,7971461740886,"Theodore","https://ap-avatar.wpscdn.com/davatar_3d24733baf745e90a7e4bdd5f77d97b2",8,"Research & Report","Applying machine learning methods to laser acceleration of protons - lessons learned from synthetic data","Study evaluates three machine learning approaches—a two-hidden-layer neural network, Support Vector Regression, and Gaussian Process Regression—for learning proton acceleration behavior in the Target Normal Sheath Acceleration regime using a synthetic dataset. The dataset is generated from a modified theoretical model (Fuchs et al. 2005) and includes controlled noise levels. After training, the methods are assessed for predictive accuracy and for practical execution on a single GPU, including memory consumption. Results clarify which models learn effectively from limited synthetic shots and how they can support laser parameter optimization and inverse problem control of proton energy spectra.","arXiv :2307 . 16036v5 [physics .plasm-ph] 16 Apr 2024  \nUnder consideration for publication in J. Plasma Phys. 1  \nApplying machine learning methods to laser acceleration of protons: lessons learned from  \nsynthetic data  \nRonak Desai 1†, Thomas Zhang 1 , John J. Felice 1 , Ricky Oropeza 1 , Joseph R. Smith 2 , Alona Kryshchenko 3 , Chris Orban 1 , Michael L.  \nDexter 4 and Anil K. Patnaik 4  \n1 Department of Physics, The Ohio State University, Columbus, OH 43210, USA  \n2 Department of Physics, Marietta College, Marietta, OH 43750, USA  \n3 Department of Mathematics, California State University Channel Islands, Camarillo, CA 93012, USA  \n4 Air Force Institute of Technology, Wright-Patterson AFB, OH 45433, USA (Received xx; revised xx; accepted xx)  \nIn this study we consider three different machine learning methods – a two-hidden layer neural network, Support Vector Regression and Gaussian Process Regression – and compare how well they can learn from a synthetic data set for proton acceleration in the Target Normal Sheath Acceleration regime. The synthetic data set was generated from a previously published theoretical model by Fuchs et al. 2005 that we modified. Once trained, these machine learning methods can assist with efforts to maximize the peak proton energy, or with the more general problem of configuring the laser system to produce a proton energy spectrum with desired characteristics. In our study we focus on both the accuracy of the machine learning methods and the performance on one GPU including the memory consumption. Although it is arguably the least sophisticated machine learning model we considered, Support Vector Regression performed very wellin our tests.  \n1. Introduction  \nThe field of ultra-intense laser science is increasingly beginning to embrace machine learning methods (Anirudh et al. 2022; Döpp et al. 2023) . This is especially true asthe repetition rates of ultra-intense laser systems increase and data acquisition systems improve (e.g. Heuer et al. (2022)) . However, to date, there has been limited use of machine learning (ML) to enhance and control proton acceleration from ultra-intense laser systems. Recently, Loughran et al. (2023) provide results from training a ML model on proton acceleration data using a laser system that operates at 1 Hz repetition rate using Bayesian optimization. Ma et al. (2021) also describe at a high level ongoing efforts towards similar goals on laser systems operating at repetition rate of a few Hz using a neural network approach.  \nThere are ultra-intense laser systems that operate at higher than 1 Hz repetition rates, and these systems are already being used in efforts to accelerate protons. Morrison et al.(2018), for example, accelerated protons to ∼2 MeV energies using few mJ ultra-intense laser pulses at a repetition rate of 1 kHz. This experiment could potentially be replicatedon the many mJ class, ∼kHz repetition rate laser systems that exist today (e.g. Cao et al.  \n† Email address for correspondence: [desai.458@osu.edu](desai.458@osu.edu)  \n2 Desai et al.  \n(2023)) . It is also true that future industrial or defense applications of ultra-intense laser systems will likely operate closer to this repetition rate regime (Palmer 2018) . Ideally, we would like to train ML models on these systems in quasi-real time. A natural question is therefore, of the ML models that are being used today, can they be quickly trained on tens of thousands of shots or more? Can this be achieved with modest computational resources such as a single GPU, or would it require a supercomputer or GPU cluster? The ability to train a ML model in quasi-real time could greatly assist efforts to optimize and/or control the properties of ion beams resulting from the laser interaction. Specifically, the ML model could help discover ways to increase the max proton energy, or it could help with the inverse problem of wanting a particular ion energy distribution and needing to know the laser parameters","cbCailEBbeocoHSC","https://ap.wps.com/l/cbCailEBbeocoHSC","pdf",3526853,1,18,"English","en",105,"# Introduction\n## Ultra-intense laser science and machine learning opportunities\n## Quasi-real-time training and inverse control of ion beams\n# Modified synthetic data generation\n## Noise modeling and relation to theoretical framework\n# Machine learning models\n## Neural network, Support Vector Regression, Gaussian Process Regression\n# Results and evaluation\n## Accuracy under noise and dataset size\n# Computational performance\n## Single-GPU memory consumption and runtime considerations","[{\"question\":\"Which machine learning methods are compared for proton acceleration learning?\",\"answer\":\"The study compares a two-hidden-layer neural network, Support Vector Regression, and Gaussian Process Regression for learning from synthetic proton acceleration data.\"},{\"question\":\"How is the synthetic dataset generated and what role does noise play?\",\"answer\":\"The dataset is generated from a modified version of the theoretical model by Fuchs et al. (2005). Noise is added intentionally at different levels to mimic variability relevant to real experiments.\"},{\"question\":\"Why is single-GPU performance included in the evaluation?\",\"answer\":\"The authors assess not only prediction accuracy but also practical training and inference feasibility on one GPU, explicitly tracking memory consumption to judge whether quasi-real-time workflows are realistic.\"}]","Applying machine learning methods to laser acceleration of protons - lessons learned from synthetic data | PDF",1785682836,45,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"applying-machine-learning-methods-to-laser-acceleration-of-protons-lessons-learned-from-synthetic-data","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/applying-machine-learning-methods-to-laser-acceleration-of-protons-lessons-learned-from-synthetic-data/118291/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Which machine learning methods are compared for proton acceleration learning?","Question",{"text":75,"@type":76},"The study compares a two-hidden-layer neural network, Support Vector Regression, and Gaussian Process Regression for learning from synthetic proton acceleration data.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is the synthetic dataset generated and what role does noise play?",{"text":80,"@type":76},"The dataset is generated from a modified version of the theoretical model by Fuchs et al. (2005). Noise is added intentionally at different levels to mimic variability relevant to real experiments.",{"name":82,"@type":73,"acceptedAnswer":83},"Why is single-GPU performance included in the evaluation?",{"text":84,"@type":76},"The authors assess not only prediction accuracy but also practical training and inference feasibility on one GPU, explicitly tracking memory consumption to judge whether quasi-real-time workflows are realistic.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]