Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks
Particle accelerators are enabling tools for scientific exploration and discovery in various disciplines. However, finding optimised operation points for these complex machines is a challenging task due to the large number of parameters involved and the underlying non-linear dynamics. Here, we intro...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-08-01
|
| Series: | Information |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2078-2489/12/9/351 |
| _version_ | 1797518827701927936 |
|---|---|
| author | Renato Bellotti Romana Boiger Andreas Adelmann |
| author_facet | Renato Bellotti Romana Boiger Andreas Adelmann |
| author_sort | Renato Bellotti |
| collection | DOAJ |
| description | Particle accelerators are enabling tools for scientific exploration and discovery in various disciplines. However, finding optimised operation points for these complex machines is a challenging task due to the large number of parameters involved and the underlying non-linear dynamics. Here, we introduce two families of data-driven surrogate models, based on deep and invertible neural networks, that can replace the expensive physics computer models. These models are employed in multi-objective optimisations to find Pareto optimal operation points for two fundamentally different types of particle accelerators. Our approach reduces the time-to-solution for a multi-objective accelerator optimisation up to a factor of 640 and the computational cost up to 98%. The framework established here should pave the way for future online and real-time multi-objective optimisation of particle accelerators. |
| first_indexed | 2024-03-10T07:34:56Z |
| format | Article |
| id | doaj.art-7f98d26e581f47faa9c523337972d91c |
| institution | Directory Open Access Journal |
| issn | 2078-2489 |
| language | English |
| last_indexed | 2024-03-10T07:34:56Z |
| publishDate | 2021-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Information |
| spelling | doaj.art-7f98d26e581f47faa9c523337972d91c2023-11-22T13:35:09ZengMDPI AGInformation2078-24892021-08-0112935110.3390/info12090351Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural NetworksRenato Bellotti0Romana Boiger1Andreas Adelmann2Paul Scherrer Institut, 5232 Villigen, SwitzerlandPaul Scherrer Institut, 5232 Villigen, SwitzerlandPaul Scherrer Institut, 5232 Villigen, SwitzerlandParticle accelerators are enabling tools for scientific exploration and discovery in various disciplines. However, finding optimised operation points for these complex machines is a challenging task due to the large number of parameters involved and the underlying non-linear dynamics. Here, we introduce two families of data-driven surrogate models, based on deep and invertible neural networks, that can replace the expensive physics computer models. These models are employed in multi-objective optimisations to find Pareto optimal operation points for two fundamentally different types of particle accelerators. Our approach reduces the time-to-solution for a multi-objective accelerator optimisation up to a factor of 640 and the computational cost up to 98%. The framework established here should pave the way for future online and real-time multi-objective optimisation of particle accelerators.https://www.mdpi.com/2078-2489/12/9/351surrogate model constructiondeep neural networkinverse neural networkcharged particle acceleratorcyclotronlinear accelerator |
| spellingShingle | Renato Bellotti Romana Boiger Andreas Adelmann Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks Information surrogate model construction deep neural network inverse neural network charged particle accelerator cyclotron linear accelerator |
| title | Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks |
| title_full | Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks |
| title_fullStr | Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks |
| title_full_unstemmed | Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks |
| title_short | Fast, Efficient and Flexible Particle Accelerator Optimisation Using Densely Connected and Invertible Neural Networks |
| title_sort | fast efficient and flexible particle accelerator optimisation using densely connected and invertible neural networks |
| topic | surrogate model construction deep neural network inverse neural network charged particle accelerator cyclotron linear accelerator |
| url | https://www.mdpi.com/2078-2489/12/9/351 |
| work_keys_str_mv | AT renatobellotti fastefficientandflexibleparticleacceleratoroptimisationusingdenselyconnectedandinvertibleneuralnetworks AT romanaboiger fastefficientandflexibleparticleacceleratoroptimisationusingdenselyconnectedandinvertibleneuralnetworks AT andreasadelmann fastefficientandflexibleparticleacceleratoroptimisationusingdenselyconnectedandinvertibleneuralnetworks |