Many measurements are required to control thermonuclear plasmas and to fully exploit them scientifically. In the last years JET has shown the potential to generate about 50 GB of data per shot. These amounts of data require more sophisticated data analysis methodologies to perform correct inference and various techniques have been recently developed in this respect. The present paper covers a new methodology to extract mathematical models directly from the data without any a priori assumption about their expression. The approach, based on symbolic regression via genetic programming, is exemplified using the data of the International Tokamak Physics Activity database for the energy confinement time. The best obtained scaling laws are not in power law form and suggest a revisiting of the extrapolation to ITER. Indeed the best non-power law scalings predict confinement times in ITER approximately between 2 and 3 s. On the other hand, more comprehensive and better databases are required to fully profit from the power of these new methods and to discriminate between the hundreds of thousands of models that they can generate. © 2015 EURATOM.

Murari, A., Peluso, E., Gelfusa, M., Lupelli, I., Lungaroni, M., Gaudio, P. (2015). Symbolic regression via genetic programming for data driven derivation of confinement scaling laws without any assumption on their mathematical form. PLASMA PHYSICS AND CONTROLLED FUSION, 57(1) [10.1088/0741-3335/57/1/014008].

Symbolic regression via genetic programming for data driven derivation of confinement scaling laws without any assumption on their mathematical form

Peluso, E;GELFUSA, MICHELA;LUPELLI, IVAN;GAUDIO, PASQUALINO
2015-01-01

Abstract

Many measurements are required to control thermonuclear plasmas and to fully exploit them scientifically. In the last years JET has shown the potential to generate about 50 GB of data per shot. These amounts of data require more sophisticated data analysis methodologies to perform correct inference and various techniques have been recently developed in this respect. The present paper covers a new methodology to extract mathematical models directly from the data without any a priori assumption about their expression. The approach, based on symbolic regression via genetic programming, is exemplified using the data of the International Tokamak Physics Activity database for the energy confinement time. The best obtained scaling laws are not in power law form and suggest a revisiting of the extrapolation to ITER. Indeed the best non-power law scalings predict confinement times in ITER approximately between 2 and 3 s. On the other hand, more comprehensive and better databases are required to fully profit from the power of these new methods and to discriminate between the hundreds of thousands of models that they can generate. © 2015 EURATOM.
2015
Pubblicato
Rilevanza internazionale
Articolo
Esperti anonimi
Settore FIS/01 - FISICA SPERIMENTALE
English
Con Impact Factor ISI
http://www.scopus.com/inward/record.url?eid=2-s2.0-84913529650&partnerID=40&md5=b81a9ce77c8530c6c81dce61c3868ab9
Murari, A., Peluso, E., Gelfusa, M., Lupelli, I., Lungaroni, M., Gaudio, P. (2015). Symbolic regression via genetic programming for data driven derivation of confinement scaling laws without any assumption on their mathematical form. PLASMA PHYSICS AND CONTROLLED FUSION, 57(1) [10.1088/0741-3335/57/1/014008].
Murari, A; Peluso, E; Gelfusa, M; Lupelli, I; Lungaroni, M; Gaudio, P
Articolo su rivista
File in questo prodotto:
File Dimensione Formato  
ppcf_murari.pdf

solo utenti autorizzati

Licenza: Copyright dell'editore
Dimensione 733.79 kB
Formato Adobe PDF
733.79 kB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2108/116596
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 27
  • ???jsp.display-item.citation.isi??? 23
social impact