Negative Ion Beam profile estimation on STRIKE calorimeter by means of Newton's Method. The method concerns the optimisation of 573 Synthetic and Experimental Images where there are 5 beamlets of the negative ion beam on STRIKE tiles and each beamlet is well approximated by a 2-d Gaussian, which depends on 5 parameters (such as amplitude, positions x and y of the centroids, sigma-x and sigma-y). The objective of Newton's method applied to this case was to minimize the mean squared error between the functions describing the original image and the reconstructed one. In the conclusions it was seen that this image regularization method is very precise, but however its convergence depends a lot on the goodness of the initial solution and it may not be that this was the best method best that suits this case. Machine learning optimization methods can be tested and used in the future.
Stima del profilo del fascio di ioni negativi sul calorimetro STRIKE attraverso l'applicazione del metodo di Newton. Il metodo riguarda l'ottimizzazione di 573 immagini Sintetiche e Sperimentali che raffigurano ciascuna 5 impronte termiche del fascio di ioni negativi su delle tegole di STRIKE e ogni fascetto è ben approssimato da una gaussiana 2-d, che dipende da 5 parametri (ampiezza, posizioni x e y dei centroidi, sigma-x e sigma-y). L'obiettivo del metodo di Newton applicato a questo caso era quello di minimizzare l'errore quadratico medio tra le funzioni che descrivono l'immagine originale e quella ricostruita. Nelle conclusioni si è visto che questo metodo di regolarizzazione dell'immagine è molto preciso, ma che tuttavia la convergenza dello stesso dipende molto dalla bontà della soluzione iniziale e che potrebbe non essere il metodo che meglio si adatta a questo caso. In futuro potranno essere testati e utilizzati metodi di ottimizzazione del Machine Learning.
Negative Ion Beam profile estimation on STRIKE calorimeter by means of Newton’s Method
DEGAN, ELEONORA
2022/2023
Abstract
Negative Ion Beam profile estimation on STRIKE calorimeter by means of Newton's Method. The method concerns the optimisation of 573 Synthetic and Experimental Images where there are 5 beamlets of the negative ion beam on STRIKE tiles and each beamlet is well approximated by a 2-d Gaussian, which depends on 5 parameters (such as amplitude, positions x and y of the centroids, sigma-x and sigma-y). The objective of Newton's method applied to this case was to minimize the mean squared error between the functions describing the original image and the reconstructed one. In the conclusions it was seen that this image regularization method is very precise, but however its convergence depends a lot on the goodness of the initial solution and it may not be that this was the best method best that suits this case. Machine learning optimization methods can be tested and used in the future.File | Dimensione | Formato | |
---|---|---|---|
Degan_Eleonora.pdf
accesso aperto
Dimensione
2.92 MB
Formato
Adobe PDF
|
2.92 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/52844