This thesis investigates shooting methods as numerical techniques for solving boundary value problems arising in optimal control. We review the main variants: single shooting, multiple shooting, and NOC-based approaches. Moreover, the research discuss their practical differences in terms of stability, convergence, and implementation. The methods are then implemented in Python and applied to the Ramsey–Cass–Koopmans growth model, an infinite-horizon optimal control problem in economics. The numerical experiments illustrate how classical shooting techniques can recover the optimal saddle-path dynamics and highlight the trade-offs between single and multiple shooting, comparing them in terms of convergence to the solution, terminal errors, and sensitivity to the initial guess.
Questa tesi analizza i metodi di shooting come tecniche numeriche per risolvere problemi ai valori al contorno che emergono nel controllo ottimo. Vengono esaminate le principali varianti: single shooting, multiple shooting e approcci basati su NOC. Inoltre, la ricerca discute le loro differenze pratiche in termini di stabilità, convergenza e implementazione. I metodi vengono poi implementati in Python e applicati al modello di crescita di Ramsey–Cass–Koopmans, un problema di controllo ottimo a orizzonte infinito in economia. Gli esperimenti numerici illustrano come le tecniche classiche di shooting possano ricostruire la dinamica ottimale del saddle-path e mettono in evidenza i compromessi tra single e multiple shooting, confrontandoli in termini di convergenza alla soluzione, errori terminali e sensibilità alla scelta iniziale.
The Shooting Method: An Application to the Ramsey Optimal Control Problem
PREATONI, ELISA
2025/2026
Abstract
This thesis investigates shooting methods as numerical techniques for solving boundary value problems arising in optimal control. We review the main variants: single shooting, multiple shooting, and NOC-based approaches. Moreover, the research discuss their practical differences in terms of stability, convergence, and implementation. The methods are then implemented in Python and applied to the Ramsey–Cass–Koopmans growth model, an infinite-horizon optimal control problem in economics. The numerical experiments illustrate how classical shooting techniques can recover the optimal saddle-path dynamics and highlight the trade-offs between single and multiple shooting, comparing them in terms of convergence to the solution, terminal errors, and sensitivity to the initial guess.| File | Dimensione | Formato | |
|---|---|---|---|
|
Preatoni_Elisa.pdf
accesso aperto
Dimensione
3.32 MB
Formato
Adobe PDF
|
3.32 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/20.500.12608/105446