Classic theory in mechanism design focuses on standard domains, in which there is no possibility for the mechanism to observe the realization of events. However, there are lots of practical situations, e.g. Web applications in which the user can somehow interact with the environment, for example through a click, in which it is possible to do so. In these uncertain and interdependent settings, where some random events will realize at runtime (uncertainty) with a probability that may depend on all the agents’ types (interdependence), many classical mechanisms fail to guarantee, among other properties, truthfulness. The designer can exploit the presence of the events defining different payments for any possible situation, using what are called Execution Contingent mechanisms, with the aim of recovering the desired properties of classical mechanisms. Up to now these settings have not been systematically studied and all the Execution Contingent mechanisms present in the literature are constructed ad hoc for the specific situation. Hence, in this work we give a general model for uncertain and interdependent settings and, focusing on Groves mechanisms, we show under which conditions it is possible to generalize classical mechanisms to this settings and to import the desired properties. VII
La teoria classica nel mechanism design si concentra su domini, cosiddetti standard, in cui non c’`e possibilit`a per il meccanismo di osservare la realizzazione di eventi. Tuttavia, in molte situazioni pratiche, come ad esempio applicazioniWeb in cui l’utente `e in grado in qualche modo di interagire con l’ambiente, magari attraverso un click, questo `e possibile. In questi scenari incerti ed interdipendenti, in cui alcuni eventi casuali si realizzano durante l’esecuzione (incertezza) con una probabilit`a che potrebbe dipendere dal tipo di tutti gli agenti (interdipendenza), molti meccanismi standard non riescono a garantire, tra le altre propriet`a, la compatibilit`a degli incentivi. Il designer pu`o sfruttare la presenza degli eventi definendo pagamenti diversi per ogni possibile situazione, utilizzando i cosiddetti meccanismi contingenti agli eventi, con lo scopo di recuperare le propriet`a desiderabili dei meccanismi classici. Finora questi scenari non sono stati studiati sistematicamente e tutti i meccanismi contingenti agli eventi presenti in letteratura sono costruiti ad hoc per la specifica situazione. Per questo, in questo lavoro di tesi ci proponiamo di fornire un modello generalizzato per scenari incerti ed interdipendenti e, concentrandoci sui meccanismi di Groves, mostriamo sotto quali condizioni `e possibile generalizzare i meccanismi classici e importare le propriet`a desiderate.
Execution-contingent Groves mechanisms for uncertain and interdependent settings
GALLI, NOEMI
2016/2017
Abstract
Classic theory in mechanism design focuses on standard domains, in which there is no possibility for the mechanism to observe the realization of events. However, there are lots of practical situations, e.g. Web applications in which the user can somehow interact with the environment, for example through a click, in which it is possible to do so. In these uncertain and interdependent settings, where some random events will realize at runtime (uncertainty) with a probability that may depend on all the agents’ types (interdependence), many classical mechanisms fail to guarantee, among other properties, truthfulness. The designer can exploit the presence of the events defining different payments for any possible situation, using what are called Execution Contingent mechanisms, with the aim of recovering the desired properties of classical mechanisms. Up to now these settings have not been systematically studied and all the Execution Contingent mechanisms present in the literature are constructed ad hoc for the specific situation. Hence, in this work we give a general model for uncertain and interdependent settings and, focusing on Groves mechanisms, we show under which conditions it is possible to generalize classical mechanisms to this settings and to import the desired properties. VIIFile | Dimensione | Formato | |
---|---|---|---|
Thesis.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: testo della tesi
Dimensione
599.9 kB
Formato
Adobe PDF
|
599.9 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/140075