El Departamento de Control de Gestión y Sistemas de Información (DCS) tiene el agrado de invitar a usted al vigésimo octavo coloquio a realizarse este jueves 17 de Julio del 2014, de 15:00 a 16:30 en la sala 503 (Diagonal Paraguay 257).

El expositor será el Profesor Asistente Ph.D. Sigifredo Laengle, del DCS de la Facultad de Economía y Negocios.

Título - Non-negativity of information value in games, symmetry and invariance

Expositor - Sigifredo Laengle (Universidad de Chile)

Abstract - In the context of optimization problems of an agent, “having more information without additional cost is always beneficial” is a classic result by D. Blackwell (1953). Nevertheless, under strategic interaction this is not always true. Under what conditions more information is (socially) beneficial in games? How the characteristics of the players and the interaction among them affects the information value? Existing literature varies between two ends: on the one hand, we find works that calculate the information value of particular cases not always easy to generalize; on the other hand, there are also abstract studies which make difficult the analysis of more concrete applications. In order to fill this gap, we calculated the information value in the general case of constrained quadratic games in the framework of Hilbert spaces; we determined the conditions to assure its no-negativity; and studied how characteristics of the players affect their value. This work presents the most important results, a discussion of them, and a revision of the possible applications. As a result of our work we found a close relationship between the symmetry of information and the mathematical property of invariance. Such property is the base to calculate the information value and demonstrating its non-negativity. Furthermore, we found how the information value depends or not of some players characteristics. Such results improve the understanding of general conditions that assure the non-negativity of information value. In the immediate thing, this work can be extended to the study of other payoff functions with more general constraints.