STOCHASTIC TEAMS WITH NONCLASSICAL INFORMATION REVISITED: WHEN IS AN AFFINE LAW OPTIMAL? .

Rajesh Bansal, Tamer Basar

Research output: Contribution to journalConference articlepeer-review

Abstract

A parameterized family of two-stage stochastic control problems with nonclassical information patterns is considered including the counterexample of H. S. Witsenhausen (see SIAM J. Contr. , vol. 6, no. 1, pp. 131-147, 1968). It is shown that the parameter region can be partitioned into two regions, in one of which the optimal solution is linear whereas in the other it is inherently nonlinear. In the nonlinear region, the best piecewise-constant solution does not always outperform the best linear solution, whereas a linear plus piecewise-constant policy does lead to a uniformly better performance. Numerical computations complement the study.

Original languageEnglish (US)
Pages (from-to)45-50
Number of pages6
JournalProceedings of the American Control Conference
DOIs
StatePublished - 1986

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'STOCHASTIC TEAMS WITH NONCLASSICAL INFORMATION REVISITED: WHEN IS AN AFFINE LAW OPTIMAL? .'. Together they form a unique fingerprint.

Cite this