Data-driven optimal control with a relaxed linear program
نویسندگان
چکیده
The linear programming (LP) approach has a long history in the theory of approximate dynamic programming. When it comes to computation, however, LP often suffers from poor scalability. In this work, we introduce relaxed version Bellman operator for q-functions and prove that is still monotone contraction mapping with unique fixed point. spirit approach, exploit new build program (RLP). Compared standard formulation, our RLP only one family constraints half decision variables, making more scalable computationally efficient. For deterministic systems, trivially returns correct q-function. stochastic systems continuous spaces, solution preserves minimizer optimal q-function, hence retrieves policy. Theoretical results are backed up simulation where solve sampled versions LPs data collected by interacting environment. general nonlinear observe again tends preserve minimizers LP, though relative performance influenced specific geometry problem.
منابع مشابه
Program Partitioning for a Control/data Driven Computer Program Partitioning for a Control/data Driven Computer ?
The paper examines the problem of dataaow graph partitioning aiming to improve the eeciency of macro-dataaow computing on a hybrid control/data driven architecture. The partitioning consists of dataaow graph synchronization and scheduling of the synchronous graph. A new scheduling algorithm, called Global Arc Minimization (GAM), is introduced. The performance of the GAM algorithm is evaluated r...
متن کاملCONTROL OF CHAOS IN A DRIVEN NON LINEAR DYNAMICAL SYSTEM
We present a numerical study of a one-dimensional version of the Burridge-Knopoff model [16] of N-site chain of spring-blocks with stick-slip dynamics. Our numerical analysis and computer simulations lead to a set of different results corresponding to different boundary conditions. It is shown that we can convert a chaotic behaviour system to a highly ordered and periodic behaviour by making on...
متن کاملcontrol of chaos in a driven non linear dynamical system
we present a numerical study of a one-dimensional version of the burridge-knopoff model [16] of n-site chain of spring-blocks with stick-slip dynamics. our numerical analysis and computer simulations lead to a set of different results corresponding to different boundary conditions. it is shown that we can convert a chaotic behaviour system to a highly ordered and periodic behaviour by making on...
متن کاملData-Driven Program Completion
We introduce program splicing, a programming methodology that aims to automate the commonly used workow of copying, pasting, and modifying code available online. Here, the programmer starts by writing a “dra” that mixes unnished code, natural language comments, and correctness requirements in the form of test cases or API call sequence constraints. A program synthesizer that interacts with a...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Automatica
سال: 2022
ISSN: ['1873-2836', '0005-1098']
DOI: https://doi.org/10.1016/j.automatica.2021.110052